ILLUMINATE ®

PUBLICATIONS
Contact us:
E-mail – contact@adiuvodiagnostics.com
Website – https://www.adiuvodiagnostics.com
Original Article
The International Journal of Lower
Extremity Wounds
A Clinical Study to Evaluate 1–8
© The Author(s) 2021
Autofluorescence Imaging of Diabetic Foot Article reuse guidelines:
sagepub.com/journals-permissions

Ulcers Using a Novel Artificial Intelligence DOI: 10.1177/15347346211047098

journals.sagepub.com/home/ijl

Enabled Noninvasive Device

Vijay Viswanathan, MD, PhD, FICP, FRCP(London),

FRCP(Glasgow)1 , Senthil Govindan, MBBS,
MS(General Surgery), MCH(Plastic Surgery)1,
Bamila Selvaraj, MPT(Neurology), MIAP, Dip. in Podiatry1,
Secunda Rupert, MD2, and Raghul Kumar, M Sc3

Abstract
Diabetic foot ulcers, with worldwide prevalence ranging from 12%-25%, are an important cause of nontraumatic lower limb
amputation. Evidence-based assessment of early infection can help the clinician provide the right first line treatment thus
helping improve the wound closure rate. Illuminate®, a novel point of care device working on multispectral autofluores-
cence imaging, helps in the rapid identification and classification of bacteria. This study was aimed to evaluate the diagnostic
accuracy of the device in detecting bacterial gram type against standard culture methods. A total of 178 patients from a
tertiary care center for diabetes was recruited and 203 tissue samples were obtained from the wound base by the plastic
surgeon. The device was handled by the trained investigator to take wound images. The tissue samples were taken from the
color-coded infected region as indicated by the device’s Artificial Intelligence algorithm and sent for microbial assessment.
The results were compared against the Gram type inferred by the device and the device was found to have an accuracy of
89.54%, a positive predictive value of 86.27% for detecting Gram-positive bacteria, 80.77% for Gram-negative bacteria, and
91.67% for no infection. The negative predictive value corresponded to 87.25% for Gram-positive, 92% for Gram-negative,
and 96.12% for no infection. The Results exhibited the accuracy of this novel autofluorescence device in identifying and
classifying the gram type of bacteria and its potential in significantly aiding clinicians towards early infection assessment
and treatment.

Keywords
diabetic foot ulcers, illuminate®, gram-positive bacteria, gram-negative bacteria, multispectral imaging, autofluorescence,
artificial intelligence

Introduction
The International Diabetes Federation estimates that the
expected prevalence of diabetes mellitus will rise to more 1
MV Hospital for Diabetes and Prof M Viswanathan Diabetes Research
than 570 million in 2030.1 Diabetic foot ulcers (DFU), Centre, Chennai, Tamil Nadu
2
and diabetic foot infections are major contributors of world- Stem Cell Research Centre, Government Stanley Medical College &
wide morbidity and mortality as they lead to limb amputa- Hospital, Chennai, Tamil Nadu
3
Adiuvo Diagnostics Private Limited, Chennai, Tamil Nadu
tion in 12%-25% of individuals with diabetes.2 DFU
initially begins with peripheral neuropathy and finally lead Corresponding Author:
to ulcerations. Secondary infections leads to chronically Vijay Viswanathan, M.V. Hospital for Diabetes and Prof. M. Viswanathan
Diabetes Research Centre (WHO Collaborating Centre for Research,
infected wounds over a period of time.3 Prompt diagnosis
Education and Training in Diabetes) (IDF centre for Excellence in Diabetes
of bacteria in these wounds and timely treatment help care), No. 4, West Madha Church Street, Royapuram, Chennai, Tamil
speed up the wound healing process. However, it is clini- Nadu, India.
cally difficult to diagnose the causative pathogens causing Email: drvijay@mvdiabetes.com
2 The International Journal of Lower Extremity Wounds

secondary infections. Routine microbial-based culture
methods, the current standard for identifying and classifying
bacteria infecting the wounds, requires taking a swab or
biopsy from the infected region and take 2 to 3 days for
diagnosis, causing a delay in providing first-line treatment
and/or inappropriate usage of antibiotics leading to the
development of bacterial resistance.4
Autofluorescence-based multispectral imaging has
recently emerged as a useful technique for the identification
of bacteria and involves shining multiple wavelengths of
light on wounds and collecting their emission responses
thus helping understand the pathophysiology of the wounds.
Fluorescence methods have been found to be one of the
most straightforward, noncontact, and sensitive methods for
the detection of biomolecules (intrinsic fluorophores) such
as tryptophan, nicotinamide adenine dinucleotide hydride
(NADH), flavins, etc.5 Different bacteria contain varying
amounts of these fluorophores in differing microenviron-
ments which help in distinguishing different types of bacte-
ria.6 This study was carried out to evaluate the accuracy of
Illuminate®, one such autofluorescence-based point of a
care imaging device that uses a multispectral imaging techni-
que to shine multiple excitation sources on the region of the
wound and collects spectral images. An Artificial
Intelligence (AI) algorithm then runs on the collected spectral
images to compare the biomolecule fluorophore intensity
toward Gram-type bacterial classification. The results
obtained from this device were compared against the current
gold standard method of bacteriological culture.
T2DM with acute and/or chronic DFU were recruited into
the study. Patients’ peri lesion noninfected site served as
their own control.
Illuminate device and Imaging Procedure with
Illuminate®
Intrinsic fluorescence exhibited by bacteria and fungi are
attractive methods for the rapid identification of infection
on wounds without any additional steps. Previous studies
have reported that bacteria have characteristic emission fluo-
rescence when excited in the UV and blue regions of light
contributed mainly by metabolic and infectious markers
such as NADH, Flavin and Porphyrin.8 Shelly et al,9 reported
fluorescence fingerprinting by characterizing each bacteria’s
excitation and characteristic emission wavelengths for rapid
and accurate identification of infection especially
Pseudomonas species and Staphylococcus aureus. On the
other hand, Ammor et al5 developed a method towards
species level bacterial detection leveraging the intrinsic fluo-
rescence of aromatic amino acids and nucleic acids, suggest-
ing that autofluorescence spectral analysis can be a rapid and
inexpensive technique. Based on the above principles of
using autofluorescence imaging of bacteria, a novel
imaging device named Illuminate® (Adiuvo Diagnostics
Private Limited, Chennai, India) (Figure 1) was developed,
combining multiple light sources (370, 395, and 415 nm)
to shine on the wound region and collects multispectral
images in approximately 30 sec to detect infection on
wounds. An image processing algorithm then analyses the

Methods
Study Design
An interventional single arm comparative study was con-
ducted at a tertiary care center for diabetes after obtaining
institutional ethics committee approval (IEC/N-009/10/
2017) registered with the Clinical Trial Registry of India
(Reg. No. CTRI/2018/10/016147). Both acute and chronic
wounds were considered for imaging and the total sample
size required for detection of infection using the device
was estimated as 140 (for 99% confidence level, assumed
prevalence of infection as 70% and a precision of 10%.7
Consecutive patients attending the outpatient department
were recruited after screening for inclusion criteria and
obtaining both informed and written consent forms. The
persons with type 2 diabetes (T2DM) and with new and/or
chronic DFU with nonintact skin (stasis ulcers both arterial
and venous) were included in the study. The persons with an
existing autoimmune disease and skin diseases such as
eczema, psoriasis in the area close to the wounds were
excluded. We excluded patients with osteomyelitis, as the
bone region is high autofluorescent in nature and needs
additional algorithm training. A total of 178 persons with Figure 1. Illuminate® imaging device.
Viswanathan et al 3

spectral images, and a pretrained machine learning algorithm Laboratory Standards Institute (CLSI) guidelines.13 The
compares the relative intensity between different autofluores- results of gram-staining, bacterial identification, and its drug
cence biomarkers10 thus enabling classification of bacteria susceptibility profile were obtained from the microbiology
into Gram-positive and Gram-negative in 2 min.10,11 The lab in a report format for each patient sample.
device is noncontact, does not require the addition of any
reagents and removes the burden of interpreting spectral
images. The AI Engine runs on the device and can even be Statistical Analysis
used in remote places with no internet access. The device pro-
The estimates of diagnostic tests such as true positives, true
vides a comprehensive report along with automated wound
negatives, false positives, false negatives were first evalu-
measurement options including a progressive report, which
ated and used for the calculation of sensitivity, specificity,
can be synced with the hospital’s central database. The
accuracy, positive predictive value, and negative predictive
device is easy to use and requires training to image the
value. The results obtained from the imaging device were
wound regions consistently without shaking.
compared with the results of the tissue culture to calculate
Illuminate device was used to collect Multispectral
accuracy, sensitivity, specificity, positive predictive value,
wound images either in a dark room or using a black
and negative predictive value. The statistical analysis was
hood covering the wound region. The wound imaging was
performed using Python software.
done by the trained investigator. Before imaging, wounds
were first cleaned with normal saline solution. The study
participants’ details were entered in the Illuminate
Results
Imaging Application of the device. The user was guided
to image the wound at a distance of 10 to 12 cm from the Clinical Characteristics of the Study Participants
area of interest. A series of 16 multispectral images were
automatically captured by the device in ∼20 sec. The user The imaging of the wound was done for a total of 178
was then prompted to trace the region of the wound post persons with DFU. The tissue sample was taken from 203
image capture. In under 2 min, an algorithm was processed sites and analyzed. Out of 178 study participants, 144 were
in the background and displayed a color-coded wound male and 34 were female with a median age of 59 years.
image along with a Gram type of bacteria (if infected) 13 Patients were on prior antibiotics usage and were included
along with the length and breadth of the wound. in the study (Table 1). About 60% of the participants were
found to have third-grade ulcers based on the Wagner’s
wound classification score, while 34.3% had grade 2 and
Specimen Collection
Tissue samples were taken with Alley’s tissue holding Table 1. Basic and Clinical Profile of the Study Participants.
forceps and scissors by the attending plastic surgeon.
These tissues were collected before and after debridement No. of participants (%)
Variable (n = 178)
from the color-coded region as indicated by the device.
Specimens were collected from all the study participants Age (years)a 59 (12.5)
and placed in a sterile transport container and sent to the Gender (M: F) 144:34
microbiology laboratory in the hospital for routine standard Duration of diabetes
culture and biochemical methods for identifying and classi- <5 years 39 (22.0)
fication of bacteria. 5-10 years 51 (28.6)
11-15 57 (32.0)
>15 31 (17.4)
Culture Wound classification
All tissue samples were cultured in the microbiology lab. All Grade 1 11 (6.2)
tissue samples were analyzed initially by Gram-staining and Grade 2 61 (34.3)
bacterial identification using conventional aerobic phenotypic Grade 3 106 (59.5)
methods by plating into various enriched and differential HbA1c (%)
growth mediums and biochemical assays.12 Bacterial load <7 23 (13.0)
7-8 43 (24.1)
was estimated by plate count semiquantitative analysis
>8 112 (62.9)
method and then graded as scanty, light, moderate, or heavy
Use of antibiotics prior to imaging
(1+, 2+, 3+ or 4+) of which moderate and heavy growth indi-
Yes 13 (7.3)
cated a significant bacterial load (ie, greater than 100 000 per No 165 (92.7)
gram of tissue). The antibiotics susceptibility profile was then
a
obtained using the disk diffusion method as per Clinical Median (IQR).
4 The International Journal of Lower Extremity Wounds

Figure 2. Percentage of treatment procedures.

6.2% had grade 1 ulcers respectively. A total of 131 patients
underwent debridement, while 20 participants were directed
towards cleaning and dressing, 17 towards sequestrectomy,
3 underwent exostectomy, 1 escharotomy, 1 bursa excision,
and 16 patients had to undergo an amputation. 39 wounds
had exudates. Around 63% of the study participants were
found to have an HbA1c level >8%.
The percentage of treatment procedures is shown in
Figure 2.
Organisms Isolated in Culture
From the 203 tissue samples collected, 14 clinically
relevant pathogens were isolated (Table 2) while 28 cases
showed no growth which included reference sites as well.
63 samples revealed one Gram-negative bacteria, 53 had
one Gram-positive bacteria, while 6 tissues had 2-gram-
negatives, 4 samples had 2-gram-positives and 49 patients
had both gram-positive and gram-negative bacteria
(Table 3).

Illuminate Inference
Table 2. Details of Pathogens Isolated From Wound Samples by
Culture Methods. The device color codes Gram-positive bacteria with red
(Figure 3b) and gram-negative with green (Figure 3a, c).
S. No Bacterial Isolates Count
Tissue culture gram type results were compared against
1 Acinetobacter SPP 4
2 Citrobacter Feundii 2
Table 3. Identification of Gram-Positive and Gram-Negative
3 E coli 19
Organisms in Wound Samples by Culture Methods vs Illuminate
4 E coli ESBL 12 Device.
5 Enterococcus SPP 48
6 Enterococcus faecalis 2 Results by illuminate
7 Klebsiella SPP 35 method (Index)
Results by culture
8 No Growth 28
Type of method Correct
9 Proteus mirabilis 25
organism (CLSI, Gram’s stain) diagnosis Misdiagnosis Total
10 Pseudomonas aeruginosa 1
11 Pseudomonas SPP 26 GN 69 63 6 69
12 Staphylococcus aureus 33 GP 57 44 13 57
13 MRSA 33 NG 28 22 6 28
14 Staphylococcus SPP 1 GPGN 49 39 10 49
15 Enterobacter SPP 1 Total 203 168 35 203

MRSA, methicillin-resistant Staphylococcus aureus; ESBL, Extended Spectrum GN, gram-negative; GP, gram-positive; NG, no growth; GPGN,
beta-lactamase. gram-positive and gram-negative.
Viswanathan et al 5

Figure 3. (a) Gram-negative bacteria-infected wound. (b) Gram-positive bacteria-infected wound. (c) Gram-negative bacteria-infected
wound. (d) Gram-positive and gram-negative bacteria-infected wound. Figures 3a-3d are examples of clinical image and Illuminate®
predicted image of Patient’s wound region. Figure 3a and 3c show a Green overlay indicating Gram-Negative infection and the
corresponding culture reports indicated Klebsiella Spp. and Pseudomonas spp. Figure 3b indicated red over overlay predicting
Gram-positive bacteria and the corresponding culture report was Staphylococcus aureus, 3d is an example of a wound infected by both
Gram-positive and negative bacteria.

the device gram type results. Illuminate device displayed
63-gram-negative bacteria, 44 gram-positive, 22 No
growths. 49 wounds were found to have a polymicrobial
infection containing both gram-negative and gram-positive
bacteria (Figureure 3d) and the device identified the same
in 39 patients. 6 ulcers were found to have only gram-
negative colonization and 4, gram-positive colonization.
Further statistical inference was made by comparing the
culture results with Illuminate device results (Table 3).
The results of the machine learning algorithm are sum-
marized in the confusion matrix provided in Table 4. The
accuracy of the device was found to be 89.54% with a pos-
itive predictive value of 80.77% for detecting gram-negative
bacteria, 86.3% for gram-positive bacteria, and 91.67%
for no growth. The negative predictive value corresponded
to 92.0% for gram-negative bacteria, 87.2% for gram-
positive bacteria, and 96.12% for no infection. The sensitiv-
ity for gram-negative bacteria was 91.30%, 77.2% for
6 The International Journal of Lower Extremity Wounds
Table 4. Confusion Matrix and Results for the Gram Type
Classification Using Illuminate® Device With the Standard
Culture-Based Test.
89.54%
Combined Accuracy GN GP
PPV 80.77 86.27
NPV 92.00 87.25
Accuracy 86.27 86.93
TPR (sensitivity) 91.30 77.19
TNR (specificity) 82.14 92.71
GN, gram-negative; GP, gram-positive; NG, no growth; GPGN,
gram-positive and gram-negative; PPV, positive predictive value; NPV,
negative predictive value; TPR, true positive rate; TNR, true negative rate.
Gram-positive, and 81.5% for No growth. Specificity for
Gram-negative was 82.14%, 92.71% for gram-positive,
and 98.4% for No Growth. Multiclass area under the
curve (AUC) score calculated using the One-versus-One
method was 0.86.
Discussion
Understanding the pathophysiology of DFU is essential for
effective treatment and prevention of its downstream com-
plications. Clinical Signs and Symptoms Score continue to
be the recommended way of detecting wound infection.14
However, this assessment can be very subjective as demon-
strated by various studies. An accurate, objective method
would be a boon to the doctors to better understand the
infection levels and to provide tailored antibiotic treatments.
Our study findings showed that the sensitivity for gram-
negative and gram-positive bacteria was 91.3% and 77.2%
respectively. A sensitivity of 81.5% was found for no
growth of any bacteria. The specificity of the test of the
device was 82.1% for gram-negative organisms, 92.7% for
gram-positive, and 98.4% for no growth of the organisms.
AUC score of 0.86 also indicates a significant discriminating
ability of the device to the presence of pathogens.
A pilot study conducted in a wound care clinic of Ireland
on the efficacy of a bacterial fluorescence imaging device
(Moleculight) showed a sensitivity of 100% and specificity
of 78%.15 The same study demonstrated that the device
has sensitivity and specificity of 100% in identifying the
Pseudomonas spp. by visually interpreting the cyan fluores-
cence from the autofluorescence images. This study also
showed the effect of appropriate selection of antibiotics
and further recovery of the wound within a 2-week
period. It is to be noted that these images need to be manu-
ally interpreted.16 and cannot automatically distinguish
Gram-positive and Gram-Negative bacteria.
Our study of 178 patients using the Illuminate® imaging
device can detect infections and differentiate the gram type
of bacteria using multispectral image information and AI.
The device has an accuracy of 89.54%, a positive predictive
value of 86.27% for detecting Gram-positive bacteria,
80.77% for Gram-negative bacteria, and 91.67% for no
infection. The negative predictive value corresponded to
NG 87.25% for Gram-positive, 92% for Gram-negative, and
96.12% for no infection.
91.67 The study conducted by DaCosta RS., et al showed that
96.12 only 52.5% of the bacterial load was accurately identified by
95.42
clinical assessment whereas 74.6% of the occurrences of
81.48
wound infection was correctly identified by fluorescent
98.41
imaging-based technique.17 Another study has re-confirmed
the high accuracy in the detection of the pathogenic infec-
tion of the wound by an autofluorescence imaging device
compared to the clinical evaluation of signs and symptoms
in detecting the wounds. It also demonstrated an improved
sensitivity of the combined use of bacterial fluorescence
imaging with clinical signs and symptoms (72%) compared
to the use of only clinical signs and symptoms (22%) in the
identification of wounds with moderate to heavy bacterial
loads (≥104CFU/g).18
Predominantly Gram-positive bacteria such as
methicillin-resistant Staphylococcus aureus19, methicillin
sensitive S. aureus, Streptococcus species are typically iden-
tified in the acute stage of wounds while gram-negative bac-
teria such as Escherichia coli, Pseudomonas aeruginosa,
Klebsiella spp., are detected in chronic wounds.20–22 Our
study showed the presence of Enterococcus spp. (17.8%)
majorly followed with the presence of Klebsiella spp.
(13%) and Staphylococcus aureus (12.2%). A bacterial con-
centration >104 CFU/g of tissue is necessary to cause infec-
tion in complex extremity wounds.23,24 Infections may then
lead to a delay in wound healing, thus early detection of bac-
teria above 104 CFU/g of tissue becomes imperative to
prevent complications.
The high sensitivity and specificity observed in the
present study, along with an exploratory study done
earlier10 reiterates that this novel device can be used as a
screening device for early infection detection and under-
standing the gram type of bacteria colonizing the wound.
Definitive tests such as culture, polymerase chain reaction
take time are reagent intensive, and are error prone due to
reasons such as error during sample transport, proximity
to a diagnostic lab, etc. This device can thus be used as a
screening tool to overcome these shortcomings. The
device can also help monitor the treatment efficacy over a
period of wound treatment and identify regions to take a
tissue sample. From our results, we could see a marked
decrease in bacterial fluorescence spatially, before and
post debridement and there was a 93% organism correlation
from the culture report obtained. Hence it can be used to
assess wounds qualitatively post debridement.11
Currently, wounds are assessed by visual inspection, and it
is difficult to track wound infection and healing post debride-
ment. More importantly, an instant decision on the gram type
Viswanathan et al
of bacteria infecting the wound can aid the doctors towards the
first line of antibiotic prescription. This device will reduce the
time and aid the clinicians ineffective treatment by understand-
ing the microbial burden on the wound region in real time.
Manual interpretation of images is very tough and requires
trained personnel and possesses some challenges in ensuring
the accuracy of decisions. Incorporation of machine learning
algorithm aids in easier understanding of infection regions.
In future, with more images, genus and species level
machine learning training can also be performed which can
guide towards effective antibiotic treatment.
Limitations
Bone, tendon, and fat along with betadine and other cleaning
solutions are autofluorescent in nature. Hence wounds are to
be cleaned with normal saline before imaging. Also, the
device cannot be used to interpret infections in the bone/fat
region. The light sources used here have a penetration depth
from 0.5 mm to ∼1 mm and hence the device cannot under-
stand the status of infection in closed wounds. In the case of
polymicrobial infection, the device displays the gram type
that is more in number in that particular spatial region.
Conclusion
Illuminate®, a novel hand device developed based on multi-
spectral imaging of autofluorescence exhibited by bacterium
has proven to be an early screening device in the identifica-
tion and classification of bacteria. The device looks at both
metabolic and infectious markers of the underlying pathogen
exhibiting a good overall accuracy and thus does not detect
commensals. The sensitivity to detect no infection wounds
is high and hence the prescription of antibiotics can be pro-
vided judiciously. The device’s high sensitivity to detect
gram-negative bacteria may aid physicians toward the
administration of agents with activity against gram-negative
bacteria. The high specificity to detect Gram-positive bacte-
ria can eliminate any burden faced by clinicians to combat
gram-positive infections. In addition, infections in the peri-
wound region and away from the wound region can be
detected by the device, thus aiding clinicians in the better
debridement process and improved understanding of the
wound infection levels qualitatively. These results observed
are closer to the gold standard culture methods and thus aids
the clinicians towards the first line treatment. This device will
also be helpful in convincing the patients easily to proceed
with treatment by showing the infection status of the
wound empirically in the outpatient clinical setting. The non-
invasive and less time-consuming procedure are the added
advantages of the device.
7
Acknowledgments
We thank the support from Ms Malathi, for her patient assistance,
the microbiology team for providing insights from culturing per-
spective, Mr Anand Kumar, Dr Satyavani, and Ms Suganya for
their support during the trial.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect
to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, author-
ship, and/or publication of this article.
Ethical Approval
Institutional Ethics Committee approval (IEC/N-009/10/2017) was
obtained before the initiation of the study.
ORCID iD
Vijay Viswanathan https://orcid.org/0000-0001-9116-3937
References
1. IDF diabetes ATLAS 9th edition 2019. n.d. Retrieved April
6, 2021, from https://www.diabetesatlas.org/upload/resour-
ces/material/20200302_133351_IDFATLAS9e-final-web.pdf
(Accessed date: 12-04-2021).
2. Boulton AJM, Armstrong DG, Hardman MJ, et al. Diagnosis
and Management of Diabetic Foot Infections. Arlington, VA:
American Diabetes Association; January 2020. doi:10.2337/
db2020-01
3. Rezende KF, Ferraz MB, Malerbi DA, et al. Predicted annual
costs for in patients with diabetes and foot ulcers in a develop-
ing country - a simulation of the current situation in Brazil.
Diabet Med. 2010;27(1):109-112. doi:10.1111/j.1464-5491.
2009.02871.x
4. Oliveira AF, Oliveira Filho H. Perfil microbiológico e de
resistência antimicrobiana no pé diabético infectado. J Vasc
Bras. 2014;13(4):289-293. doi:10.1590/1677-5449.0015
5. Ammor MS. Recent advances in the use of intrinsic
fluorescence for bacterial identification and characterization.
J Fluoresc. 2007;17(5):455-459. doi:10.1007/s10895-007-
0180-6
6. Awad F, Ramprasath C, Mathivanan N, Aruna PR, Ganesan S.
Optical fiber-based steady state and fluorescence lifetime
spectroscopy for rapid identification and classification of bac-
terial pathogens directly from colonies on Agar plates. Int Sch
Res Notices. 2014;2014:1–7. doi: 10.1155/2014/430412
7. Pourhoseingholi MA, Vahedi M, Rahimzadeh M. Sample size
calculation in medical studies. Gastroenterol Hepatol Bed
Bench. 2013;6(1):14-17.
8. Estes C, Duncan A, Wade B, Lloyd C, Ellis WJ, Powers L.
Reagentless detection of microorganisms by intrinsic fluores-
cence. Biosens Bioelectron. 2003;18(5-6):511-519. doi:10.
1016/S0956-5663(03)00008-3
8 The International Journal of Lower Extremity Wounds
9. Shelly DC, Quarles JM, Warner IM. Identification of fluores-
cent Pseudomonas species. Clin Chem. 1980;26(8):1127-
1132. doi:10.1093/clinchem/26.8.1127
10. Radhakrishnan G, Gupta AG, King J, Ganvir D. Rapid hand-
held screening device to detect skin and soft tissue infections.
In: Choi B, Zeng H, eds. Photonics in Dermatology and
Plastic Surgery 2020. SPIE; 2020:1–9. Accessed Feb 22
2021. Available from: https://www.spiedigitallibrary.org/
conference-proceedings-of-spie/11211/2546404/Rapid-hand
held-screening-device-to-detect-skin-and-soft-tissue/10.1117/
12.2546404.full
11. Shah BM, Ganvir D, Sharma YK, et al. Utility of a real-time
fluorescence imaging device in guiding antibiotic treatment in
superficial skin infections [published online ahead of print,
2021 Feb 24]. Indian J Dermatol Venereol Leprol. 2021:1-
6. doi:10.25259/IJDVL_856_19
12. Bauer AW, Kirby WMM, Sherris JC, Turck KM. Antibiotic
susceptibility testing by a standardized disk method. Amer J
Clin Path. 1966;45(4):493-496. doi:10.1093/ajcp/45.4_ts.493
13. Wayne P, Clinical Laboratory Standards Institute.
Performance standards for antimicrobial disk susceptibility
tests; Approved standard—9th ed. CLSI document M2-A9.
Clinical Laboratory Standards Institute; 2006. 26;1.
14. International Wound Infection Institute (IWII). Wound infec-
tion in clinical practice. Wound Int. 2016. Retrieved July 31,
2018 from https://www.woundsinternational.com/resources/
details/iwii-wound-infection-clinical-practice.
15. Hurley CM, McClusky P, Sugrue RM, Clover JA, Kelly JE.
Efficacy of a bacterial fluorescence imaging device in an out-
patient wound care clinic: a pilot study. J Wound Care.
2019;28(7):438-443. doi:10.12968/jowc.2019.28.7.438
16. Rennie MY, Dunham D, Lindvere-Teene L, Raizman R, Hill
R, Linden R. Understanding real-time fluorescence signals
from bacteria and wound tissues observed with the
MolecuLight i: XTM. Diagnostics (Basel. 2019;9(1):22.
Published 2019 Feb 26. doi:10.3390/diagnostics9010022
17. DaCosta RS, Kulbatski I, Lindvere-Teene L, et al. Point-of-
care autofluorescence imaging for real-time sampling and
treatment guidance of bioburden in chronic wounds: first-in-
human results. PLoS One. 2015;10(3):e0116623. doi:10.
1371/journal.pone.0116623
18. Serena TE, Harrell K, Serena L, Yaakov RA. Real-time bacte-
rial fluorescence imaging accurately identifies wounds with
moderate-to-heavy bacterial burden. J Wound Care. 2019;
28(6):346-357. doi:10.12968/jowc.2019.28.6.346
19. Guo Y, Ramos RI, Cho JS, Donegan NP, Cheung AL, Miller
LS. In vivo bioluminescence imaging to evaluate systemic and
topical antibiotics against community-acquired methicillin-
resistant Staphylococcus aureus-infected skin wounds in
mice. Antimicrob Agents Chemother. 2013;57(2):855-863.
doi:10.1128/AAC.01003-12
20. Sadeghpour Heravi F, Zakrzewski M, Vickery K, G
Armstrong D, Hu H. Bacterial diversity of diabetic foot
ulcers: current status and future prospectives. J Clin Med.
2019;8(11):1935. doi:10.3390/jcm8111935
21. Sankar N, Khaja Moinuddin S, Mohan S. Microbial etiology
of diabetic foot ulcers: swab versus tissue culture. Int J Surg
Sci. 2020;4(2):146-148. doi:10.33545/surgery.2020.v4.i2c.
408
22. Huang Y, Cao Y, Zou M, et al. A comparison of tissue versus
swab culturing of infected diabetic foot wounds. Int J
Endocrinol. 2016;2016:8198714. doi:10.1155/2016/8198714
23. Jneid J, Cassir N, Schuldiner S, et al. Exploring the Microbiota
of diabetic foot infections with culturomics. Front Cell Infect
Microbiol. 2018;8:282. doi:10.3389/fcimb.2018.00282
24. Williams DT, Hilton JR, Harding KG. Diagnosing foot infec-
tion in diabetes. Clin Infect Dis. 2004;39(Suppl 2):S83-S86.
doi:10.1086/383267
 ijdvl.com Case Report

Utility of a real‑time fluorescence imaging Corresponding author:

Dr. Aayush Gupta,
Department of Dermatology,
device in guiding antibiotic treatment in Dr. D. Y. Patil Medical College,
Hospital and Research Centre,
superficial skin infections Pimpri, Pune, Maharashtra, India.
aayushggupta@gmail.com

Received: October, 2019

Bhavika M. Shah, Devina Ganvir, Yugal K. Sharma, Shahzad Beg Mirza1, Accepted: February, 2020
Published:
R. N. Misra1, Preeti Kothari, Sweety Darall, Jitendra S. Bhawalkar2,
Aayush Gupta
DOI:
Departments of Dermatology, 1Microbiology and 2Community Medicine, Dr. D. Y. Patil Medical College, Hospital 10.25259/IJDVL_856_19
and Research Centre, Pune, Maharashtra, India

PMID:
***
Abstract
The prescription of antibiotics empirically without confirmation of an infective etiology is on the rise.
Administration of appropriate antibiotics can be guided by real‑time fluorescence imaging using a
point‑of‑care device. These composite images show the presence, type and the burden of infection.
The time saved by this method over microbiological testing, especially in resource‑poor settings, can
lead to a paradigm shift in treatment by facilitating prompt and adequate antimicrobial therapy, surgical
debridement as well as follow‑up. Thumbnail sketches of a series of four cases highlighting different
scenarios in which a fluorescent imaging device utilizing artificial intelligence and machine learning
was found useful is presented in this report.

Key words: Antimicrobial stewardship, fluorescence imaging, Illuminate, infections, wound care

Introduction techniques, all organisms may not be isolated especially in

Most of the skin and soft tissue infections resolve with
treatment, however, some worsen forming chronic ulcers that
impair not only the patients’ quality of life but strain an already
overburdened healthcare system. An escalating incidence
of secondarily infected chronic wounds has led to increased
hospitalization of patients as well as antibiotic usage.1
The incidence of acute and chronic wounds in India is 10.5
and 4.5 per 1000 population respectively.2 Secondary bacterial
infections commonly complicate the healing of diabetic
ulcers leading to the loss of a lower limb every 30 seconds.3
Initial clinical evaluation and microbiological correlation of
an infected wound typically takes up to 10 days. An already
tedious recovery may get prolonged by incorrect therapy
consequent to a cursory examination, pending culture and
sensitivity results, or inaccurate reports following improper
swabbing. Moreover, despite proper swabbing/aseptic
comparison with that of deep tissue biopsies (current gold
standard).4,5
We used a novel handheld imaging device, “Illuminate,”
developed by Adiuvo Diagnostics (Chennai, India), which
aims to accelerate the treatment of skin infections by
rectifying some of the above‑mentioned lacunae. This
device leverages autofluorescence, artificial intelligence
and machine learning to accurately classify/locate the
microorganisms and provide semiquantitative data regarding
the infective burden of the wound. Pointed to the region of
interest, the device captures 16 serial images at different
excitation emissions and produces a clinical image overlaid
with the superimposed infection, if any, marking the
presence of gram‑positive bacteria in red, gram‑negative in
green, and fungus in blue. Four cases where the device was
found invaluable are summarized below.

How to cite this article: Shah BM, Ganvir D, Sharma YK, Mirza SB, Misra RN, Kothari P, et al. Utility of a real‑time fluorescence imaging device in
guiding antibiotic treatment in superficial skin infections. Indian J Dermatol Venereol Leprol 2020;1-6.

This is an open-access article distributed under the terms of the Creative Commons Attribution-Non Commercial-Share Alike 4.0 License, which allows others to remix, tweak,
and build upon the work non-commercially, as long as the author is credited and the new creations are licensed under the identical terms.

© 2020 Indian Journal of Dermatology, Venereology and Leprology - Published by Scientific Scholar 1
Shah, et al.
Case Reports
Case 1
A 73‑year‑old female  had multiple and painful non‑healing
recalcitrant ulcers on both thighs [Figure 1a] for 9 months,
despite multiple courses of tablet amoxicillin/clavulanic
acid, tablet azithromycin, and injectable amikacin as per
sensitivity reports of the culture isolates of Staphylococcus
epidermidis and Klebsiella. The ulcers were well defined,
punched out, 0.5 to 3 cm in diameter with erythematous
margins and seropurulent discharge. A punch biopsy revealing
neutrophilic infiltrate was described as ‘non‑specific’.
Repeated use of Illuminate for many days consistently failed
to capture the fluorescence indicative of any pathogenic
organisms, although repeated pus swabs showed growth
of coagulase‑negative Staphylococcus [Figure 1b]. Deep
excisional biopsy thence demonstrated a dense band
of neutrophilic infiltrate and perivascular lymphocytes
suggestive of pyoderma gangrenosum. The patient ultimately
responded well to corticosteroids (systemic and intralesional)
without any antibiotic coverage.
Case 2
A 58‑year‑old female, known diabetic for 10 years developed
an ulcer over her left great toe [Figure 2a] along with difficulty
in walking for 4 months. Escherichia coli were isolated but
despite the complete course of cefoxitin as per sensitivity,
Figure 1a: Clinical image of recalcitrant thigh ulcers
2 Indian Journal of Dermatology, Venereology and Leprology | Volume XX | Issue XX | Month 2020
Fluorescence imaging in antimicrobial stewardship
the ulcer did not heal. Cyan and green fluorescence in
two different areas of the ulcer examined by Illuminate
indicated the co‑colonization of Pseudomonas species and
Escherichia coli and the same was subsequently confirmed by
image‑guided swabbing [Figures 2b and c]. Prompt clinical
response followed with the use of ciprofloxacin.
Case 3
A 45‑year‑old HIV‑positive patient presented with a crusted
ulcer on left shin for 15 days. [Figure 3a] Initial assessment
with Illuminate   suggested a high gram‑positive bacterial
load  (subsequently confirmed to be methicillin‑sensitive
Staphylococcus aureus) and hence cotrimoxazole was given
empirically. Seven days later, the sensitivity report confirmed
it to be the appropriate choice, during which serial assessment
demonstrated reduced fluorescence at every visit indicating
a satisfactory response at the end of 10 days of treatment,
obviating the need for repeated wound swabs [Figures 3b‑d].
Case 4
A 48‑year‑old male, known diabetic, presented with a
malodorous, discharging ulcer over the right toe since 1
month. [Figure 4a] Earlier culture of pus swab taken from
the center of the lesion yielded no growth. Clusters of
methicillin‑resistant Staphylococcus aureus were isolated
from an image‑guided swab taken from the periphery of the
ulcer [Figure 4b] The wound was surgically debrided with
repeated imaging till the device failed to show any fluorescence,
signifying appropriate debridement depth [Figure 4c].
Discussion
Studies have shown that significant bacterial load in wounds is
correctly identified only in 52.5% instances by clinical assessment
as against 74.6% with fluorescent imaging‑based techniques.6
Antibiotic resistance, a growing global concern can be prevented,
at least in part, by antimicrobial stewardship practices.7,8 All of
the key factors listed by the European Wound Management
Association contributing to antimicrobial misuse in wound care,
Figure 1b: Lack of fluorescence on imaging with “Illuminate”
Shah, et al.
namely— diagnostic uncertainty regarding presence of bacterial
infection, clinician ignorance regarding class and duration of
antibiotic to be used, and patient demands for prescription of
inappropriate antibiotics can be overcome by a point‑of‑care
device providing immediate guidance.9
Our case series underscores that real‑time fluorescence
imaging of microorganisms can greatly facilitate the
appropriate use of antibiotics. The patient with pyoderma
gangrenosum, having received unnecessary and prolonged
antibiotic therapy due to a misleading clinical appearance,
responded promptly to systemic steroids, due to the instant
result provided by this device.
Figure 2a: Ulcer on the great toe with slough
Figure 2c: Fluorescence highlighting areas of bacterial colonization with
gram‑positive species (red) and Pseudomonas
Indian Journal of Dermatology, Venereology and Leprology | Volume XX | Issue XX | Month 2020
Fluorescence imaging in antimicrobial stewardship
Swabbing from inappropriate sites demonstrated false‑negative
culture results in cases 2 and 4. Sapico et al. reported only a
63% concordance of isolates between skin biopsy samples
taken from the periphery and in the center of 25 pressure
sores.10
The resistant‑to‑local‑cleaning periphery of wounds was
suggested to be a better swabbing site for accurate detection
of isolates in some cases. Also, a mixed infection may go
undetected by a single swab taken even from the area
of maximal clinical involvement. Image‑guided sample
collection overcomes this problem.
Figure 2b: Composite algorithmic image after processing demonstrating
co‑colonization
Figure 3a: Encrusted ulcer on the left shin
3
Shah, et al.
Figure 3b: Fluorescence at initial assessment
The reduced bioburden following appropriate antibiotics/
wound debridement seen on serial imaging in cases 3 and 4
not only led to rapid healing but also supplanted the subjective
clinical and/or time‑consuming microbiological assessment.
Illuminate leverages the autofluorescence property
exhibited by pathogens [such as nicotinamide adenine
dinucleotide hydrogen  (NADH), flavins] and infectious
markers (pyoverdine, porphyrin) present in bacteria and
fungus. NADH has a strong emission peak at 470 nm
producing blue fluorescence, while flavin and porphyrin peak
at around 525 nm and 620 nm corresponding to green and red
fluorescence, respectively. Gram‑negative organisms have
a higher intensity of NADH and flavin than gram‑positive
organisms thus making it possible to differentiate between
gram types. Pseudomonas possesses pyoverdine, a unique
marker having blue‑green color enabling its detection.
An image processing and machine learning algorithm
processes the spectral images to detect and classify pathogens
based on the intensity of autofluorescence variation amongst
these biomarkers. It can detect bacterial loads as less as
103 CFU/gm (colony‑forming unit per gram) allowing
for early intervention (Indian patent No. 323440, 2019).
Contaminants, if present, are usually at a lower density.
Contamination may also occur at later stages of sample
collection, processing, or in the laboratory.11
4 Indian Journal of Dermatology, Venereology and Leprology | Volume XX | Issue XX | Month 2020
Fluorescence imaging in antimicrobial stewardship
Figure 3c: Day 5 of follow-up
The skin itself has a natural fluorescence contributed by
collagen/elastin; bacteria interact with the tissue via heme
pathway and produce porphyrin displayed as red fluorescence
while fungi have a characteristic blue fluorescence due to
chitin and dityrosine linkages. Pseudomonas aeruginosa
produces a greenish‑cyan fluorescence signal due to the
presence of pyoverdine. There are spectral and textural
changes between the green illumination of specific
organisms (especially Pseudomonas giving a cyan color) and
surrounding dermis, which enable their identification. The
device overlays this composite autofluorescence image on
regions of infections. Gram‑positive bacteria are displayed in
red and gram‑negative in green by comparing the difference
in autofluorescence intensity.
However, limitations of the technology exist in detecting
deep‑seated bacterial infections and differentiating pathogenic
from commensal bacteria. While an approximation of
significant bacterial load is possible; the determination of
accurate levels thereof still requires tissue culture. The device
also cannot act as a surrogate for culture and sensitivity,
which is often required to choose appropriate antibiotics in
the era of drug resistance.
Point‑of‑care devices may substantially aid the evaluation of
wounds and help institute appropriate treatment promptly by
instant confirmation of infective etiology thereby reducing
Shah, et al. Fluorescence imaging in antimicrobial stewardship

Figure 4a: Ulcer over the right great toe with serous discharge

Figure 3d: Day 10 of follow-up

Figure 4c: Reduction in fluorescence evident after the initial cleaning

Declaration of patient consent

The authors certify that they have obtained all appropriate
patient consent.

Financial support and sponsorship

Nil.

Figure 4b: Highlighted areas showing clusters of bacterial colonization Conflicts of interest
primarily over the periphery of the wound
There are no conflicts of interest.

the investigative/financial burden, especially in resource‑poor References

settings where facilities for microbiological evaluation may 1. Singh  B, Singh  S, Khichy  S, Ghatge  A. Clinical presentation of
not be available. soft‑tissue infections and its management: A study of 100 cases. Niger

Indian Journal of Dermatology, Venereology and Leprology | Volume XX | Issue XX | Month 2020 5
Shah, et al.
J Surg 2017;23:86‑91.
2. Gupta N, Gupta SK, Shukla VK, Singh SP. An Indian community‑based
epidemiological study of wounds. J Wound Care 2004;13:323‑5.
3. Cheng CF, Sahu D, Tsen F, Zhao Z, Fan J, Kim R, et al. A fragment
of secreted Hsp90α carries properties that enable it to accelerate
effectively both acute and diabetic wound healing in mice. J Clin Invest
2011;121:4348‑61.
4. Bowler  PG, Duerden  BI, Armstrong  DG. Wound microbiology and
associated approaches to wound management. Clin Microbiol Rev
2001;14:244‑69.
5. Spear M. When and how to culture a chronic wound: A culture is a
valuable tool in wound care if used correctly. Wound Care Advisor
2014;3:23‑5.
6. DaCosta RS, Kulbatski I, Lindvere‑Teene L, Starr D, Blackmore K,
Silver JI, et al. Point‑of‑care autofluorescence imaging for real‑time
sampling and treatment guidance of bioburden in chronic wounds:
6 Indian Journal of Dermatology, Venereology and Leprology | Volume XX | Issue XX | Month 2020
Fluorescence imaging in antimicrobial stewardship
First‑in‑human results. PLoS One 2015;10:e0116623.
7. Shallcross LJ, Howard SJ, Fowler T, Davies SC. Tackling the threat
of antimicrobial resistance: From policy to sustainable action. Philos
Trans R Soc Lond B Biol Sci 2015;370 (1670):20140082.
8. Bell  BG, Schellevis  F, Stobberingh  E, Goossens  H, Pringle  M.
A  systematic review and meta‑analysis of the effects of antibiotic
consumption on antibiotic resistance. BMC Infect Dis 2014;14:13.
9. Lipsky BA, Dryden M, Gottrup F, Nathwani D, Seaton RA, Stryja J.
Antimicrobial stewardship in wound care: A Position Paper from the
British Society for Antimicrobial Chemotherapy and European Wound
Management Association. J Antimicrob Chemother 2016;71:3026‑35.
10. Sapico FL, Ginunas VJ, Thornhill‑Joynes M, Canawati HN, Capen DA,
Klein NE, et al. Quantitative microbiology of pressure sores in different
stages of healing. Diagn Microbiol Infect Dis 1986;5:31‑8.
11. Radhakrishnan  G, King  J, Meenatchi  U, Gupta A. Indian Patent No.
323440. IP India; 2019.
Indian Journal of Surgery
https://doi.org/10.1007/s12262-021-03190-6

ORIGINAL ARTICLE

A New Device and Technology for Detecting Bacterial Infection and its

Gram Type in Diabetic Foot Ulcer
Janakrai N. Parekh1 · Payal Soni2 · Mahendra Kumar Meena3 · Chetan Kumar Tandel3 · Geethanjali Radhakrishanan4

Received: 6 January 2021 / Accepted: 24 November 2021

© Association of Surgeons of India 2021

Abstract
The correct diagnosis and classification of bacterial infection are important for initiating appropriate antibiotic therapy
thus leading to a reduction in the incidence of antibiotic-resistant microbial infection. There are many methods to detect
wound bacterial infections. The gold standard method, at present, remains the tissue/swab culture along with a sensitivity
test. However, new fluorescence-based methods of detecting bacterial infection in a POC (point of care) setting have proved
themselves to be simple, prompt, accurate, and affordable. Many commonly infecting bacteria have a characteristic emission
auto fluorescence when excited with ultraviolet and blue light. A newly developed hand-held device captures the emitted
fluorescence and records the spectral signature of different bacteria. The fluorescence is exhibited because of the metabolic
and infectious biomarkers present in a bacterium, e.g., pseudomonas produces pyocyanin, and pyoverdine. Finally, an in-
built image processing and machine learning algorithm detects this native fluorescence produced by the bacteria and helps
classify them according to their Gram type. We studied 100 patients with diabetic foot ulcers, comparing the results of the
routine tissue culture with the Gram type report generated at POC by this novel device and found that this device was able
to detect infection with 99.24% sensitivity and 82.35% specificity with that of the gold standard culture test.

Keywords  Auto fluorescence · Non-invasive infection screening · Wound infection · Gram type bacteria classification

Introduction administer the right antibiotics at the right time; in other

words, to develop a POC (point of care) diagnostic test that
The prevalence of antimicrobial resistance remains a major helps identify the presence of bacterial infection [2]. A rapid
problem all over the world. Its potential impact on health identification of bacteria in wounds, blood, or body excre-
care is as significant as that of global warming [1]. Many tions will lead to treatment directed towards the specific
challenges exist in the way of overcoming antimicrobial pathogen.
resistance, with the most important being the ‘creation of At present, the most common method to detect wound
a cost effective, accurate, rapid, and easy to use method for infection is to take a deep tissue biopsy/swab from the
detection of bacterial infections that allows clinicians to ulcer and perform a culture and sensitivity test. For this, a
BSL-2 facility is required in the microbiology laboratory
and it takes about 24 to 48 h to know the status of the infec-
* Chetan Kumar Tandel tion and the type of bacteria. Though many phenotypic and
deptofsurgeryvalsad@gmail.com molecular techniques for microbial detection are available,
Janakrai N. Parekh they are time-consuming, reagent intensive, and require
janak.nvs@gmail.com trained personnel, not making it suitable for rapid screen-
1 ing. Recent advancements leveraging native fluorescence of
Department of Surgery, GMERS Medical College, Valsad,
Gujarat 396001, India bacterial cellular molecules have been found to be sensitive
2 and effective for the rapid screening of bacterial infection
Department of Microbiology, GMERS Medical College,
Valsad, Gujarat 396001, India and classification [3].
3 A startup based in India has developed and patented
Department of Surgery, GMERS Medical College, Valsad,
Gujarat 396001, India (IN Patent–323,440) a novel handheld POC device, (Illu-
4 minate®), which utilizes auto fluorescence of bacteria for
Adiuvo Diagnostics, Chennai, India

13
Vol.:(0123456789)

Fig. 1  Illuminate: image of the
novel handheld device
its detection [4] (Fig. 1). Each bacterium has characteristic
emission fluorescence when excited with different wave-
length of light [5, 6]. The device uses multispectral imaging
combined with an advanced computational algorithm and
proprietary state of the art AI engine for spatial mapping and
classification of pathogens. The device captures the spectral
signatures of metabolic growth markers along with markers
released when a micro-organism causes infection, to detect
and assess the bacterial Gram type. The device leverages the
auto fluorescence property exhibited by pathogens [such as
nicotinamide adenine dinucleotide hydrogen (NADH), fla-
vins] and infectious markers [Pyoverdine, Porphyrin] present
in bacteria and fungus [16] Use of such handheld device
has been done successfully in diagnosing skin and soft tis-
sue infection—bacterial and fungal infection—and has been
reported in literature also [7, 8].
This newly developed hand-held device takes non-contact
images of the ulcer, preferably in a dark room or under a
dark hood, from 7 to 10 cm away. The device detects the
presence or absence of infection in the wound, maps the
exact site of infection spatially and classifies the Gram type
of bacteria within 2 min. The device can also measure the
dimensions of the ulcer, so wound healing can be measured
during follow-up visits. The test report can be immediately
generated, and the ‘test’ data can be saved and transferred
either using a USB drive or through the cloud, connecting
with the hospital records.
A clinical study on 100 patients of diabetic foot ulcers
was conducted using this novel handheld device to detect the
presence of infection, classify infecting bacteria according to
13
Indian Journal of Surgery
their Gram type and compare the results obtained with the
results of standard tissue culture method.
Patients and Methods
This study was carried out at GMERS Medical College &
Hospital, Valsad (Gujarat) by the Department of Surgery
and department of Microbiology, between April 2018 and
November 2019. Permission from the Institutional Human
Ethics Committee was obtained for conducting this study
(No. MCV/IHEC/10/18) and the study was registered
with CTRI (Clinical Trial Registry of India) (Reg. No
CTRI/2018/10/016147).
Though the device can be used for imaging any wound,
we decided to include only diabetic foot ulcer patients
for this study as they are most seen patients in General
Surgery OPD and wards in any teaching hospital. These
patients also remain in ward for many days enabling easier
follow-ups.
Male and female patients, from the OPD or ward, of
all age groups, having ulcer(s) on the foot/leg along with
diabetes were included in this study. Patients unwilling to
give consent or uncooperative in nature were excluded.
Pregnant patients and patients on chemotherapy were
excluded from the study as follow-up in these patients
sometimes becomes difficult.
All foot ulcers were clinically examined, washed with
saline and a deep tissue biopsy was taken from a loca-
tion which was suspected to be infected by the clinician
on visual inspection by either a small scoop or a no. 22
Indian Journal of Surgery
blade without the help of the device. After this, the ulcer
was imaged by this device either in a dark room or under
a black hood and another tissue sample was taken from
the site where the device showed maximum infection.
These samples are (1) blindly taken and (2) the device-
guided samples were sent to the microbiology laboratory
for culture and sensitivity testing. It is to be noted that the
microbiologist did not know the device result to avoid any
bias. The Gram type displayed immediately by the device
was recorded in a prepared proforma. After 48 h, culture
reports were compared with the device report (Gram type).
Figs. 2 and 3 show the images taken by the device and
clearly indicating the site of infection, the Gram type of
bacteria infecting the wound and size of the ulcer.
Fig. 2  Example of Gram-nega-
tive bacteria-infected wound
Fig. 3  Example of Gram-posi-
tive bacteria-infected wound
A total of 100 patients (88 males and 12 females) with
104 wound sites were recruited for the study and a total of
163 session images were taken; some ulcers needed to be
imaged after few days of antibiotics therapy when the treat-
ing doctor clinically felt that infection was still not under
control. Some patients were reimaged after debridement of
the ulcer while some patients had multiple ulcers. Thus, all
ulcers were imaged and tissues were taken separately from
each ulcer.
Out of the 163 session images taken, 149 sessions were
considered for study as 14 images had to be removed due
to poor imaging (6 sessions), presence of a bony region (3
sessions), and/or interference due to betadine (5 sessions)
on the wound.
13
 Indian Journal of Surgery

Table 1  Comparison of results obtained by standard culture method and 2 show the comparison of Gram type obtained in 149
and by new device samples by culture method and by the device. The device
Gram type Results of Gram type Results of Gram type imaged report displayed Gram-negative bacteria in green
obtained by culture obtained by new overlay and Gram-positive as red overlay. If two types of
method device Gram-negative bacteria were present in the wound, the
Gram positive 11 20 device depicted them as Gram-negative bacteria only as
Gram negative 119 112 it does not differentiate the different types of species of
No growth 17 15 bacteria. This is true for Gram-positive bacteria also. In
Polymicrobial 02 02 case of polymicrobial infection comprising of both Gram-
Total 149 149 positive and Gram-negative bacteria, the device displays
the Gram type of more abundant bacteria in the region.
Sensitivity of the new device is 99.24%, which means
Totally from 104 sites tissues were taken blindly without that the diagnosis of the bacterial wound infection by the
use of the device by the clinician from the area of the ulcer new device will be 99.24% correct.
where he suspected infection. After this, same ulcer was Specificity of the device is 82.35%, which correctly
imaged by the device and tissues were taken from the sites identify those samples without infection.
on ulcer, where the device was indicating infection spatially. Cohen’s Kappa (K) was also tested to determine the
Forty-five more sites were detected by the device where clini- agreement between culture method and new device
cally there was no suspicion of infection. So totally, 149 tis- method. We found K = 0.86 (P < 0.001), which means
sue samples were taken under the device guidance and totally highly significant agreement was seen between the cul-
149 tissue samples were sent to microbiology department for ture method and the new device method. So, new device
culture and sensitivity testing. The results obtained by the perfectly diagnoses the bacterial presents.
device in all 149 cases about the Gram type of infection were Table  3 shows association among Gram-negative,
recorded in proforma. Microbiologist was not aware of the Gram-positive, and no growth by culture and device
Gram type of bacteria identified by the device. method.
(There were 2 polymicrobial infection sites diagnosed
by culture method and by device method; so, for easy cal-
Results culation 1 is added into Total calculation of Gram-positive
and 1 is added into total calculation of Gram-negative
The culture report of 149 tissue samples were obtained bacteria).
from the microbiology department and compared with Cramer’s V is used to measure the strength of the asso-
the Gram type of infection shown by the device. Tables 1 ciation which is 0.765 (P value < 0.01).

Table 2  Screening test culture Screening test culture (standard)/new device Growth according to culture report
(standard)/new device
Growth seen No growth seen Total

Growth according to new device Growth seen 131 (T.P) 3 (F.P) 134
No growth seen 1 (F.N) 14 (T.N) 15
Total 132 17 149
Sensitivity =  T.P/(T.P + F.P) 99.24%
Specificity =  T.N/(F.N + T.N) 82.35%
Positivity predictive value =  T.P/(T.P + F.N) 97.76%
Negative predictive value =  T.N/(F.P + T.N) 93.33%

Table 3  Comparison of Culture (standard)/new device Growth according to culture report

detection of Gram-negative,
Gram-positive, and no growth Gram negative Gram positive No growth
by the new device and culture
method Growth according to new device Gram negative 109 (90.8%) 1 (8.3%) 3 (17.6%)
Gram positive 10 (8.3%) 11 (91.7%) 0 (0.0%)
No Growth 1 (0.8%) 0 (0.0%) 14 (82.4%)

P value < 0.001, highly significant association.

13
Indian Journal of Surgery
There is highly significant association between the cul-
ture method and new device method for diagnosing infec-
tion and Gram type of bacteria. We applied Cramer’s V test
to find out strength of association between culture method
and new device method.
Discussion
Bacterial identification is a growing field of interest in
microbiology. Traditional methods of bacterial identifica-
tion include simple biochemical tests like catalase, oxidase,
and substrate utilization test that mostly rely on phenotypic
identification [5]. On the other hand, modern methods for
microbial identification include PCR, microarray-based
identification, immunological methods, chemical analytical
methods etc. [2]. Modern methods usually complement but
may sometimes replace traditional methods [6].
One promising method for detection of bacterial infec-
tion is by using the inherent fluorescence of bacteria [3].
Previous studies have shown fluorescence-based detection
to be sensitive and effective for food born and environmen-
tal microorganisms and have even been able to distinguish
between different cell types. However, this powerful tech-
nique is not available for large-scale deployment in clinical
use to detect infection [3].
Fluorescence techniques are well developed in flow
cytometry [9]. Flow cytometry utilizes either the intrinsic
fluorescence (auto fluorescence) of cellular molecules or
requires the addition of extrinsic fluorescent stains or anti-
body probes and is being increasingly employed for rou-
tine analysis in environmental and industrial microbiology
[10, 11]. Fluorescence techniques can identify potentially
pathogenic or toxic microorganisms in environmental water
and foods [12, 13]. More specifically, within medicine,
fluorescence characterization has provided insights in the
development and progression of cancerous tissue [14, 15].
Giana et al. (2003) [7] successfully discriminated between
the clinically important Escherichia coli., Enterococcus
faecalis, and Staphylococcus aureus using fluorescence
methods.
Adiuvo Diagnostics Pvt. Ltd., Chennai, has used the prin-
ciples of auto fluorescence of clinically important bacteria
to make a hand-held device (Illuminate) and we at GMERS
Medical College and Hospital, Valsad, conducted this study
for validation of the said device. The makers of the device
first quantified the complete fluorescent response of many
clinically relevant bacteria to identify the most promising
fluorescent features. In the future, more data will lead to
the addition of more wavelengths of light and making the
real-time detection of the species of the bacteria possible
along with fungus detection as well. More variants of the
hand-held device can be constructed to detect presence of
bacteria on various surfaces like OT table, anesthesia trolley,
or in ICU like railing of bed, B.P cuff, and nurse’s computer.
This screening device has multiple advantages.
(1) The presence of infection in the ulcer and Gram type
of the infecting bacteria can be immediately diagnosed.
So, appropriate antibiotics based on Gram type of bac-
teria can be started using the hospital antibiogram.
(2) The device can automatically measure the size of the
wound, helping the clinician during follow up. The
treating doctor can immediately know the status of
infection and the reduction in the size of the ulcer, if
any.
(3) It can detect bacterial loads as less as ­103 CFU/gm
(Colony forming unit per gram) allowing for early
intervention [8].
(4) In our study, when tissues sample were taken under the
guidance of device, the report showing “no growth”
were significantly less. “No growth” was reported in
28 patients when tissues were taken blindly without
using the device and only 15 patients when tissues were
taken under guidance of the device. This difference is
statistically significant. Overall accuracy of the device
is 93.20% in diagnosing Gram type of infection.
(5) The device is useful for getting immediate feedback
during debridement of an infected wound in OT.
(6) Before grafting the partial thickness skin graft on recip-
ient area, the device can screen the area and show any
infection pocket on the recipient area, preventing graft
rejection.
The device has certain disadvantages.
(1) Dark room or black hoods required for imaging the
wound. Wound cannot be imaged in presence of ambi-
ent light.
(2) Device cannot determine the species of the bacteria at
present, whereas standard culture method can tell us
the species of the bacteria while the device can only
differentiate infecting bacteria into Gram-negative or
Gram-positive.
(3) When there is polymicrobial infection the device can
not differentiate various species of bacteria and it will
show the Gram type of abundant bacteria. It becomes
difficult to calculate exactly the number of polymicro-
bial infections in a particular wound.
Conclusion
From the many newly developed techniques for detection of
bacterial infection, techniques using fluorescence character-
istic of bacteria can be simple, cost-effective, and speedy.
13

This newly developed device is a prime example of the same.
At present, this device shows only Gram type of bacteria and
not the species of bacteria. In the future with data available,
the algorithm in the device will also be capable of differen-
tiating between the genus/species of infecting bacteria. The
accuracy of the device is around 93% which is acceptable.
Health care industry should have the advantage of deploying
this device in hospitals or the clinic.
Acknowledgements  We are thankful to Dr. Diwakar Sharma, Stat-
istician and Assistant Professor, Community Medicine Department,
GMERS Medical College, Valsad, for helping us to prepare the
manuscript.
References
1. Sally D, Fowler T, Watson J, Livermone D, Walker D (2013)
Annual report of the chief medical officer: Infection and the rise
of antimicrobial resistance. Lancet 381(9878):1000–1609
2. Varadi L, Luo Jia Lin, Hins DE, Perry JD, Anderson RJ, Orenga S,
Paul W (2017) Methods for detection and identification of patho-
genic bacteria Ground water chem. Soc Rev 46:4818–32
3. Dartnell LR, Roberts JA, Moore G, Ward JM, Mullet JP (2013)
Fluorescence characterization of clinically Important bacteria.
PloS One 8(9):e75270. www.​ploso​ne.​orgse​pt2013/​vol.8/​issue9/​
e75270
4. Amor MS (2007) Recent advances in the use of intrinsic fluores-
cence for bacterial identification and Characterization. J Fluoresc
17(5):455–59. https://​doi.​org/​10.​1007/​s10895-​007-​0180-6
5. Pj Modeiros, Weixu Liztein XU, Wilson BC, Rosen C, Linden R
(2015) Point of care autofluorescence imaging for real time sam-
pling and treatment guidance of bio burden in chronic wounds:
first in human results. PLOS One 10(3):9
6. Murihead K, Horan G (1985) Flow cytometry: Present and future.
Nat Biotechnol 3:56. https://​doi.​org/​10.​1038/​nbt04​85-​337
7. Radhakrishnan Geethajali, Gupta Ayush, King John, Ganvir
Davina (2020) Rapid handheld screening device to detect skin
and soft tissue infection. Proc SPIE 2020:11211. https://​doi.​org/​
10.​1117/​12.​25464​04
13
Indian Journal of Surgery
8. Shah BM, Ganvir D, Sharma Misra SB, Misra RN, Kothari P
et al (2020) Utility of a real –time fluorescence imaging device in
guiding antibiotic treatment in superficial skin infections. Indian
J Dermatol Venereol leprol 2020(1):6
9. Patra D, Mishra AK (2001) Investigation on simultaneous analysis
of multi component polycyclic aromatic Hydrocarbon mixtures
in water samples: a simple synchronous fluometric method. Tal-
anta 55:143–153. https://​doi.​org/​10.​1016/​50039-​9140(01)​00404.​
PubMe​d1896​3596
10 Chapman GV (1993) Instrumentation for flow cytometry. J Immu-
nol Methods 243:3–12. https://​doi.​org/​10.​1016/​50022-​1759(00)​
224-6
11. Vermes I, Hannen C, Reutelingsperger C, (2000) flow cytometry
of apoptotic cell death. J Immunol Methods 243:167–190. https://​
doi.​org/​10.​1016/​50022-​1759(00)​PubMe​d1098​61414
12. Zigarmann M, Abert M, Miller M, Frimmel FH (2010) use of
fluorescence fingerprints for the estimation of Bloom forma-
tion and toxin production of microcystis aeruginosa. Water Res
444:195–204. https://​doi.​org/​10.​1016/j.​waters.​2009.​09.​035 (Pub
Med 19818983s)
13. Georgakoudi I, Jacobson BC, Miiler MG, sheets EE, Badizadegan
k etal, (2002) NAD(P)H and collagen as in vivo Quantitative fluo-
rescent Biomarkers of Epithelial precancerous changes. Cancer
Res 62:682–687
14. Diagaradjane P, yaseen MA, yu J, wong MS, Anvari B, (2005)
Autofluorescence characterization for the early diagnosis of neo-
plastic changes in DMBA/TPA induced mouse skin carcinogen-
esis. Lasers Surg Med 37:382–395. https://​doi.​org/​10.​1002/​15m.​
20248.​PubMe​d1624​0416
15. Giana H, silveria L, Zangaro R, Pacheco MTT, (2003) Rapid iden-
tification of bacterial species by fluorescence spectroscopy and
classification through principal components Analysis. J fluoresce
13:489–493. https://​doi.​org/​10.​1023/​13JOF​L0000​008059.​74052.​
36
Publisher's Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
 PROCEEDINGS OF SPIE
SPIEDigitalLibrary.org/conference-proceedings-of-spie

Rapid handheld screening device to

detect skin and soft tissue infections

Radhakrishnan, Geethanjali, Gupta, Aayush, King, John,

Ganvir, Devina

Geethanjali Radhakrishnan, Aayush Gupta, John King, Devina Ganvir, "Rapid

handheld screening device to detect skin and soft tissue infections," Proc.
SPIE 11211, Photonics in Dermatology and Plastic Surgery 2020, 112110V
(19 February 2020); doi: 10.1117/12.2546404

Event: SPIE BiOS, 2020, San Francisco, California, United States

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 03 Mar 2020 Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

 Rapid Handheld Screening Device to Detect Skin and Soft Tissue
Infections
Geethanjali Radhakrishnana, Aayush Guptab, John Kinga, Devina Ganvirb
a
Adiuvo diagnostics private limited, Chennai, India, Chennai, India
b
Department of Dermatology, Dr DY Patil Hospital, Pune

ABSTRACT

Skin and soft tissue infections (SSTIs) are one of the most common infections in India affecting 10-12% of Indian
population. They are caused by a variety of bacteria and fungus, which makes it harder to diagnose and propose an effective
treatment immediately especially in low resource settings due to the lack of access to qualified physicians. Management
of SSTIs requires early expert infection assessment and remains a major challenge for the clinicians. A hand-held device
is developed leveraging the inherent autofluorescence properties of the bacterial and fungal species that can non-invasively
and rapidly identify the pathogens on SSTI using multispectral imaging followed by image processing and machine
learning algorithms. The device can classify the gram type of bacteria with > 85% accuracy.

Keywords: Multispectral imaging, SSTI, Wound, Handheld device, Multispectral imaging, Machine Learning, Gram
type classification, autofluorescence1

1. INTRODUCTION

The principal barrier against microbial invasion is the skin [1]. It provides the first line of defense and interacts with the
external environment constantly and is colonized with a diverse population of microbes. Gram-positive species such as
Staphylococcus epidermidis, Corynebacterium species, Staphylococcus aureus and Streptococcus pyogenes are the typical
species that colonize the skin above the waistline. Both gram positive and gram-negative bacteria colonize the skin below
the waistline.

Skin and soft tissue infections (SSTIs) which includes all pathologic infections of the skin, range from simple impetigo to
rapidly progressive necrotizing fasciitis [2], foot ulcers, surgical infections and so on. Diagnosing the exact extent and
causative organism of the disease is critical for the successful management of a patient through successfully treating the
soft tissue infection [3].

SSTI has an incidence rate of about 24.6 per 1000 person per year along with a prevalence rate in hospital of 7% to 10 %
[4]. It’s important to identify cases which require immediate intervention to the ones that are less severe thus presenting
diverse challenges. The problem is more compounded in patients who have diabetes mellitus and AIDS where a mild
infection can easily progress into a rapidly advancing life-threatening condition. Chronic ulcers, like those associated with
neuropathy are also associated with unique challenges. Hence, infection care remains a major clinical challenge and
presents an enormous burden to health care worldwide. It is important to understand the infection immediately when a
patient is presented with SSTI, identify the causative organism to deal with them effectively.

Currently, assessments of skin and soft tissue infections and determining the underlying pathogens causing infection is
typically visual inspection based followed by swab analysis and culture method. Culture method is considered the gold
standard technique, but the process is laborious, time-consuming, and requires a BSL-2 facility for identification thus
leading to prescription of generic antibiotics resulting in antimicrobial resistance. In case of fungus, the turnaround time

Geethanjali@adiuvodiagnostics.com; www.adiuvodiagnostics.com, Phone +91-8015316313

Photonics in Dermatology and Plastic Surgery 2020, edited by Bernard Choi,

Haishan Zeng, Proc. of SPIE Vol. 11211, 112110V · © 2020 SPIE
CCC code: 1605-7422/20/$21 · doi: 10.1117/12.2546404

Proc. of SPIE Vol. 11211 112110V-1

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 03 Mar 2020
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 for a culture technique is even longer and requires a dermatologist for assessment. Only 6000 dermatologists cater to 121
crore population [5] making it difficult for diagnosis in low resource setting areas. Recently PCR is used for rapid detection,
but they require trained physicians and cannot be used for continuous monitoring of an SSTI. Commercially available
wound management devices do not provide comprehensive information about specific pathogens infecting the wound and
are expensive. There is a huge unmet need to understand pathophysiology of a wound at an early stage, especially
complications of diabetes foot ulcers leading to increased risk of chronic wounds and amputations. Similarly, hospital
acquired infections at surgical sites are becoming increasingly prevalent and needs preventive measures.

We propose a non-invasive device that can be used in real-time to detect the causative microorganism gram type and
continuously monitor the SSTI along with their corresponding pathogen load to achieve the above-mentioned unmet need
and improve the quality of life of patients.

Previous research findings show that bacteria have characteristic native fluorescence contributed by metabolic markers
such as tryptophan, tyrosine, Nicotinamide Adenine Dinucleotide (NADH), Flavin molecules during its metabolic phase
and siderophore porphyrin during its infectious phase. It is also well known that Pseudomonas aeruginosa in addition
produces siderophore pyoverdine during its heme acquisition pathway [6]. Fungus has chitin and di-tyrosine linkages
which are autofluorescence in nature as well [7]. Metabolic signals, which are indicators of live cells, fluoresce in the blue-
green region. The fluorescence emitted is directly proportional to the concentration of metabolites and thus to the average
number of live cells [8]. Flavins which fluoresce in the green region and protoporphyrin IX, which fluoresce in the red
region are found in both live and dead cells.

Leveraging this, we wanted to further analyze the change in fluorescence intensity contributed by each clinically relevant
bacterium during its growth phase and while acquiring heme from host and if such an information can be used in classifying
the gram type of bacteria.

We have analyzed auto-fluorescence characteristics of clinically relevant bacteria and fungi. From the obtained data,
characteristic excitation and emission wavelengths are chosen to distinguish various bacteria into gram type and identify
few fungi. To aid the ease of use in low resource settings, we have developed an affordable multispectral imaging platform
capable of capturing images of the wound region. A clinical study was conducted using the hand-held device on SSTI
patients at Dr. DY Patil Hospital, Pune, India. The device was able to distinguish gram type of clinically relevant bacteria
and fungus with more than 85% accuracy.

2. FLUORESCENCE SPECTRA OF CLINICALLY RELEVANT PATHOGENS

An Excitation Emission Matrix Spectra (EEMS) is a three-dimensional graph to analyse fluorescence properties exhibited
by an unknown molecule at various excitation wavelengths [9]. Clinically relevant pathogens samples were obtained from
Microbial Type Culture Collection (MTCC), American Type Culture Collection (ATCC) and subcultures taken from
Patients and used for EEMS analysis. Gram staining was done before and after inoculation from the subcultures to ensure
no contamination. Clinically relevant pathogens that are most common in Indian settings were identified as Staphylococcus
aureus, Enterococcus faecalis, Escherichia coli, Klebsiella sp., Proteus sp., and Pseudomonas aeruginosa. In fig.1, we
have shown results obtained from Pseudomonas aeruginosa and Escherichia coli, belonging to the gram-negative type
and Staphylococcus aureus, a gram-positive bacterium and Malassezia sp. of fungal origin for comparison.

Both intracellular and extracellular metabolic biomarkers are known to cause autofluorescence on excitation with
wavelengths between 200 nm to 450 nm with emission wavelengths between 300 nm to 800 nm. The intensity of
fluorescence contributed by each biomarker is inferred from the EEMS recorded. Tryptophan, NADH, and Flavin
molecules contribute to the majority of autofluorescence in specified wavelength region [10].

Proc. of SPIE Vol. 11211 112110V-2

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 03 Mar 2020
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Pseudomonas aeruginosa Staphylococcus aureus

Malassezia sp.

Figure 1: Excitation Emission Matrix of clinically relevant bacteria and fungus.

Bacterial pathogens require iron-containing heme to cause the disease. Pathogens have elaborate strategies to acquire heme
from host, and this is required to cause an infection. To understand the biomarkers and any change in autofluorescence
intensity during heme acquisition pathways, these clinically relevant pathogens were induced with aminolaevulinic acid
(ALA). From Figure 2, one can understand that while pyoverdine is very evident in Pseudomonas aeruginosa, porphyrin
intensity is higher in Staphylococcus aureus, followed by a double peak between 400-500nm in Escherichia coli and
Malassezia sp. has significant porphyrin secretion between 550 nm-600 nm.

Proc. of SPIE Vol. 11211 112110V-3

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 03 Mar 2020
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Pseudomonas aeruginosa + ALA Staphylococcus aureus + ALA

Malassezia sp. + ALA

Figure 2: Excitation Emission Matrix of clinically relevant bacteria induced with ALA.

We compared the EEMS before and after ALA induction (see fig. 2) and compared the intensities contributed by
biomarkers: Flavin, NADH, Porphyrin and Pyoverdine. There were significant differences in Porphyrin production for
gram-positive, gram-negative bacteria1 and fungus. We used this information for classification of gram type of bacteria
and identification of fungus.

3. DEVELOPMENT OF MULTISPECTRAL FLUORESCENCE IMAGING DEVICE

Based on the identified excitation and emission wavelengths from EEMS graphs of various Gram positive, Gram negative
bacteria and Fungi, we have developed a portable handheld multispectral imaging device, referred to as “IlluminateR ”,
that illuminates the SSTI region with multiple excitation sources and capturing the emitted auto fluorescence.

3.1 Device Block Diagram:

The handheld device comprises of the following components: Fluorescence imaging module, Image acquisition module,
Image processing module, Display module and Control system.

Proc. of SPIE Vol. 11211 112110V-4

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 03 Mar 2020
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Fig. 3: Block diagram of handheld device for capturing auto fluorescence from the diabetic wound region

3.1.1 Fluorescence Imaging Module:

Fig. 4: Block diagram of Fluorescence Imaging Module

The components of fluorescence imaging module comprise of multi-wavelength UV LEDs (UV-A, UV-B and visible)
driven by a LED driver board, an optical filter wheel with emission filters over a CMOS imaging sensor forming a
multispectral imaging module. The filter wheel is rotated by a nano servo motor. The whole module is controlled by the
control system which sets excitation/emission wavelength in a sequence and captures multispectral image of the wound.

3.1.2 Control System:

The control system consists of a quad core ARMv8 Cluster and is implemented on a BCM2837 software. The control
system receives commands from the Display Module running on the Android Smart Phone. It controls the fluorescence
imaging module, image acquisition module, image processing module and display module.

3.1.3 Image Acquisition Module:

Image acquisition is done using 13MP CMOS sensor present on the Android smartphone. The frame rate and exposure for
images captured is set automatically by the control system prior to image acquisition.

Proc. of SPIE Vol. 11211 112110V-5

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 03 Mar 2020
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 3.1.4 Display module:

An application runs on the smartphone screen which enables user to enter patients details, set SSTI area on to the capture
window, evoke capture of images, lasso the wound region after capture and finally also to display colour code wound map
with details of infected region. Red to display gram positive bacteria, green to display gram negative bacteria, blue for
fungi. The interface also enables physician to share and print the report, retrieve patient data any time and for adding notes.

3.1.5 Image Processing Module:

The image processing module analyses the auto-fluorescence response collected from the spectral images and analyses
various biomolecular fluorescence intensities contributed by NADH, Flavin, pyoverdine, Chitin, porphyrin and any
unknown molecule. After the spectral images are captured, the image processor implements an image orientation mapper
to correct for the orientation mismatches. Each image is oriented and aligned with each other and the mean intensity in the
region of interest for each filter is calculated and subsequently used for machine learning.

3.1.6 Pathogen Classifier Model:

Fig. 5: Block diagram of the process flow implemented in the handheld device for pathogen classification.

A machine learning approach is followed in order to build the Pathogen classification model as shown in Figure 5. Spectral
images are captured from illuminate device and sent into a decision matrix step where the images are filtered based on hue
and threshold to extract autofluorescence regions from noise and other unwanted signals. The filtered regions are called
ROI, which are sent as a swab or tissue to microbiology lab and a report is obtained with organism name which is used for
labelling. The corresponding ROI spectra in fed into a Machine learning algorithm which predicts the pathogen gram type
and displays the result. In case of fungus a simple thresholding is done, and sent for ML.

Random Forest Classifier is used here with a 70:30 training/test data split. Data is shuffled randomly before applying the
random classifier. During the training and test split, the data is stratified to ensure equal percentage of each classes is
represented in the training and test set to avoid data bias. Five-fold cross validation is also done to calculate the accuracy
and standard deviation values.

4. CLINICAL STUDY TO VALIDATE THE MULTISPECTRAL IMAGING DEVICE

A clinical study was conducted in collaboration with Dr. DY Patil Hospital, Pune, India. The study was approved by
institutional ethics clearance “DYPU/EC/159/2018”and registered under Central Trial Registry of India (CTRI) with
register number “CTRI/2018/10/016147”.

Proc. of SPIE Vol. 11211 112110V-6

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 03 Mar 2020
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 It was a single arm interventional study where the device results were compared against a gold standard culture method in
case of bacteria or sent for KOH staining in case of fungus. A swab/deep tissue or skin scrapping were sent for microbiology
assessment in the region where the device shows presence of bacteria/fungi. The device report was compared against the
microbiology report. Patients with fungal infections, surgical site infections, contact Dermatitis, stasis ulcer both arterial
and venous, diabetic foot ulcer, sterile ulcers, any scars/suture site, sinuses/fistulas were included in the study. Patients
with inability to consent were excluded from the study.

Informed consent was obtained from 182 patients and were imaged using the device. 33 bacterial patients having various
type of SSTI and 149 dermatology related fungal infections as shown in (Table- A) were imaged using the device.

Species Number
CONS 1
Klebsiella sp. 3
MRSA 7
Pseudomonas sp. 11
Staphylococcus aureus 3
Propionibacterium sp. 4
Tinea sp. 81
P.versicolor 68
No infection – Sterile wounds 5

Table A: Summary of bacteria isolated through standard culture method.

The device was mounted on a tripod to avoid any vibration/shaking during the imaging and lights were turned off to
maintain a dark room setting and the device is held 10-15 cm from the wound region while imaging. For smaller diameter
wounds of ~5 cm2, the distance was less than 7 cm. Images were captured automatically in less than 30 seconds and an
additional 1 minute for the machine learning algorithm to display the results. The images were stored in the memory card
within the device. 70% ethanol was used to decontaminate the device before each imaging. Additionally, a black hood was
given if the lights cannot be turned off to minimize noise due to ambient lighting.

Series of images were taken with the “IlluminateR” imaging device on the affected SSTI region. At the end of imaging, a
spatially mapped color-coded image was displayed indicating possible presence of pathogen regions, where a swab/deep
tissue biopsy/scraping were taken and cross-validated against standard visual inspection followed by
microbiology/microscopy wherever applicable (See table 1).

Previously a machine learning model was trained on 121 bacterial patients and this model was ported into the device. This
model was able to predict 14 gram positive sites and 11 gram negatives with greater than 90% sensitivity as per results
shown in Table B. The device also picked up 1 polymicrobial site and showed no fluorescence in 5 patients where there
was no growth of bacteria in culture as well. Further tests were performed on the no infection patients and the physician
identified 2 patients associated with a rare condition called Pyoderma gangrenosum. The device was also used to check
bacterial presence on patients with Acne. The specificity was 100% and accuracy 93% in case of binary classification
between infection and no infection. In case of fungus, the accuracy is greater than 70% towards classification between
Tinea species and Pityriasis versicolor using morphological features alone. In future, accuracy will be improved combining
both spectral and morphological features. In the near future, the device needs will be tested against more reference images
of different skin colors to compensate for background autofluorescence to make the model more robust.

Proc. of SPIE Vol. 11211 112110V-7

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 03 Mar 2020
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Microbiology result Total no. of sites Imaging device result

Correct classification Incorrect classification

Gram positive bacteria 15 14 1

Gram negative bacteria 12 11 1

Co-colonization 1 1 0

Microbiology result Total no. of sites Imaging device result

Correct classification Incorrect classification

Presence of bacteria 28 28 0

Absence of bacteria 5 5 0

Table B: Results of the wound imaging for gram type classification, presence and absence of bacteria using the portable
handheld imaging device.

5. LIMITATIONS

Imaging needs dark room setup and extra care needs to be taken for no external light to interfere with imaging sequence
else the model predicts random artifacts as infections. Staples, fibers from cloth, sutures, bones and tendons has
autofluorescence on their own and leads to false fluorescence. The device has been fine-tuned with increased sensitivity
to pick up infections in open wounds. The device should only be used for imaging closed skin or aberrations with caution
due to high autofluorescence contributed by collagen and elastin on skin. Sometime edge artifacts contribute to false
overlay, where decision has to be made by comparing the spectral images.

6. CONCLUSION

The portable handheld device, IlluminateR, based on multispectral auto fluorescence imaging can be used as an important
tool in SSTIs for guided swabbing, assessment of an infection, understanding its microbiome pattern and aiding in the right
first line treatment during the golden hour of wound healing. The device helps to identify infected from non-infected
wounds through a color coded over lay image and classifies the infected bacterial wounds broadly according to their gram
type. The device also identifies fungal infection regions in dermatology related skin infections. The results of the study
with >85% accuracy demonstrates attractive potential of the device in any SSTI settings. Since the results of the gram type
classification are instantaneous, the device significantly helps doctors in effective first line antibiotic treatment thus
preventing antibiotic misuse. In future, the device will be tested on more patients and the model will be fine-tuned to
increase the accuracy.

Proc. of SPIE Vol. 11211 112110V-8

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 03 Mar 2020
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 Acknowledgments

We would like to thank Department of Dermatology, Microbiology, Surgery at Dr.DY Patil Hospital, Pune. We also thank
BIRAC, Department of Biotechnology, Government of India for the Grant aid and support to develop and validate the
device. We also thank Villgro Innovation Foundation and Venture Center, Pune for their constant guidance. NCL, Pune
and Golden Jubilee Biotech park for Woman, Chennai for giving access to their Spectrofluorometer.

REFERENCES

[1] Ki, V., & Rotstein, C. (2008). Bacterial Skin and Soft Tissue Infections in Adults: A Review of Their Epidemiology,
Pathogenesis, Diagnosis, Treatment and Site Of Care. Canadian Journal of Infectious Diseases and Medical
Microbiology, 19(2), 173–184. doi: 10.1155/2008/846453
[2] Park, Y. J., & Lee, H. K. (2018). The Role of Skin and Orogenital Microbiota in Protective Immunity and Chronic
Immune-Mediated Inflammatory Disease. Frontiers in Immunology, 8. doi: 10.3389/fimmu.2017.01955
[3] Park, Y. J., & Lee, H. K. (2018). The Role of Skin and Orogenital Microbiota in Protective Immunity and Chronic
Immune-Mediated Inflammatory Disease. Frontiers in Immunology, 8. doi: 10.3389/fimmu.2017.01955
[4] Leong, H. N., Kurup, A., Tan, M. Y., Kwa, A. L. H., Liau, K. H., & Wilcox, M. (2018). Management of complicated
skin and soft tissue infections with a special focus on the role of newer antibiotics. Infection and Drug Resistance, Volume
11, 1959–1974. doi: 10.2147/idr.s172366
[5] Skin diseases to grow in India by 2015: Report. (2014, May 2). BioSpectrum Bureau.
[6] Nitzan, Y., Salmon-Divon, M., Shporen, E., & Malik, Z. (2004). ALA induced photodynamic effects on Gram positive
and negative bacteria. Photochemical & Photobiological Sciences, 3(5), 430. doi: 10.1039/b315633h
[7] Smail, E. H., Briza, P., Panagos, A., & Berenfeld, L. (1995). Candida albicans cell walls contain the fluorescent cross-
linking amino acid dityrosine. Infection and Immunity, 63(10), 4078–4083. doi: 10.1128/iai.63.10.4078-4083.1995
[ 8 ] Kilungo, A. P., Carlton-Carew, N., & Powers, L. S. (2013). Continuous Real-time Detection of Microbial
Contamination in Water using Intrinsic Fluorescence. Journal of Biosensors & Bioelectronics, S12(01). doi: 10.4172/2155-
6210.s12-002
[9] Blough, N. V., & Vecchio, R. D. (2002). Chromophoric DOM in the Coastal Environment. Biogeochemistry of Marine
Dissolved Organic Matter, 509–546. doi: 10.1016/b978-012323841-2/50012-9
[10] Dartnell, L. R., Roberts, T. A., Moore, G., Ward, J. M., & Muller, J.-P. (2013). Fluorescence Characterization of
Clinically-Important Bacteria. PLoS ONE, 8(9). doi: 10.1371/journal.pone.0075270

Proc. of SPIE Vol. 11211 112110V-9

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 03 Mar 2020
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N
DOI: 10.35279/jowm2022.23.03.06

Clinical significance of a
novel imaging device to
evaluate infection on wounds:
Performance comparison
with culture method
and metagenome sequencing
Rajesh Kesavan1, Changam Sheela Sasikumar2
1 Department of Podiatric Surgery, Hycare Super Specialty Hospital, Chennai, Tamil Nadu, India. Adjunct Faculty, SRM Institute of Science and
Technology, Kattankulathur, Tamil Nadu, India.
2 Department of Clinical Research, Hycare Super Specialty Hospital, Chennai, Tamil Nadu, India. Managing Partner - SS Clini Research LLP, Adjunct Faculty,

Department of Biochemistry, Saveetha Dental College, Saveetha Institute of Medical & Technical Sciences, Saveetha University Chennai, Tamil Nadu, India.

Correspondence: Dr Rajesh Kesavan, hycareforwound@gmail.com

Conflict of interest: None

Keyword s:
Autofluorescence, diabetic foot ulcers, machine learning, pathogens, metagenome sequencing,
point-of-care device

ABSTRACT
Background
Diabetic foot and lower limb complications affect
40–60 million people globally. A rapid method is
needed to understand the bioburden and type of
infecting bacteria on diabetic foot ulcers.
Aim
To compare the accuracy and efficacy of a novel
multispectral imaging device, against the standard
culture method and correlate bacterial bioburden
levels with metagenome sequencing.
Method
A clinical study was conducted on diabetic foot ulcer
patients. Wounds, post-debridement, were imaged
using the multispectral imaging device and the re-
port, containing spatially mapped regions of bac-
terial burden along with their bacterial gram type,
was compared with the culture sensitivity report. Ad-
ditionally, metagenome sequencing was done on a
subset of the patient samples.
Results
A total of 157 patients were imaged, and 177 deep
tissue biopsies were taken from colour-coded regions
(Gram Positive/Gram Negative infected) identified by
the machine learning algorithm. The class-averaged
accuracy of the device was found to be 90.4% for
gram-positive, -negative, polymicrobial and for no
bacterial burden. A total of 26 biopsies were also
sent for 16S rRNA sequencing. Of these, cultures
were positive in 17 and correlated with the species
identified through 16S rRNA results in 16 cases. In
five cases where there was no growth in culture,
the multispectral imaging device could still detect
the presence of bacteria as confirmed by 16S rRNA
results (using 50 reads as a cut-off) and thus be
used as an adjunct for monitoring wounds’ bacterial
burden over time.
Conclusions
The novel multispectral imaging device can be used
to effectively understand the bioburden in a wound
of clinical significance and the gram type of infect-
ing bacteria.

182 J OU R NAL O F WO UN D M A N AG EM EN T
OFFIC IAL JOURNAL OF THE EU RO PEA N WO U N D M A N AGEM EN T A SSO C IAT IO N
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N
Implications for clinical practice
The autofluorescence imaging device can assist doc-
tors in evaluating wounds’ bacterial burden level,
spatial bioburden extent and the gram type of bac-
teria present, thus aiding in effective debridement
and proper wound-management protocols.
INTRODUCTION
Diabetes, one of the most prevalent chronic diseases,
is projected to affect more than 700 million peo-
ple globally by 2050 according to the International
Diabetes Federation (1). People in underdeveloped
and low-income countries are more prone to develop
diabetes, and about 12–25% of those are at risk of
developing diabetic foot ulcers as well, which can
further lead to chronic wounds, amputation and even
death (2-6). Wound bacterial burdens are known to
be initially composed predominantly of gram-posi-
tive bacteria (such as Methicillin-susceptible/resist-
ant Staphylococcus aureus, Streptococcus), but as the
infection progresses, they become colonised by gram-
negative bacteria (such as Pseudomonas aeruginosa,
Klebsiella species) (7-12).
Wounds frequently differ in their clinical character-
istics, making the diagnosis of infections through
signs and symptoms alone difficult (13). In hospital
settings, the current standard method for detecting
bacterial burden requires a swab or deep tissue biopsy
for culturing microorganisms, followed by biochemi-
cal methods for species recognition and antibiotic
susceptibility testing, which usually takes 3–5 days.
In short, these methods fail to provide the immediate
clinical information (such as the presence/absence
of bacterial burden and gram type of bacteria) re-
quired for first-line treatment (14). In addition, the
culture sensitivity methodology has other limitations,
including a bias to certain species, depending on the
choice of growth medium and incubating conditions;
a requirement for special apparatus for anaerobic bac-
teria; and an inability to pick up the full diversity of
infecting organisms (2). Accurate identification of
polymicrobial species infecting the wound can only
be achieved using genotypic methods such as16s
rRNA sequencing, shotgun sequencing and so on
(15-16). However, these methods are cumbersome,
costly and have yet to be adopted in mainstream clini-
cal practice (14).
The ability to offer targeted treatment during a first
consultation offers a tremendous advantage for timely
wound healing. With proper and timely management
J OU R NAL O F WO UN D M A N AG EM EN T
OFFI C IAL JOURNAL OF THE E U RO PEA N WO U N D M A N AG EM EN T A SSO C IAT IO N
of a wound, clinicians can achieve resolution in >90%
of mild to moderate soft tissue infections (7); there-
fore, there is a significant need for a modality that
focuses on the early assessment and classification of
pathogenic gram types infecting a wound.
Multispectral imaging, which involves shining mul-
tiple wavelengths of light and collecting the emission
response, has proven to be a useful technique for
identifying and classifying bacteria based on their
autofluorescence (19). In addition, autofluorescence
imaging is label-free, which is advantageous in terms
of reduced complexity and cost. Studies have shown
that bacteria have characteristic emission fluorescence
when excited in the UV and blue regions of light,
contributed by metabolic and infectious markers such
as NAD(P)H, flavins, porphyrins and pyoverdine (in
the case of Pseudomonas) and so on (19-20). Previous
studies have also shown that gram-positive bacteria
typically have more fluorescence intensity in the red
region, due to an increase in the release of porphyrins,
compared to gram-negative bacteria. Similarly, some
gram-negative bacteria have increased NAD(P)H and
flavins, when compared to gram-positive bacteria (21-
22). In addition, Pseudomonas aeruginosa, which is
one of the most predominant pathogens present in
diabetic foot infections (21), has clearly distinguish-
able autofluorescence contributed by pyoverdine.
These key differences in the relative concentration
of biomarkers and autofluorescence in various spec-
tral bands can potentially be used to design a rapid,
non-invasive diagnostic technique for the presence/
absence of bacterial burden and the classification of
the gram type of the bacteria infecting the wounds.
Fluorescence imaging has emerged as a promising
point-of-care technique for monitoring bacterial
burden in wounds (21,22). Fluorescence imaging,
in combination with clinical signs and symptoms,
has been used to effectively monitor wounds with
moderate-to-heavy bacterial loads (23). Similarly, sur-
gical sites with a high prevalence of bacterial burden
have been assessed with point-of-care fluorescence
imaging devices (24). These results show that when
the standard clinical signs and symptoms are assessed
in conjunction with the inputs from fluorescence im-
aging, bacterial burden diagnostic accuracy improves.
A bacterial fluorescence imaging system was also used
for effective wound debridement, which results in
accelerated wound healing (25). Further studies have
demonstrated that the use of fluorescence imaging
to monitor wounds results in reduced antibiotic use
(>30%), a reduction in antimicrobial wound dress- 
183
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N

Figure 1: Overview of Illuminate multi-spectral autofluorescence imaging platform for pathogen

detection and gram-type classification

Multiple Light source Biomarker - Emission  Machine Learning Algorithm

illumination Metabolic - NADH, Flavin compares biomarker intensity
Infectious - Porphyrin, Pyoverdine

Gram positive Gram negative

bacteria bacteria

ings (>45%) and improved wound healing rates
(>20%) over 12 weeks (26). Overall, the improved
healing rates are projected to reduce annual wound
care costs by 10% in National Healthcare System,
United Kingdom. Recently, guidelines were evolved
for the detection of bacterial burden in wounds using
fluorescence imaging based on the Delphi consensus
protocol (27).
The current point-of-care devices for fluorescence
imaging require careful interpretation. In addition to
autofluorescence from bacterial bioburden, human
tissue has its own autofluorescence, contributed to,
for example, by collagen or elastin, thus it is impera-
tive to distinguish pathogenic autofluorescence from
this background noise. Advanced machine learning
techniques trained on data from skin and wound
regions can overcome the above limitations by pro-
viding correction for these interfering factors.
Innovation
To the best of our knowledge, lluminate® is the first
device to use multispectral autofluorescence imag-
ing integrated with machine learning to provide spa-
tial–temporal information on the presence/absence
of bacterial burden and gram type of bacteria infect-
ing a wound and to provide the ability to monitor
a wound continuously (Figure 1). We conducted a

Table 1: Age, gender, and ulcer grade distribution

Age No. of participants (%) N = 157

≤ 35 years 4 (2.55 %)
36–45 years 16 (10.19 %)
46–55 years 39 (24.84 %)
56–65 years 44 (28.03 %)
≥66 years 54 (34.39 %)
Median 60
Q1 51
Q3 69
IQR 18
Gender (M: F) 104:53
Wound Classification
Grade 1 5
Grade 2 33
Grade 3 60
Grade 4 59

184 J OU R NAL O F WO UN D M A N AG EM EN T
OFFIC IAL JOURNAL OF THE EU RO PEA N WO U N D M A N AGEM EN T A SSO C IAT IO N
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N
Clinical Generated
Figure 2(a):
Gram negative bacteria infected wound
Patient ID: 6_64
Highlighted color-coded region is the Illuminate
prediction of the infection and gram type. Tissue
biopsy region is taken from this highlighted region
and the culture results are compared with the
Illuminate device result.
Device Report: Gram Negative Bacteria
Culture Report: Pseudomonas spp., Heavy
Growth.
Clinical Generated
Figure 2(c):
Gram Negative bacteria infected wound
Patient ID: 21_184
Device Report: Gram negative bacteria
Culture Report: Pseudomonas sp. - Heavy Growth
clinical study on diabetic foot ulcer patients to un-
derstand the clinical significance of this device for
bacterial burden identification and accuracy of the
gram type classification of bacteria in wounds, to aid
doctors in administering first-line treatment.
Clinical problem addressed
Currently, the assessment of wounds for the presence
J OU R NAL O F WO UN D M A N AG EM EN T
OFFI C IAL JOURNAL OF THE E U RO PEA N WO U N D M A N AG EM EN T A SSO C IAT IO N
Clinical Generated
Figure 2(b):
Gram negative bacteria infected wound
Patient ID: 15_24
Device Report: Gram Negative Bacteria
Culture Report: Klebsiella species - Moderate
growth.
Clinical Generated
Figure 2(d):
Gram positive bacteria infected wound
Patient ID: 20_34
Device report: Gram positive bacteria
Gram Staining Report: Gram positive cocci in pairs
& short chains
Culture Report: Enterococcus sp. - Scanty Growth
or absence of bacterial burden based on visual char-
acteristics alone is subjective and can be erroneous.
In addition, information on the gram type of bacteria
infecting the wound using standard microscopy and
culture tests is both time-consuming and reagent-
intensive. Understanding the bacterial burden and
infection status and monitoring these infections in
a wound over time is of the utmost importance for 
185
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N
Clinical Generated
Figure 2(e):
Gram positive bacteria infected wound
Patient ID: 22_30
Device Report: Gram positive
Culture Report: Staphylococcus sp. - Moderate
Growth
enabling wound closure. Further, identifying bacte-
rial burdens that are of clinical relevance is equally
important, as wounds have high loads of commen-
sals present. A bacterial load >104 CFU/g is usually
considered to be of clinical significance, and earlier
fluorescence imaging methods have been able to de-
tect bacterial burden above this value effectively (23).
With the imaging device, the algorithm thresholds
are adjusted to detect a bacterial load >104 CFU/g.
MATERIALS AND METHODS
Study design
An interventional single-arm comparative study
was carried out at a tertiary care centre in Chennai,
India after obtaining institutional ethics committee
approval (IEC 031#HYC/IEC/2018). The study was
registered with the Clinical Trial Registry of India
(Reg. No. CTRI/2018/10/016147). Adults (>18
years) diagnosed with diabetic foot ulcers and who
were willing to give their consent were included in the
study and imaged on their first visit, and during any
follow-up visits to the hospital. Patients’ diabetic sta-
tus was confirmed from various standard biochemical
tests that measure blood sugar levels. Patients with
infections under intact skin and/or with pre-existing
skin conditions, such as eczema or psoriasis in areas
close to the wounds, were excluded from the study.
Imaging procedure with the multispectral
autofluorescence imaging device
All multispectral images were collected using the de-
vice in either a dark room, or by covering the wound
with a black hood to minimise interference due to
186 J OU R NAL O F WO UN D M A N AG EM EN T
OFFIC IAL JOURNAL OF THE EU RO PEA N WO U N D M A N AGEM EN T A SSO C IAT IO N
Clinical Generated
Figure 2(f ):
Gram positive bacteria infected wound
Patient ID: 17_32
Device Report: Gram positive
Culture Report: Staphylococcus sp.
ambient light. Patient details were entered into the
device, and the wounds were cleaned with normal
saline solution before imaging. The device was held
10–12 cm from the wound region. A guide icon on
top of the application page, which uses information
from the distance sensor integrated into the device,
ensured that the required distance of ~10 cm was
maintained during imaging. The capture button was
then pressed, to record a series of >15 multispectral
autofluorescence images in <20 seconds. Upon suc-
cessful imaging, the user was prompted with a white
light clinical image to ‘lasso’ the region of interest
to assess the bacterial burden. The built-in image
processing (to compensate for the slight movement
of patients during the imaging and identify regions
of interest) and an edge-inferencing machine learn-
ing algorithm (to detect and classify bacteria causing
infections) then displayed a colour-coded overlaid
image of the wound, along with the infected region
(if any) and the gram type of bacteria. The device
also displayed the length, breadth and area of the
wound based on proprietary wound segmentation
algorithms. A guided swab/tissue sample was taken
from the infected region predicted by the device and
sent for microbial culture, and the presence/absence
of the bacterial burden, along with the gram type,
were compared with the report generated from the
device. Patients were imaged progressively, and a pro-
gressive report tracking/monitoring the wound was
also displayed to get a quick understanding of the
wound status during follow-up visits.
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N

Table 2: Organisms isolated from tissue biopsy samples

Organism type Occurrence in tissue samples

Pseudomonas spp. 40
Enterococcus spp. 22
Klebsiella spp. 20
E. coli 15
Proteus mirabilis 12
Morganella morganii 11
Staphylococcus aureus 13
Non-fermenting gram negative bacilli 6
Streptococcus spp. 7
Enterobacter spp. 5
Acinetobacter spp. 1
Aeromonas spp. 1
Citrobacter koseri 1
MRSA 1
Candida Spp. 1

Image processing and machine
learning algorithms
After collecting multispectral autofluorescence im-
ages, thresholding was applied to identify the regions
exhibiting higher autofluorescence, and spectral
intensity features were extracted from each of the
threshold images and sent for machine learning.
The algorithm’s image-processing thresholds were
adjusted to detect bacterial loads, typically >104
CFU/g. The machine learning algorithm based on
random forests (a highly accurate ensemble classi-
fier) was trained using an 80:20 data split ratio. The
hyperparameters, such as the number of trees, tree
depth, split criterion and so on, were optimised for
maximum accuracy. A five-fold cross-validation was
done to assess the accuracy of pathogen detection
and gram type classification.
After two minutes, the machine learning algorithm
could detect and classify the gram type of bacteria,
and the built-in application displayed a wound re-
port, as shown in Figure 2a–f. Both clinical and de-
vice’s post-processed wound images were displayed
for easier comparison. Gram-positive bacteria are
colour-coded in red, and gram-negative bacteria are
in green.
Specimen collection
Following debridement, a deep-tissue biopsy was
taken from the colour-coded region of the wound
area, as indicated by imaging device. Specimens were
placed in sterile transport containers and delivered to
the certified microbiology laboratory for gram stain-
ing, culture and antimicrobial susceptibility testing.
The laboratory personnel were blind to the findings
of the device. Some samples were also sent for 16s
rRNA sequencing, for comparison of the culture re-
sults for bacterial species identification. Samples were
also taken from regions of no fluorescence, to serve
as controls. No antimicrobial agents were adminis-
tered into the wounds before specimen collection.
Cultures
All tissue biopsies obtained post debridement were
cultured at a National Accreditation Board for Test-
ing and Calibration Laboratories-accredited labo-
ratory and analysed initially by gram-staining and
conventional aerobic methods by plating into various
enriched and differential growth medias. The colo-
nies were then sent for biochemical assay for species
recognition. Bacterial load was estimated by a plate
count semi-quantitative analysis and then graded as
scanty, light, moderate or heavy (1+, 2+, 3+ or 4+),
wherein moderate and heavy growth indicated a sig-
nificant bacterial load (i.e., greater than 100,000 per
gram of tissue). An antibiotics susceptibility profile
was then obtained using the disk diffusion method,
as per Clinical and Laboratory Standards Institute
guidelines.


J OU R NAL O F WO UN D M A N AG EM EN T
OFFI C IAL JOURNAL OF THE E U RO PEA N WO U N D M A N AG EM EN T A SSO C IAT IO N
187
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N

GN Table 3: Confusion matrix for infection and gram type classification

87 2 1 2
GP

3 19 1 8
True label
NG

16 2 24 8
GPGN

5 2 8 13

GN GP NG GPGN
Predicted label

16S rRNA sequencing
16S rRNA gene amplicon sequencing was performed
on a subset of 26 patient samples at a certified labo-
ratory. DNA was isolated from the tissue samples
using the EXpure microbial DNA isolation kit. PCR
was used for selective amplification using 27F and
1492R primers targeting the V3 and V4 regions.
Unincorporated PCR primers and dNTPs from the
PCR products were removed using the Montage PCR
cleanup kit. The quality, quantity (~100 ng/μl) and
formulation of the PCR product was checked using
a Qubit Fluorometer 3.0. Nanopore sequencing was
subsequently used on a 1 µg DNA template, and the
results were analysed using the EPI2ME 16S analysis
workflow. Finally, 16S rRNA sequencing results were
compared with the results obtained from the culture
method and the fluorescence imaging device.
Statistical analysis
The sample size was calculated to be 164 at a 99%
confidence level, with a 10% margin of error. Cor-
relation between the gram type obtained from fluo-
rescence imaging device and the tissue biopsy culture
was assessed using Cohen’s Kappa method in IBM’s
SPSS software. The results from the device were com-
pared with the culture results for gram type identifica-
tion, and the multi-class performance metrics, such
as accuracy, sensitivity and specificity were calculated
using standard formulas (28).
RESULTS
Clinical characteristics of the subjects
A total of 157 diabetic foot ulcer patients (104 males,
53 females) with a median age of 60 years were en-
rolled in the study, and 177 tissue samples were ob-
tained from these patients. Nine wounds were present
in the distal phalanges, 34 in the dorsal regions, 8
in the lateral heel region, 28 in the medial aspect of
the heel, 67 in the plantar region, 10 in the posterior
regions and 1 in the anterior tibia. In terms of the
Wagner classification scoring system, 5 patients had a
Grade 1 ulcer, 33 had a Grade 2 ulcer, 60 had Grade
3 ulcer and 59 had a Grade 4 ulcer (Table 1).
Organisms isolated in culture
Fifteen pathogens were isolated from the 177 tissue
biopsies collected (Table 2): 112 gram-negative, 43
gram-positive and 1 fungus; 42 reports came back
showing ‘no growth of bacteria’. Pseudomonas species
has the highest occurrence (40 samples), followed by
Enterococcus (22) and Klebsiella (20) species.
Inferences
The device picked up 111 wounds with gram-nega-
tive bacteria, 25 with gram-positive bacteria and 15
wounds with both gram-positive and -negative bac-
teria. A further 26 wounds had no bacterial burden.
Statistical data
Cohen’s kappa was used to determine the probability
of an agreement between gram type results obtained
from both the semi-quantitative culture test and the
multispectral imaging device in 177 tissue samples.
A good agreement was found between both meth-
ods, ĸ = .774, 95% CI, [-.0359 to .121], p< .001.
The results of the machine learning algorithm are

188 J OU R NAL O F WO UN D M A N AG EM EN T
OFFIC IAL JOURNAL OF THE EU RO PEA N WO U N D M A N AGEM EN T A SSO C IAT IO N
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N

Table 4: Machine learning performance of the Illuminate device

GN GP NG GP/GN
PPV 78.38 76.00 92.31 86.67
NPV 92.42 97.37 88.08 95.68
Accuracy 83.62 94.35 88.70 94.92
TPR (Sensitivity) 94.57 82.61 57.14 65.00
TNR (Specificity) 71.76 96.10 98.52 98.73

GN= Gram Negative; GP= Gram Positive; TP= True Positive; FP= False Positive; TN= True Negative;
FN= False Negative; TPR= True Positive Rate; TNR=True Negative Rate; PPV= Positive Predictive Value;
NPV= Negative Predictive Value

summarised in the confusion matrix provided in
Table 3. The class-averaged accuracy of the device
was found to be 90.4%, with a positive predictive
value of 78.38% for detecting gram-negative bacteria,
76.00% for gram-positive bacteria, 86.67% for pol-
ymicrobial (GP&GN) bacterial burden and 92.31%
for no significant bacteria burden (no growth). The
negative predictive value was found to be 92.42%
for gram-negative bacteria, 97.37% for gram-positive
bacteria, 95.68% for polymicrobial bacterial burden
and 88.08% for no growth. The sensitivity for gram-
negative bacteria was 94.57%, 82.61% for gram-
positive, 65% for polymicrobial and 57.14% for no
growth. Specificity for gram-negative was 71.76%,
96.10% for gram-positive, 98.73% for polymicrobial
and 98.52% for no growth. It should be emphasised
that, since the accuracy of the gold standard culture
method is not 100%, the accuracy of the fluorescence
imaging device could be potentially much higher.
The device’s imaging results and comparisons with
the culture report for various wounds with gram-pos-
itive bacteria, gram-negative bacteria and no growth
are shown in Figure 2.
16S rRNA sequencing results
To better assess the performance of the fluorescence
imaging device in the detection of bacterial burden,
16S rRNA sequencing was conducted to provide
more information on the polymicrobial species pre-
sent in the wound, as compared to a standard cul-
ture test. The results of 16S rRNA sequencing were
compared with the results obtained by the fluores-
cence imaging device and the culture method for the
presence/absence of significant bacterial bioburden.
Of the 26 biopsies compared with the 16S rRNA
sequencing and culture method, the culture method
gave positive results in 17 cases. In these 17 cases,
the species identified by the culture method corre-
lated with at least one of the species identified by
the 16S rRNA sequencing results in 16 cases. The
device also showed significant levels of bacterial bur-
den, which correlated with both the culture and 16S
rRNA results. However, the culture method showed
no growth in several cases (9 cases), where the 16S
rRNA results showed significant reads (>50 reads for
at least one organism) in five samples. Interestingly,
in four of the five cases, the fluorescence imaging
device was still able to show the presence of bacterial
burden, demonstrating the superior sensitivity of the
device, as compared to the culture.
DISCUSSION
Understanding bacterial burden is important for
the effective management of diabetic foot ulcers
and preventing downstream complications. Earlier
studies have shown that the clinical evaluation of
signs and symptoms (CSS) has a low accuracy rate for
identifying wounds with moderate to heavy bacterial
burdens (23). CSS is typically based on the clinical
signs of critical colonisation (NERDS) or infection
(STONEES) (29). The accuracy results vary widely
among the studies, ranging from <30% (23) to <80%
(29), hence, a better technique is needed for accu-
rate wound bioburden assessment. It is well known
that pathogens emit autofluorescence when excited
with UV and violet/blue regions of light during the
phase of growth and infection (21-22). This weak
autofluorescence is typically studied using standard
spectrofluorometers, which are both expensive and
bulky, making them difficult to adopt in a point-of- 

J OU R NAL O F WO UN D M A N AG EM EN T
OFFI C IAL JOURNAL OF THE E U RO PEA N WO U N D M A N AG EM EN T A SSO C IAT IO N
189
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N
care setting (30). A novel portable and handheld mul-
tispectral imaging device was developed for rapid and
non-invasive bacterial burden assessment at the bed-
side in under 2 minutes (31-31). The device detects
bacteria that have infected the tissue and are express-
ing significant metabolic activity, as the device mainly
targets NAD(P)H and flavins. The device integrates
multiple wavelengths of light both in the UV-A and
violet regions and captures weak autofluorescence at
different spectral emission bands using high optical
density to minimise background autofluorescence.
These spectral images are fed to an image processing
and machine learning algorithm that compares these
images and identifies significant bacterial burden re-
gions based on the differences in autofluorescence
intensities at various spectral bands before classifying
them based on the gram type.
This study was carried out to detect the accuracy
of the device in comparison with the current gold
standard culture method and proved that the device
is capable of accurately detecting and distinguishing
the gram type of bacteria within 2 minutes. Thus,
fluorescence imaging device can be used as a first-
line screening solution, especially when access to
microbiology labs is problematic (33). In addition,
the results show that the device has a better sensitiv-
ity and specificity in identifying bacterial burden,
compared to the standard CSS technique. The device
also enables evidence-based continuous monitoring
of patients’ wounds during hospitalisation, rather
than visual inspection, making wound tracking easier.
At present, it is difficult to accurately assess if infected
tissues have been removed completely post debride-
ment using CSS alone. Fluorescence imaging can aid
in effective debridement (25), as demonstrated in
earlier studies, and the device used in the present
study was also able to improve debridement efficiency
by providing immediate feedback regarding the status
of bacterial burden after debriding each layer. This
also saves healthy viable tissue from excessive debride-
ment, improving wound healing. Further, the device
can be used to check whether a wound area is free of
bacterial burden before tissue grafting.
There are several instances when a wound may ap-
pear infected clinically, but the culture report shows
no growth. This could be the result of concurrent
antibiotic therapy, the ability of culturing to pick only
out certain types of bacteria or a defective wound
swabbing/biopsy site (34-37). Devices such as the
novel multispectral imaging device can guide accu-
190 J OU R NAL O F WO UN D M A N AG EM EN T
OFFIC IAL JOURNAL OF THE EU RO PEA N WO U N D M A N AGEM EN T A SSO C IAT IO N
rate swabbing by identifying areas with the highest
levels of bacterial burden, leading to a better under-
standing of the overall microbial burden, especially
for polymicrobial wounds, for further therapeutic
interventions. In addition, the device helps doctors
to identify bacterial burdens away from the wound,
which can potentially be missed by visual inspection
alone.
Despite the above, the manual interpretation of mul-
ti-spectral autofluorescence images is difficult, subjec-
tive and requires significant training. This cumber-
some process might pose severe challenges in terms
of ensuring the accuracy of decisions. This is where
machine learning plays a vital role, as it corrects for
the background noise caused by tissue autofluores-
cence and automatically classifies infected regions as
gram-positive or gram-negative. The colour-coding
aids in effective spatial understanding of pathogen
presence in the wound. Typically, it is assumed that
the bacterial burden is present at the centre of the
wound, at the site of maximum slough, but the device
was frequently able to detect the presence of patho-
gens, including pseudomonas, away from wound, at
the peri wound site, providing hints about the reasons
behind delayed wound healing.
Limitations
The device has the following limitations. Bone, ten-
don and fat regions of wounds have autofluorescence
of their own, primarily contributed by collagen. The
algorithm may misrepresent these regions as infected,
due to high levels of autofluorescence. Hence, the
results from the device must be clinically correlated
in such scenarios. The presence of betadine and other
cleaning solutions that are fluorescent in nature will
also interfere with these weak autofluorescence sig-
nals; hence, the wound must be thoroughly cleaned
with saline solution before imaging. In future,
through advanced image processing and machine
learning techniques, it may be possible to compen-
sate for the interference from betadine fluorescence
and extract the bacterial fluorescence signatures to as-
sess the bioburden, especially at high bacterial loads.
Surgical sutures and cotton dressing remaining on
wounds might also predict a false bacterial burden,
hence one needs to be cautious while imaging.
CONCLUSION
The device is a first-of-its-kind device that uses mul-
tispectral autofluorescence imaging combined with
machine learning for the rapid detection of bacte-
rial burden and gram type classification on diabetic
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N
foot wounds. The device demonstrated >90% class-
averaged accuracy, compared to the current gold
standard culture method in gram type classification.
In addition, the device can pick up bacterial bur-
dens, even in cases the culture report has missed, as
confirmed by 16S rRNA sequencing. Thus, the de-
vice can guide doctors towards guided debridement,
the right first-line treatment protocol and improved
wound tracking.
IMPLICATIONS FOR
CLINICAL PRACTICE
n Doctors and clinicians can obtain an instant
understanding of the bioburden on the wound
and gram type of infecting bacteria.
n The device aids in effective wound debride-
ment.
n Doctors can optimise dosages and prescriptions
based on the bacterial type and bioburden
present on the wound.
Further research
Areas for further research could include:
n Large-scale studies correlating the performance
of the device with the results from metagenome
studies.
n Quantification of the bioburden using a
fluorescence imaging device.
n Longitudinal tracking of bioburden and
correlations with wound healing in ulcer
patients.
Acknowledgments
The authors thank Vignesh Devaraj, Clinical Re-
search Intern, who was pivotal in data collection and
documentation.
J OU R NAL O F WO UN D M A N AG EM EN T
OFFI C IAL JOURNAL OF THE E U RO PEA N WO U N D M A N AG EM EN T A SSO C IAT IO N
Funding
No exclusive funding was provided by Adiuvo Di-
agnostics or any other entity for the clinical studies.
Ethical statement
The authors are accountable for all aspects of the
work, including ensuring that questions related to
the accuracy or integrity of any part of the work are
appropriately investigated and resolved. The trial
was conducted in accordance with the Declaration
of Helsinki (as revised in 2013). The study was con-
ducted at Hycare Hospital, in Chennai, India, after
obtaining institutional ethics committee approval
(IEC 031#HYC/IEC/2018). All data were collected
with patient consent prior to their participation in the
study. The study was registered with the Clinical Trial
Registry of India (Reg. No. CTRI/2018/10/016147).
Key messages
n A novel multispectral imaging device is a first
of its kind device which uses multispectral
autofluorescence imaging combined with
machine learning for rapid detection of
bacterial burden and gram type classification
on the diabetic foot ulcers.
n The device demonstrated >90% class-averaged
accuracy as compared to the current gold
standard culture method in gram type
classification and can pick-up bacterial burden
even in cases where the culture method has
missed as confirmed by 16S rRNA sequencing.
n It can assist doctors towards right first line
treatment protocol, guided debridement, and
effective wound tracking. m
191
 S C I E N C E , P R A C T I C E A N D E D U C AT I O N
REFERENCES
1. IDF diabetes ATLAS 9th edition [Internet]. 2019.
[2102 April 6]. Available at https://www.diabetesatlas.
org/upload/resources/material/20200302_133351_
IDFATLAS9e-final-web.pdf
2. Andersen CA, Roukis TS. The diabetic foot. Surgical
Clinics of North America. 2007 Oct;87(5):1149–77.
3. Singh N, Armstrong DG, Lipsky BA. Preventing foot
ulcers in patients with diabetes. JAMA 2005;
292(2):217-28.
4. Boulton AJM, Armstrong DG, Hardman MJ, Malone
M, Embil JM, Attinger CE, Lipsky BA, Aragón-
Sánchez J, Li HK, Schultz G, Kirsner RS. Diagnosis
and Management of Diabetic Foot Infections.
Arlington (VA): American Diabetes Association; 2020
Jan. PMID: 32105420.
5. Zubair M, Ahmad J. Potential risk factors and
outcomes of infection with multidrug resistance
among diabetic patients having ulcers: 7 years study.
Diabetes Metab Syndr. 2019 Jan-Feb;13(1):414-418.
doi: 10.1016/j.dsx.2018.10.014. Epub 2018 Oct 13.
PMID: 30641735.
6. Wu M, Pan H, Leng W, Lei X, Chen L, Liang Z.
Distribution of Microbes and Drug Susceptibility in
Patients with Diabetic Foot Infections in Southwest
China. J Diabetes Res. 2018 Aug 5;2018:9817308.
doi: 10.1155/2018/9817308. PMID: 30175153;
PMCID: PMC6098928.
7. Sadeghpour Heravi F, Zakrzewski M, Vickery K, G
Armstrong D, Hu H. Bacterial Diversity of Diabetic
Foot Ulcers: Current Status and Future Prospectives. J
Clin Med. 2019 Nov 10;8(11):1935. doi: 10.3390/
jcm8111935. PMID: 31717640; PMCID:
PMC6912738.
8. Sankar N, Khaja Moinuddin S, Mohan S, Microbial
etiology of diabetic foot ulcers: Swab versus tissue
culture. Int J Surg Sci 2020; 4(2):146-8.
9. Huang Y, Cao Y, Zou M, Luo X, Jiang Y, Xue Y, et al.
A comparison of tissue versus swab culturing of
infected diabetic foot wounds. International Journal of
Endocrinology. 2016;2016:1–6.
10. Ogba OM, Nsan E, Eyam ES. Aerobic bacteria
associated with diabetic foot ulcers and their
susceptibility pattern. biomed dermatol. 2019
Dec;3(1):1.
11. Aragón-Sánchez J, Luis Lázaro-Martínez J, Pulido-
Duque J, Maynar M. From the diabetic foot ulcer and
beyond: how do foot infections spread in patients with
diabetes? Diabetic Foot & Ankle. 2012
Jan;3(1):18693.
12. Tascini C, Lipsky BA, Iacopi E, Ripoli A, Sbrana F,
Coppelli A, et al. KPC-producing Klebsiella
pneumoniae rectal colonization is a risk factor for
mortality in patients with diabetic foot infections.
Clinical Microbiology and Infection. 2015
Aug;21(8):790.e1-790.e3.
13. Hassan MA, Tamer TM, Rageh AA, Abou-Zeid AM,
Abd El-Zaher EHF, Kenawy ER. Insight into
192
OFFIC IAL JOURNAL OF THE EU RO PEA N WO U N D M A N AGEM EN T A SSO C IAT IO N
multidrug-resistant microorganisms from microbial
infected diabetic foot ulcers. Diabetes & Metabolic
Syndrome: Clinical Research & Reviews. 2019
Mar;13(2):1261–70.
14. Miyan Z, Fawwad A, Sabir R, Basit A. Microbiological
pattern of diabetic foot infections at a tertiary care
center in a developing country. J Pak Med Assoc. 2017
May;67(5):665-669. PMID: 28507348.
15. Edwards U, Rogall T, Blöcker H, Emde M, Böttger
EC. Isolation and direct complete nucleotide
determination of entire genes. Characterization of a
gene coding for 16S ribosomal RNA. Nucl Acids Res.
1989;17(19):7843–53.
16. Gunasekaran P. Assessment of major microorganisms
involved in diabetic foot infection which delays
wound healing. IAIM 2017; 4(8):87-90.
17. Ramani A, Ramani R, Shivananda PG, Kundaie GN.
Bacteriology of diabetic foot ulcers. Indian J Pathol
Microbiol 1991; 35(2):81-7.
18. Dartnell LR, Roberts TA, Moore G, Ward JM, Muller
JP. Fluorescence characterization of clinically-impor-
tant bacteria. Brody JP, editor. PLoS ONE. 2013 Sep
30;8(9):e75270.
19. Leriche F, Bordessoules A, Fayolle K, Karoui R, Laval
K, Leblanc L, et al. Alteration of raw-milk cheese by
Pseudomonas spp.: Monitoring the sources of
contamination using fluorescence spectroscopy and
metabolic profiling. J Microbiol Methods 2004 Oct;
59(1):33-41.
20. Estes C, Duncan A, Wade B, Lloyd C, Ellis W Jr,
Powers L. Reagentless detection of microorganisms by
intrinsic fluorescence. Biosens Bioelectron 2003;
18(5-6):511-9.
21. Radhakrishnan G, Gupta A, King J, Ganvir D. Rapid
handheld screening device to detect skin and soft
tissue infections. PROC SPIE 2020; 11211.
22. Rennie M, Dunham D, Lindvere-Teene L, Raizman
R, Hill R, Linden R. Understanding real-time
fluorescence signals from bacteria and wound tissues
observed with the moleculight i:xTM. Diagnostics.
2019 Feb 26;9(1):22.
23. Serena, TE, Harrell K, Serena L, Yaakov RA. Real-time
bacterial fluorescence imaging accurately identifies
wounds with moderate-to-heavy bacterial burden. J
Wound Care 2019; 28(6):346-57.
24. Sandy‐Hodgetts K, Andersen CA, Al‐Jalodi O,
Serena L, Teimouri C, Serena TE. Uncovering the
high prevalence of bacterial burden in surgical site
wounds with point‐of‐care fluorescence imaging.
International Wound Journal. 2022 Oct;19(6):1438–
48.
25. Cole W, Coe S. Use of a bacterial fluorescence imaging
system to target wound debridement and accelerate
healing: A pilot study. J Wound Care 2020;
29(Sup7):S44-S52.
26. Price N. Routine fluorescence imaging to detect
wound bacteria reduces antibiotic use and antimicro-
bial dressing expenditure while improving healing
J OU R NAL O F WO UN D M A N AG EM EN T
rates: retrospective analysis of 229 foot ulcers.
Diagnostics. 2020 Nov 10;10(11):927.
27. Oropallo AR, Andersen C, Abdo R, Hurlow J, Kelso
M, Melin M, et al. Guidelines for point-of-care
fluorescence imaging for detection of wound bacterial
burden based on delphi consensus. Diagnostics. 2021
Jul 6;11(7):1219.
28. Tharwat A. Classification assessment methods. ACI.
2021 Jan 4;17(1):168–92.
29. Woo KY, Sibbald RG. A cross-sectional validation
study of using NERDS and STONEES to assess
bacterial burden. Ostomy Wound Manage. 2009 Aug
1;55(8):40-8. PMID: 19717855.
30. Kumar AA, Hennek JW, Smith BS, Kumar S, Beattie
P, Jain S, et al. From the bench to the field in low-cost
diagnostics: two case studies. Angew Chem Int Ed.
2015 May 11;54(20):5836–53.
31. Radhakrishnan G, Gupta A, King J, Ganvir D. Rapid
handheld screening device to detect skin and soft
tissue infections. PROC SPIE 2020; 11211.
32. Shah BM, Ganvir D, Sharma YK, Mirza SB, Misra
RN, Kothari P, et al. Utility of a real-time fluorescence
imaging device in guiding antibiotic treatment in
superficial skin infections. IJDVL. 2021 Feb
24;88:509–14.
33. Lipsky BA, Senneville É, Abbas ZG, Aragón‐Sánchez
J, Diggle M, Embil JM, et al. Guidelines on the
diagnosis and treatment of foot infection in persons
with diabetes (IWGDF 2019 update). Diabetes Metab
Res Rev [Internet]. 2020 Mar [cited 2022 Nov
3];36(S1). Available from: https://onlinelibrary.wiley.
com/doi/10.1002/dmrr.3280
34. Bill TJ, Ratliff CR, Donovan AM, Knox LK, Morgan
RF, Rodeheaver GT. Quantitative swab culture versus
tissue biopsy: a comparison in chronic wounds.
Ostomy Wound Manage. 2001;47(1):34-37.
35. Levine NS, Lindberg RB, Mason AD Jr, Pruitt BA Jr.
The quantitative swab culture and smear: A quick,
simple method for determining the number of viable
aerobic bacteria on open wounds. J Trauma.
1976;16(2):89-94.
36. Demetriou M, Papanas N, Panopoulou M, Papatheo-
dorou K, Bounovas A, Maltezos E. Tissue and swab
culture in diabetic foot infections: neuropathic versus
neuroischemic ulcers. The International Journal of
Lower Extremity Wounds. 2013 Jun;12(2):87–93.
37. Bonham PA. Swab cultures for diagnosing wound
infections: a literature review and clinical guideline.
Journal of Wound, Ostomy & Continence Nursing.
2009 Jul;36(4):389–95.