Infrared Physics & Technology 52 (2009) 97–108

Contents lists available at ScienceDirect

Infrared Physics & Technology

journal homepage: www.elsevier.com/locate/infrared

Infrared thermography on ocular surface temperature: A review

Jen-Hong Tan a, E.Y.K. Ng a,*, U. Rajendra Acharya b, C. Chee c
a
School of Mechanical and Aerospace Engineering, College of Engineering, Nanyang Technological University, 50, Nanyang Avenue, 639798, Singapore
b
Department of Electronics and Computer Engineering, Ngee Ann Polytechnic, Singapore
c
Medical and Surgical Retinal Centre Department of Ophthalmology, National University Hospital, 119074, Singapore

a r t i c l e i n f o a b s t r a c t

Article history: Body temperature is a good indicator of human health. Thermal imaging system (thermography) is a non-
Received 2 December 2008 invasive imaging procedure used to record the thermal patterns using Infrared (IR) camera. It provides
Available online 18 May 2009 visual and qualitative documentation of temperature changes in the vascular tissues, and is beginning
to play an important role in the field of ophthalmology. This paper deals with the working principle,
Keywords: use and advantages of IR thermography in the field of ophthalmology. Different algorithms to acquire
Eye the ocular surface temperature (OST), that can be used for the diagnosis of ocular diseases are discussed.
Infrared
! 2009 Elsevier B.V. All rights reserved.
Thermogram
Temperature
Cornea
Ocular

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
2. The principles of infrared thermography on human eye . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
2.1. The physics of IR radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
2.2. The IR thermography system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
2.3. The measured OST. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3. Methodologies in the study of OST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.1. Manual measures in the acquisition of OST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.2. Semi-automated method in the acquisition of OST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.3. Automated method in the acquisition of OST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
4. Application of IR thermography to ocular studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.1. Studies of ocular physiologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.2. Studies of ocular diseases and surgery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.3. Studies of non-ocular diseases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

1. Introduction techniques. Galileo invented the thermoscope in 17th century and

Temperature has been used to investigate the physiological and
pathological changes in human body since 400 B.C. using different
* Corresponding author. Address: Adjunct NUH Scientist, Office of Biomedical
Research, National University Hospital of Singapore. Tel.: +65 67904455; fax: +65
67911859.
E-mail address: mykng@ntu.edu.sg (E.Y.K. Ng).
then it evolved into modern mercury-in-glass thermometer, radi-
ometers for middle-ear temperature and disposable sterile ther-
mocouple which is widely used these days. Liquid crystals such
as cholesteric esters, found in 1877 had the property of changing
color with temperatures, were used to display the distribution of
temperature on skin topographically. Such detectors were inex-
pensive, had relatively short life span and may alter surface tem-
perature due to large area of contact. Schlerian photography

1350-4495/$ - see front matter ! 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.infrared.2009.05.002
98 J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108
enabled researchers to see the convection currents surrounding the
body, but is limited to the study of heat transfer around insulated
clothing [1]. Except Schlerian photography, most of the methods
developed in past had limitations of either requiring contact with
subject or incapable of displaying distribution of temperature.
Infrared (IR) thermography is a non-contact and non-intrusive
temperature measuring technique, with an advantage of no alter-
ation in the surface temperature and capable of displaying real-
time surface temperature distribution. It was first introduced by
Lawson in 1956 to modern medicine and discovered the associa-
tion of elevated skin temperature with breast carcinoma [2] and la-
ter investigated the feasibility and potential of using IR
thermography as a tool to study breast lesions [3]. This technology
has revolutionized the field of temperature measurement in the
last 50 years and is widely employed nowadays.
In the field of thermo-fluid dynamics IR thermography was ap-
plied to measure convective heat fluxes, and for the comprehen-
sion of fluid dynamics phenomena on the flow field behaviour
over complicated body shapes [4]. The process of ice nucleation
and ice propagation in flowers of fruit trees and other frost sensi-
tive plants were studied using IR thermography in agriculture
[5]. It was used for the measurement of size, depth and thermal
resistance of materials and components [4], and also for environ-
mental monitoring such as sea, river pollution, information about
indoor climate [6], inspection of plants and assistance in the reduc-
tion of maintenance cost of mechanical equipment [4]. In poly-
graph testing, a standard security procedure favoured by US
government, IR thermography was applied to perform facial image
analysis [7].
In medical field, IR thermography has been used to assist in
decision making in open heart surgery due to its ability to provide
real-time information [8]. It was also used for the management of
neuropathic pain [9,10] and the assessment of patient response to
chiropractic care by measuring the temperature gradient in clinical
setting [11]. So far, IR system has been used to diagnose breast can-
cer [12,13], rheumatism [1], skin lesion [14], fever [15], impotence
[16] and thyroid gland disease [17]. Currently, it has been applied
to ophthalmology to diagnose eye diseases [18].
The ocular anterior anatomy and physiology nowadays can be
studied using a number of ophthalmic imaging techniques: slit
lamp biomicroscopy, confocal microscopy, corneal topography,
optical coherence tomography, computerized tomography, ultra-
sonic biomicroscopy, and magnetic resonance imaging. These
Fig. 1. Typical thermogram of normal eye.
techniques are capable of providing accurate description of ana-
tomical features and help to diagnose the ocular diseases better.
Infrared thermography is renowned for its ability to detect the
pathological and physiological changes in the eye which are ob-
scured or unreachable under anatomical examination. A typical
ocular thermogram of normal eye is shown in Fig. 1. It has been
used to study the inflammation of human lacrimal drainage system
[19], dry eye [20], carotid artery stenosis [21], glaucoma [22], uni-
lateral exophthalmos [23], Tolosa–Hunt syndrome [24], and oph-
thalmic post-herpetic neuralgia [25]. It was also used to diagnose
retinoblastoma in children [26] and vascular neuritis [27] of the
optic nerve. The OST can be used in the diagnosis of different ocular
diseases.
The invasive methods of measuring eye temperature require di-
rect contact with human cornea. Among the invasive measuring
techniques, needle probe was mainly employed. During measure-
ment, needle probe acts as a cooling fin when inserted into the
eye [28] and error inevitably exists if the penetration depth is be-
low 40 mm. Topical anesthesia is often required, and this instilled
solution often lowers OST. In addition, the penetration of needle
can be traumatic, which often induces further blood flow in eye
and thus alters OST. Hence, this invasive method, is not comfort-
able to the subjects and the reported discrepancies in temperature
in some cases can be up to 6 "C [28].
Infrared thermometry and thermography can measure OST
without causing trauma on subjects. However, they are unable to
measure the intraocular temperature. These techniques remotely
measure the emitted IR radiation and acquire temperature data
of a specific surface. During measurement there will be no alter-
ation in the surface temperature and also the data collected is of
higher precision. With this technology researchers are able to
study OST with greater ease, and accuracy.
2. The principles of infrared thermography on human eye
In general, IR thermography refers to the recording of tempera-
ture, or the distribution of temperature utilizing infrared radiation
emitted from a body surface, forming an image called thermogram.
The 2-D thermogram presents the distribution of temperature dis-
tinctly unlike IR thermometry, which gives a single temperature
value. IR thermography also differs with IR photography, where
in the latter records infrared radiation reflected back from objects
in the presence of some external infrared energy sources. The
 J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108 99

principle of IR thermography on eye involves the physics of IR radi-

ation and measurement of OST. These concepts are discussed in the
following sections.

2.1. The physics of IR radiation

Any object, when its temperature is above absolute zero, emits

electromagnetic radiation, called thermal radiation. In physics, the
amount of electromagnetic waves radiated by an object is quanti-
fied by the term spectral radiance. It describes the amount of elec-
tromagnetic waves emitted from a particular surface on the basis
of per unit solid angle. The idea of black body radiation forms
the basis of the physics of thermal radiation, though behaviours
of many objects deviate from the ideal black body extensively.
When a black body is in equilibrium at temperature T in an
environment, it absorbs and emits a specific continuous spectrum
of wavelengths and intensities. The profile of the emitted spectrum
by a black body depends only on the temperature at equilibrium in
an environment and therefore the temperature of a black body is
directly related to the wavelengths of the light that it emits.
According to Planck’s law, the spectral radiance, I (in watts per ste-
radian per square meter) of electromagnetic radiation at all wave-
lengths, k, at temperature T from a black body as a function of
wavelength is given by
2
2phc 1 projected and recorded on IR detector within the camera, is deter-
Iðk; TÞ ¼ W cm$2 lm$1
k5
e
hc
kkT $1
where k is the Boltzmann constant (1.381 % 10$23 J/K), h is the
Planck constant (6.626 % 10$34 Js), c is the speed of light
(2.998 % 108 m/s) and T is the absolute temperature. Differentiating
Planck’s law with respect to k to look for the maximum radiation
intensity, the Wien’s displacement law is obtained
2898 lK
kmax ¼
T
which mathematically illustrates a common observation: as the tem-
perature of an object rises, the color of light emitted varies from infra-
red to red to orange. Given human body temperature of 37 "C, the
emitted light peaks at 9.35 lm which falls into the infrared region.
Consider a small flat body radiating outward into a half-sphere
under an temperature-equilibrium environment, one can deduce
Stefan–Boltzmann’s law from Eq. (1) and get
j ¼ rT 4
which states that for a black body the total energy radiated per unit
surface area in a unit time is directly proportional to the fourth
power of the absolute temperature of the black body.
In reality, real objects are rarely a black body, though in certain
spectral intervals they may behave closed to the ideal black body.
In most cases real object emits only part of the radiation a black
body emits at the same temperature and same wavelength. Denote
the amount of radiation emitted by real object as Ek and Ebk for the
amount of radiation emitted by black body, the emissivity is de-
fined as
Ek
e¼
Ebk
and Eq. (3) can be rewritten for gray body with the knowledge of
emissivity as
j ¼ erT 4
2.2. The IR thermography system
Planck’s law, Wien’s displacement law, Stefan–Boltzmann’s law
and the concept of emissivity is the physics behind the working
Fig. 2. IR camera (VarioTHERM# head II, http://www.infratec.de/fileadmin/down-
loads/pdf/Flyer_Variotherm_head_en_mail.pdf).
principle of an IR thermography system. An IR thermography sys-
tem in general consists of a camera (as shown in Fig. 2), with some
detachable optics, and a personal computer that controls the cam-
era. In the camera there is an IR detector, which absorbs the IR
shining by the lenses and converts the IR signal into electrical volt-
age or current for processing. The way a desired field of view is
ð1Þ
mined by the imaging system.
Generally, there are two main sorts of imaging system available
to the IR thermography: scanner (scanning array) and focal plane
array. The focal plane array is superior to scanning array in almost
every aspect in terms of performance. Scanning array consists of
linear arrays, rastered across the desired field of view using one
detector, two lenses, one horizontal and one vertical deflection
mirrors, to construct a 2-D image as illustrated in Fig. 3a. Its work-
ð2Þ ing principle is analogous to a person looking at a view through
narrow slit, rastering both his/her head and slit in the direction
perpendicular to the slit to build up 2-D image. Focal plane array
images a desired field of view without scanning, and it works sim-
ilar to a typical camera, in which the film captures the 2-D image
directly projected by the lens at image plane (Fig. 3b).
IR detectors of IR thermography consist of an array of elements
reactive to IR. In the broader field of IR thermology, there are two
types of IR detectors: thermal and photonic. Thermal detectors
ð3Þ such as microbolometer, pyroelectric detectors and Golay cells de-
tect heat generated by incidence of IR radiation and are not wave-
length dependent. They are commonly used in bolometers such as
thermocouple and thermopile. They show slow response time and
low detection capability without the need for cooling. The photonic
detectors like quantum well infrared photodetector (QWIP) and
mercury–cadmium–telluride, are wavelength dependent, and can
detect the temperature of a remote object’s surface by measuring
the amount of IR incidence of a specific range. Unlike thermal
detectors, they offer quicker response time and better detection
performance, and therefore are favourite as IR detector for focal
plane array imaging system.
ð4Þ Most of the photonic detectors operate only at cryogenic tem-
perature, otherwise the detected signal would be swamped by
thermal noise. The IR thermography utilizing imaging system of fo-
cal plane array can be of cooled or uncooled type. Most of these
detectors are cooled evaporatively, either by liquid helium, liquid
ð5Þ
nitrogen, thermal-electric cooler, or Stirling cycle refrigerator.
Though not common, in the latest technologies microbolometer
is used in focal plane array, which have lower cost and require
no cooling.
Presently in the medical field IR thermography captures ther-
mogram at wavelength range of 8–12 lm, which falls in the region
100 J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108
Fig. 3. (a) Scanning array: 1. detector 2. lens 3. horizontal deflection mirror 4. vertical deflection mirror 5. lens 6. object 7. measuring spot. (b) Focal plane array: 1. object with
measuring spots 2. lens 3. detectors (http://www.infratec.de).
of long-wave IR (LWIR). They are further subdivided into a few
bands [29,30]: near IR (NIR), from 0.7 to 1 lm, short-wave IR
(SWIR), from 1 to 3 lm, mid-wave IR (MWIR), from 3 to 5 lm,
long-wave IR (LWIR), from 7 to 14 lm, very long-wave IR (VLWIR),
from 12 to 30 lm. NIR and SWIR are used in fiber optics telecom-
munications and long-distance communications, respectively;
guided missiles technology making use of IR heat is used in MWIR.
The LWIR is the region of ‘‘thermal imaging” which does not re-
quire external thermal source to obtain a passive isotherm of the
outside world. The IR thermography used for measurement of
OST obtains thermogram in LWIR.
2.3. The measured OST
Previous studies [31,32] have illustrated that the absorption
bands of ocular tissues are similar to water, which was opaque to
far infrared. Water is found to have an emissivity of 1, and it be-
haves like black body radiator on IR spectral above 3 lm. The emit-
ted spectrum falls in between 1 lm and 30 lm with a maximum
distance at 9 lm for ocular tissues, under normal circumstances
(32 "C). Consequently, the spectrum radiated by any ocular tissues
will be fully absorbed by ocular tissues anterior to it. In other
words, the spectrum emitted by vitreous will be absorbed by lens
that is anterior to it.
The transmission in between cornea and tear film is slightly dif-
ferent. Though water behaves similar to black body on IR spectrum
above 3 lm, for thin water layer the degree of absorption depends
on the thickness of layer. The transmittance of IR on spectrum
8 lm to 13 lm is about 30% when the layer thickness is 10 lm.
When the layer thickness was 40 lm or above, the transmittance
approximates to zero [33,34]; and tear film thickness was found
to be around 40 lm [35]. Consequently, the spectrum radiated by
cornea was also wholly absorbed by tear film. The temperature
of tear film was measured by thermography during the evaluation
of OST [34,36,37]. This study was performed using finite element
model of heat transfer in human eye [38].
A recent study [39] verified the above idea and showed that,
OST was mainly related to tear film stability, rather than other
parameters such as central corneal thickness, corneal curvature
or depth of anterior chamber.
In clinical thermography, studies have indicated that an angle of
viewing of curved anatomical surface beyond 90" will lead to a
reduction in measured temperature of 4 "C or more, due to the var-
iation of emissivity at different angle of viewing. For the measure-
ment of OST by IR thermography, it is assumed that, the angle of
viewing on any part of the cornea and sclera during measurement
with respect to the thermography is within p=4, so that the error
induced due to the variation of emissivity at different angle of
viewing is negligible. Furthermore, as the angle of view becomes
greater, the amount of reflected thermal radiation from anatomical
surface increases and the error in temperature measured grows.
Such resultant error is found to be negligible for the case where an-
gle of view falls within p=4. Therefore, the OST measured can be
comfortably taken as temperature of tear film even after consider-
ations of error incurred by angle of view and reflected thermal
radiation.
3. Methodologies in the study of OST
IR thermography captures temperatures and its variation over
ocular surface. Generally, it is either displayed in gray-scale or
RGB color palette. RGB thermogram provides a better visual repre-
sentation of temperatures for direct inspection, which relies on
users experience and judgment during the study of OST. Gray-scale
IR thermogram is useful in the quantitative analysis of the OST.
OST is the temperature of a specific location acquired on corneal
surface or some defined region [19,23,40–43]. Usually, it is studied
by comparisons on several regions in normal and diseased eyes.
3.1. Manual measures in the acquisition of OST
Efron et al. have estimated the geometric centre of cornea, and
measured surface temperature at every 0.5 mm increments on
either side of corneal surface horizontally [44]. In total there are
11 points running across the anterior surface, as illustrated in
Fig. 4. The method enables the study of temperature profile of
OST, using a regression polynomial
DT ¼ ax2 þ bx þ c ð6Þ
2
where a = 0.01 "C/mm , b = 0.003 "C/mm, c = 0.01 "C, and x is the
distance from the geometric centre of cornea in mm.
Morgan et al. have [21,36,45] approached the acquisition of OST
by different method. Five 10 % 10 boxes were placed in five differ-
ent anatomical locations along horizontal meridian running across
the estimated centre of cornea as shown in Fig. 5. One was placed
at the estimated centre of cornea, two at the limbal position, and
another two at the nasal and temporal conjunctiva (2 mm from
the limbus). The area of each box was approximately 1 mm2. Mean
OST was used for statistical analysis. In one of the studies [45],
 J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108
Fig. 4. Methodology by Efron et al. in the acquisition of OST [44].
Fig. 5. Methodology by Morgan et al. in the acquisition of OST [21,36,45].
radial temperature difference (RTD) was proposed to represent the
variation in temperature across the cornea. It is the temperature
difference, in between the average temperature value at the centre
of cornea, and the mean temperature value of the sites of two lim-
bal positions.
There were other similar measures to study OST. Instead of
boxes, points were placed at five different anatomical locations
along the horizontal meridian. Five points were placed on centre
of cornea, internal and external canthus, half-way from the inter-
nal canthus and nasal limbus, half-way from the temporal limbus
and external canthus (Fig. 6) [46,47]. In another scheme [48] five
points equally placed along a horizontal line running through cen-
tre of cornea, connecting medial and lateral canthi and those points
were placed (Fig. 7).
In some studies either a squared 10 % 10 pixels box [42,49], or a
squared 20 % 20 pixels box [50] with an actual area of 3.3 mm2, or
an area of 4 mm2 [20] or an encircled region of 4.4 mm diameter
[51], or a small circle [52] at the centre of cornea were used to
study OST.
A circular region was defined to estimate the centre of cornea
and the radius of the circle was either a fixed value [39] or was ac-
quired by subjective judgment [25,53]. Purslow et al. have re-
corded temperature data from 23 points across the anterior eye,
as illustrated in Fig. 8, and this data was grouped into five regions
Fig. 6. Methodology by Galassi et al. in the acquisition of OST [46,47].
101
Fig. 7. Methodology by Sodi et al. in the acquisition of OST [48].
Fig. 8. Methodology by Purslow et al. in the acquisition of OST [54].
for subsequent analysis: central, superior, inferior, nasal, and tem-
poral [54]. Tan et al. [55] have estimated the geometric centre of
cornea, and OST were measured by a total of 20 points across the
anterior eye: geometric centre of cornea, three points inferiorly
and four points superiorly at 2 mm separation, six points nasally
and temporally at 2 mm separation, respectively [55] (Fig. 9).
Chang et al. [56] have studied Graves’ ophthalmopathy by acquir-
ing local temperatures of lateral orbit (reference point), upper eye-
lid, caruncle, medial conjunctiva, lateral conjunctiva, lower eyelid,
and cornea, as shown in Fig. 10.
3.2. Semi-automated method in the acquisition of OST
A semi-auto method was developed to acquire OST on thermo-
gram using standard procedure [57], as shown in Fig. 11. Thermo-
gram of eye (in OEM data format) was converted to gray-scale jpeg
file. Then the image was manually cropped to consist only of eye,
and resized to a standard size of 400 % 200 pixels. An algorithm
to detect the circular cornea was proposed, with corneal radius
being one-fourth of the length of the entire eye. Temperature pro-
file in the middle of the cornea was plotted, as illustrated in Fig. 12.
Fig. 9. Methodology by Tan et al. in the acquisition of OST [55].
102 J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108
Fig. 10. Methodology by Chang et al. in the acquisition of OST [56].
The variation in temperature over the cornea was used to study
temperature deviation along cornea (TDC)
X user is required to select location for placement of initial contour,
TDC ¼ f ðx; yÞ $ f ðx; y þ 1Þ
y
3.3. Automated method in the acquisition of OST
An automated method to acquire OST was developed [58,59], as
illustrated in Fig. 13. In that method, localization of eye and cornea
was achieved by snake algorithm and target tracing function [58]
without any manual intervention. Genetic algorithm was utilized
to perform search for minimum on target tracing function.
Snake [60] is an active contour consisting of a series of points
(dubbed snake points), moving under forces of gradient vector flow
(GVF) [61] to lock onto nearby edges, and thereby delineates the
shape of eye if its initial contour is of appropriate shape and placed
in some suitable locations. GVF, an intense force field pushing
snake contour to reach its minimum in an energy functional (pro-
posed by Kass et al.), is derived from edge map. In the automated
method, the algorithm acquires edge map by applying Gaussian
blur on the ocular thermogram (gray-scale image).
For a snake locking into the desired feature, it requires the
snake to be initially placed closed to the feature of interest, and
the shape of the snake must be carefully selected. In fact, even in
some more sophisticated methods such as spline-based shape
matching [62], diffusion snakes [63], and active shape models
Fig. 11. Methodology by Acharya et al. in the acquisition of OST [57].
Fig. 12. Pixel profile across the cornea (corresponds to temperature profile of
cornea).
[64], these deformable template matching techniques are local
optimizers and require good initial position or clean image to avoid
incorrect localization [65]. In other words, for some applications
ð7Þ also the size or the shape of contour, in order to get the correct
localization. There were cases where user has to monitor the
expansion or convergence of snake.
The eye in ocular thermogram is not fixed at the centre of the
image, and may appear anywhere in the image. For some eyes
which have smaller size, they occupy just some part of the image
instead of most of the spaces in thermogram. Given these circum-
stances, local optimizers are not a viable option if any user initial-
ization is to be avoided.
In the automated method, the problem is overcome by using a
target tracing function. Target tracing function evaluates few con-
verged snakes, and from them only a few, or probably just one of
them correctly localize the eye. The converged snake which gives
target tracing function the lowest value, is the snake that accu-
rately localizes the eye. In other words, the algorithm performs a
search for minimum on target tracing function to get the right
snake to correctly localize the eye. Genetic algorithm is utilized
to perform this search process.
The radius and the centre of cornea are derived from the final
resultant snake points. The centre of cornea is the same as centroid
of the snake, and radius of cornea is acquired on the ground of the
snake points with the use of a formula derived [66]. Fig. 13 shows
result of the automatic detection of eye and cornea using genetic
snake algorithm.
 J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108
Fig. 13. Methodology by Tan et al. in the acquisition of OST [58,59].
4. Application of IR thermography to ocular studies
In ophthalmology, IR thermography has been applied to study
ocular physiologies, ocular diseases and surgery. Several correla-
tions between OST and a number of physiological and pathological
changes in eye have been studied with greater ease through IR
thermal imaging. The OST is not only affected by physiological
and pathological changes in eye, but also some of the external
factors.
Room temperature was shown to influence OST [39,67]; a rise
in 1 "C room temperature may lead to an increase of 0.15 "C to
0.2 "C in OST [68,69]. The OST was reported to decrease with an in-
crease in the air flow [70], therefore uniform room temperature
and humidity are required to minimize the discrepancies in studies
of OST.
It was reported that, OST increases when an eye is infected with
anterior uveitis [36,71], and indicates a negative correlation for car-
otid artery stenosis [19,67,72]. The degree of hyperemia in bulbar
conjunctiva (examined in terms of grade of redness using
McMonnies scale) were positively correlated to OST [73]. It is in
agreement among most of the investigators that, a large inter-ocular
temperature difference indicates the presence of eye disease.
4.1. Studies of ocular physiologies
Mapstone described thermographic patterns in normal, ische-
mic and hyperemic eyes using Bofor IR camera system [40].
Wachtmeister also conducted a similar study, investigating both
normal and diseased eyes [41]. He found out that, the affected
eye was warmer than the normal eye for anterior and posterior
diseases.
Since then a number of researchers looked into the application
of IR thermography as a diagnostic tool, to detect eye diseases
using thermal asymmetry [74]. They have compared thermogram
of both sides at the orbito-ocular region and concluded that, ocular
disease will be present, if a difference of 0.5 "C exists in between
the left and right eye. However, this method was shown to be
incorrect [74]. In one study involving 96 normal subjects, half of
them exhibited asymmetry at orbito-ocular region [74]. Hence
symmetry alone, may not be a good tool to differentiate normal
103
and diseased eye. Some characteristic patterns of asymmetry in
orbito-ocular thermogram was proposed for normal and patholog-
ical eyes [74].
Efron et al. have used a wide-field color-coded infrared imaging
device, to observe the variation in temperature across the ocular
surface and the temporal stability of the central cornea tempera-
ture [44]. They have observed the presence of ellipsoidal iso-
therms, with a major horizontal axis for most cases, concentric
on a temperature apex that was slightly inferior to the geometric
centre of the cornea in most of the OST [44]. Limbus was found
to be 0.45 "C warmer than geometric centre of cornea, and the rate
of corneal cooling after a blink was positively correlated to the
amount of time an eye can remain open.
Another study similar to Efron et al. was conducted to show that
limbus was 0.23–0.43 "C warmer than the geometric centre of cor-
nea [55]. All subjects recruited were Chinese, illustrating ‘bowl-
shaped’ horizontal OST profile and ‘spoon-shaped’ vertical OST
profile [55]. It was shown that Chinese eyes have lower tear vol-
ume and tear stability compared to Caucasian eyes [75–78]. These
findings suggested brown eyes in general have lower temperature
at limbus compared to blue eyes [55].
OST was shown to be decreasing by $0.01 "C per year through-
out life, and this rate of reduction increases after middle age [79]. It
was revealed that, 95% normal subjects showed an inter-ocular
temperature difference of less than 0.62 "C and more than 0.62 "C
for disease subjects [36].
Human lacrimal drainage system and the effect of non-contact
corneal esthesiometer (NCCA) air stimulus were also investigated
[42]. The anatomy and patency of human lacrimal drainage system
were usually examined using radionucleotide lacrimal scanning
and dacryocystography. But, the subjects were exposed to radia-
tions due to these techniques. Raflo et al. have employed IR ther-
mography together with lacrimal irrigation, for the purpose of
visualizing tear ducts in control and patients with obstructive
epiphora [80].
Corneal innervation has a number of specialized nerve types:
mechano-sensory, polymodal, mechano-heat and ‘cold’ neurons.
Traditional methods of assessing corneal nerve function, such as
using nylon thread of fine metal wire cause trauma to cornea.
NCCA was proposed as a new method to stimulate corneal nerves
104 J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108
utilizing controlled air pulse, though the mode of stimulation was
unclear to researchers.
Murphy et al. have conducted a study to investigate the mode of
stimulation by NCCA and concluded that, the rate of change in cor-
neal surface temperature was due to the principal mode of stimu-
lation by NCCA [42]. In other words, during stimulation both A @
(mechano-sensory) and C (temperature) fibers are likely to re-
spond to air-pulse stimulus signals from C fibers.
Dynamic IR imaging was used in the NCCA stimulation study,
and the corneal surface temperature was measured in a sequence
of 32 images with 0.25 s in between each image of subjects after
exposure to air-pulse stimulus [42]. In another recent study, dy-
namic IR imaging was utilized to investigate the relationship be-
tween OST and some physical parameters of anterior eye, such as
corneal topography, corneal thickness, bulbar hyperemia, and tear
film stability [39]. Initial OST after blinking was found to be corre-
lated with body temperature and tear film stability, as assessed by
non-invasive breakup time.
4.2. Studies of ocular diseases and surgery
Exophthalmos is a bulging or protruding eyeball. It is a medical
condition that is often seen in severe thyroid eye disease (Graves’
ophthalmopathy), an auto-immune inflammatory disorder which
affects the orbit of the eye, in patients with or without thyroid dis-
order. A number of different cases such as left endocrinal exoph-
thalmos, metastasis of left orbit were subjectively assessed with
IR thermography [23]. The number of subjects in each case was
too small to perform the statistical analysis [23].
In a later study, IR thermal imaging was utilized to determine
the inflammatory state and follow-up effect of methylprednisolone
pulse therapy in patients with Graves’ ophthalmopathy [56]. Local
temperatures such as lateral orbit (which is defined as a reference
point), upper eyelid, caruncle, medial conjunctiva, lateral conjunc-
tiva, lower eyelid and cornea were measured in patients with
Graves’ ophthalmopathy and normal control subjects. Clinical
activity score and local temperatures in some 11 patients were also
measured before and after methylprednisolone pulse therapy. The
study showed that, temperature difference between lateral orbit
and other target area such as caruncle, medial conjunctiva, lateral
conjunctiva and lower eyelid of Graves’ ophthalmopathy patients
were significantly higher than those of normal subjects [56]. Posi-
tive correlation (correlation coefficient = 0.8, n = 22) was found in
between the difference in the sum of the temperatures before
and after treatment, and the change in clinical activity score. The
IR thermal imaging was thought to be helpful in evaluating the fol-
low-up effect of methylprednisolone pulse therapy [56].
In some other studies, the effect of corneal temperature on the
insudation of lipoprotein and its mobility within corneal tissue
were studied with IR thermography [43]. They found that, the local
capillary permeability was increased when the regional differences
in temperature was greater, bringing lipoprotein insudation prefer-
entially into this warmer corneal region and making it less mobile
and relatively inert [43].
Studies on ocular diseases and surgery were conducted, with
the aid of IR thermography on post-herpetic neuralgia (PHN),
photorefractive keratectomy (PRK) and dry eye. In patients with
Herpes Zoster Ophthalmicus (HZO), it was shown that the affected
eye was warmer than the fellow eye, giving a large inter-ocular dif-
ference in temperature [81]. However, for patients with estab-
lished PHN the affected side was colder than the other side [81].
Another study indicated that for the affected side in PHN the entire
ocular surface was cool with a loss of isotherms [25] and reduction
in corneal sensitivity of more than 50% [23].
A rise in the temperature was reported during the PRK treat-
ment [49]. Studies have shown that, at 38.7 "C 50% of corneal col-
lagens denature and a complete denaturation was reached at 40 "C
[82]. The studies were conducted to investigate whether the tem-
perature for denaturation of corneal collagen on corneal surface
was routinely reached throughout the PRK operation [49]. OST
was found to reach the threshold at which the corneal collagen
denatures during treatment and there was no correlation reported
in between any surgical parameters and temperature changes.
It was reported that subjects with dry eye showed greater OST
and radial temperature difference than normal subjects [45]. Dur-
ing cooling of ocular surface, the dry eye subjects showed faster
rate of cooling than normal eye [20]. It was in agreement with
other investigators that dry eye patients have higher evaporation
rate [83,84]. The effect of conjunctival hyperemia associated with
dry eyes seems to outweigh the cooling effect given by the in-
creased evaporation, hence dry eyes were found to have higher
OST [45].
However, in another studies it was shown that the centre part
of corneal temperature of dry eyes was lower than normal eyes,
though mean OST was not evaluated [53]. Similar studies con-
ducted by research group in Japan [50,85], showed that central cor-
neal temperature in dry eye patients were higher than normal
control groups. Such differences were attributed to the different
population of dry eye patients recruited in studies [53].
Besides, Craig et al. have also indicated that, the temperature
variation factor and mean osmolality were higher in dry eye groups
than control subjects [53]. Temperature variation factor was found
to be inversely correlated to central corneal temperature. Hence,
the ocular surface having higher variation in temperature tends
to have lower central corneal temperature.
Corneal surface temperature decreases exponentially after eye
opens, and it asymptotically approaches a constant value after
some time [50]. This can be modeled by the following formula:
TðtÞ ¼ ðT 0 $ T 1 Þe$kt þ T 1 ð8Þ
In which T is the corneal surface temperature after the eye opens for
a period of t. T 0 is the temperature immediately the eye opens; T 1 is
the corneal surface temperature at equilibrium; k is the tempera-
ture coefficient. By measuring temperature values within a box of
20 % 20 pixels, or 3.3 mm2, placed at the centre of cornea over a
period of 30 s, it was found that for normal blinking the mean k va-
lue in dry eye patients (5.6 ± 2.9 per s) was significantly less than in
normal control subjects (9.3 ± 5.0 per s) [50]. It was suggested that k
value might reflect tear film stability [34].
A later studies employed principle similar to the one mentioned
above to diagnose dry eye using IR thermography. Twenty six nor-
mal and 82 dry eye patients were recruited in that studies. An
encircled region having diameter of 4.4 mm (22 pixels) was posi-
tioned at the centre of eye to acquire OST [51]. The study showed
that, the diagnosis of dry eye patients was accurate with a sensitiv-
ity of 79%, specificity of 75% with 0.841 area under ROC (receiver
operating characteristics) curve 0.42 "C/s [51].
Furthermore, dry eye was studied in conjunction with the treat-
ment of acupuncture and IR thermography to determine the effect
of acupuncture [86]. It was reported that acupuncture affects the
temperature of the pre-corneal tear film, and lowers the tempera-
ture at the middle of cornea.
Some studies have suggested a correlation in between OST and
ocular blood flow. In monkeys, intraocular pressure (IOP) was
found inversely related to ocular perfusion pressure and ocular
temperature [87]. Corneal temperature was determined to be pos-
itively correlated to the ipsilateral values of end diastolic velocity
(in left and right eyes, respectively), and it also positively correlates
to the resistivity index of left and right eyes, respectively; inter-
ocular difference in corneal temperature was positively correlated
with the difference in end diastolic velocity, and was found to be
negatively correlated with resistivity index [88]. The above
 J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108
study was done with IR thermometer instead of IR thermographer
[88].
Galassi et al. have evaluated the OST of patients with primary
open-angle glaucoma (POAG) and control group through IR ther-
mography [46]. They have investigated the correlation between
OST, IOP and retrobulbar hemodynamics in conjunction with the
use of color Doppler imaging (CDI). Temperatures of five anatomi-
cal points (internal and external canthus, half-way from the inter-
nal canthus and nasal limbus, centre of the cornea, half-way from
the temporal limbus and external canthus) of POAG patients were
found to be lower than of healthy control group. These tempera-
tures were significantly correlated to resistivity index [46]. Their
results highlight the influence of retrobulbar hemodynamics on
OST [46]. In other study, higher OST was observed for central reti-
nal vein occlusion (CRVO) compared to normal subjects and lower
OST was shown for ischemic CRVO eyes compared to non-ischemic
ones [48].
The effects of two glaucoma surgeries, namely deep sclerec-
tomy and trabeculectomy, on bulbar hemodynamics and corneal
surface temperature were investigated [47]. Corneal surface tem-
perature was acquired [46]. Three months after operation, the cor-
neal surface temperature increased in both type of surgeries, and a
negative correlation between postoperative changes in ophthalmic
artery resistivity index and corneal surface temperature was ob-
served in both types of surgeries [47]. They have suggested that
IR thermography was useful in the evaluation of vascular outcome
of glaucoma surgery [47].
The cataract patients with an anterior capsulotomy, had corneal
temperature significantly higher not only just after the surgery, but
also at the 30th-day after the operation [89]. The length of surgery
lasted for more than 40 min, and IR thermometer was used in the
study. No follow-up investigation with IR thermographer was done
to elucidate further detail.
IR thermography was employed to analyze and compare differ-
ent cataract surgery procedures based on the phacoemulsification
system. In one of the studies, phacoemulsification probes from
Alcon Legacy AdvanTec, Bausch & Lomb Millennium and AMO sov-
ereign WhiteStar were placed in air, and testing condition was set
in a way such that a corneal burn might be produced during cata-
ract surgery [90]. It was found that, Millennium and Sovereign
WhiteStar generated more heat and therefore led to higher tem-
perature than Legacy AdvanTec phacoemulsification system [90].
Three different cataract surgeries performed in vivo with phac-
oemulsification (phacoemulsification with the traditional Sover-
eign system without WhiteStar technology, phacoemulsification
with the Sovereign WhiteStar system, and phacoemulsification
through micro incision cataract surgery with the bimanual Sover-
eign WhiteStar system) were compared using IR thermography
[91]. Among these three techniques phacoemulsification through
micro incision cataract surgery with the bimanual Sovereign
WhiteStar system was determined to have the lowest thermal im-
pact on eye [91]. In another study, intraoperative thermal levels at
wound site were studied, with Sovereign WhiteStar system and
Legacy Advantec and NeoSoniX system. Legacy AdvanTec and
NeoSoniX system produced grater mean temperature change at
wound site and higher mean peak temperatures in patients [92].
These studies showed that IR thermography has facilitated effec-
tive analysis on cataract surgery with phacoemulsification.
4.3. Studies of non-ocular diseases
Carotid artery stenosis (CAS) was investigated, using color-
coded infrared ocular thermography [21]. They have shown that,
the OST was negatively correlated with the degree of CAS and
the relative difference in CAS (CAS of right eye minus that of left
eye) was also negatively correlated with the relative difference in
105
OST (OST of right eye minus that of left eye). They have concluded
that the circle of Willis and the other anastomoses within the brain
are not able to compensate for the reduction in blood flow and
hence led to a cooler eye on the affected side [21].
5. Discussion
IR thermography is an efficient tool not only to capture temper-
atures of corneal surface, but also to detect and visualize any subtle
changes on the OST. Eye is a delicate organ and highly susceptible
to external variations. The non-intrusiveness of this technique has
encouraged many researchers to use this method to study ocular
physiology, instead of other imaging techniques and methods for
measuring the OST.
Zeiss pioneered the use of thermometry in the measurement of
ocular temperature [93]. However, Mapstone [32] had first intro-
duced IR thermal imaging into the field of ophthalmology. Since
its inception, several researchers have looked into thermal asym-
metry or anomalies in thermal distribution, both on the eyes and
region surrounding the eyes. They have tried to establish diagnos-
tic criteria from these asymmetry and anomalies. Due to the poor
resolution of IR thermal imaging at that time, the eye on thermo-
gram was not clearly discernible. Hence, it was not possible to pro-
pose diagnostic criteria based on these asymmetry and anomalies.
Due to the rapid advancement in the IR thermography, more de-
tailed thermal distribution of the anterior eye can be obtained.
The better technology has prompted researchers to focus their
studies on the thermographic pattern of the anterior eye, instead
of the entire orbito-ocular region. The horizontal temperature pro-
file passing through the geometric centre of eye was illustrated,
and more recently, the vertical temperature profile is also
described.
Finite element analysis on human eye has shown that, the cen-
tre of the corneal surface has the lowest temperature [94]. How-
ever, experimental studies showed that the coolest point on the
cornea is slightly inferior to the geometric centre of cornea
[36,44,55], and this observation was evident in most of the normal
subjects [44]. Researchers suggested that the presence of upper
eyelid, which is a source of heat, shifted the temperature apex
(the coolest point) inferiorly instead of being at the centre of
cornea.
Besides, the shape of isotherms predicted by finite element
analysis was not the same as the experimental studies. The pres-
ence of eyelid has made the isotherms as an elliptical shape, in-
stead of a circular shape [94].
So far thermographic pattern of normal subjects’ OST have been
reported. But for other ocular diseases the corresponding thermo-
graphic pattern has not been reported. Subjective assessment was
the only method employed on studies of OST of diseased eye con-
ducted during 1970s and 1980s; for later studies, mean tempera-
tures on one or several sites (in this context, a site refers to a
point or an area) over the anterior eye were utilized to study ocular
diseases.
From the viewpoint of image processing, when one acquires
mean temperature, standard deviation or median over a region
or an area on ocular thermogram, he/she is in fact acquiring some
first-order spatial statistics of texture analysis on the thermal im-
age. As an example, to obtain the mean temperature in the above
case, the average pixel intensity over the region is calculated and
converted to the corresponding temperature. The average pixel
intensity is the first-order spatial statistics.
Often researchers employed mean temperature as a measure to
get a ‘‘middle” or ‘‘expected” temperature value of the site of inter-
est, and made comparison with the corresponding value obtained
from normal subjects. Texture analysis was not actually considered
in their studies. But in coming future such technique may prove to
 106
Table 1
Summary of methods used to estimate OST.

Studies Types of Methods Main findings

OST
acquisition
Efron et al. Manual 11 Points running across the anterior surface Average difference in temperature between the GCC and the limbus is 0.45 "C
[44]
Morgan et al. Manual Five 10 % 10 boxes placed in five different anatomical locations along horizontal In [21], a significant negative correlation was found between OST and the degree of CAS [21]
[21,36,45] meridian running across the estimated centre of cornea In [36], 95% normal subjects were found to have inter-ocular temperature differences between $0.57 and
0.62 "C [36]
In [45], dry eye patients were found to have greater mean OST than normal subjects, and RTD was also
found to be greater in dry eye group [45]
Galassi et al. Manual Five points placed on centre of cornea, internal and external canthus, half-way from the For all five points temperature of patients of POAG are lower than temperature of healthy control group
[46,47] internal canthus and nasal limbus, half-way from the temporal limbus and external [46]

J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108

canthus A negative correlation between postoperative changes in ophthalmic artery resistivity index and corneal
surface temperature was observed 3 months after deep sclerectomy and trabeculectomy surgeries.
Sodi et al. Manual Five points equally placed along a horizontal line running through centre of cornea, In central retinal vein occlusion (CRVO) eyes and in fellow, non-affected eyes, OST values were lower than
[48] connecting medial and lateral canthi in controls [48]
Murphy et al. Manual A squared 10 % 10 pixels box placed at the centre of cornea In [42], using the NCCA air pulse, the principal mode of corneal nerve stimulation was the rate of
[42], temperature change of the corneal surface [42]
Betney et al. In [49], no correlation was demonstrated between any surgical parameters and changes either in absolute
[49] or relative temperature [49]
Mori et al. Manual A squared 20 % 20 pixels box placed at the centre of cornea The rate of decline in corneal temperature for normal blinking in patients with dry eye was significantly
[50] less than that in normal subjects [50]
Chiang et al. Manual An encircled region of 4.4 mm diameter (22 pixels) Diagnostic of dry eye using infrared thermal imager system has reached a sensitivity of 79.3%, a specificity
[51] of 75 [51]
Ng et al. [52] Manual A small circle placed at the centre of cornea OST was found to decrease with age at a rate of 0.0383 "C per year [52]
Cardona et al. Manual A circular region centred at the corneal centre In [25], patients with HZO in acute stage were found to show high inter-ocular temperature differences,
[25] with the affected side being warmer than the other side [25]
Purslow and Manual In [39], strong positive correlations were found in between initial OST and both body (forehead)
Wolffsohn temperature and tear film stability [39]
[39]
Craig et al. Manual In [53], dry eye patients were found to have lower central cornea temperature and higher mean
[53] osmolality [53]
Purslow et al. Manual 23 Points placed across the anterior eye OST is greater eye with hydrogel and greater still with silicone hydrogel contact lenses in situ, regardless
[54] of modality of wear [54]
Tan et al. [55] Manual 20 Points placed across the anterior eye, lined up in the shape of ‘‘+” Vertical OST profile was noted to be a ‘spoon-shaped’; for Chinese the average difference in temperature
between the GCC and the limbus is 0.23–0.43 "C [55]
Chang et al. Manual Acquire local temperatures of lateral orbit, upper eyelid, caruncle, medial conjunctiva, Temperature difference between lateral orbit and three other target areas (caruncle, medial conjunctiva,
[56] lateral conjunctiva, lower eyelid and cornea lateral conjunctiva and lower eyelid) of GO patients were found significantly higher than those of normal
subjects [56]
Acharya et al. Semi-auto Image was manually cropped to consist only of eye, the cornea was then detected by Average OST and temperature deviation across the cornea decreases with age [57]
[57] algorithm developed
Tan et al. [58] Automated The eye was localized by genetic snake algorithm, and the corneal diameter and location The algorithm was able to localize 90% of IR ocular thermogram [58]
were derived from the resultant snake points
 J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108
be an alternative approach to effectively study thermographic pat-
tern of normal and diseased eye, other than descriptively or quan-
titatively accounting the pattern of isotherms.
The study of OST on IR thermogram needs more analysis tools
and efforts to deliver better diagnostic value. It is known that eye
thermogram lacks corneal boundary, which is clearly visible in
optical image. This makes the quantification of corneal tempera-
ture (or OST) a difficult task to achieve due to the difficulty in cor-
neal localization, and this problem is approached with numerous
methods as introduced in Section 3 and summarized in Table 1.
For manual acquisition of OST, the centre of cornea is located by
subjective judgment, by placing few boxes or points at other loca-
tions with respect to the centre [20,42,49,50,52]. In some other
studies, a circular region was defined at the centre of cornea. These
methods do not indicate the actual corneal temperature and accu-
racy depends on the number and position of the boxes.
Semi-automated method is an improved method. It provides a
set of rules and procedures to define and locate cornea, though
in some cases there is some slight error in accuracy with regard
to the location of corneal region. However, human intervention is
required before the start of the proposed algorithm to crop the
eye exactly in the proper location.
The snake algorithm coupled with target tracing function [58]
localizes the cornea and eye automatically. This recent method
can be used to evaluate the OST and hence the eye diseases. And
the acquired temperature profile (similar to Fig. 12) can be used
as tool to diagnose the ocular diseases.
Eye is a delicate organ, and ocular IR thermography brings no
harm to subjects’ eye under study or diagnosis. But it is unable
to provide thermal information and description of anatomical fea-
tures beneath the anterior eye. Unlike ultrasound imaging, IR ther-
mography cannot provide description (or image) on the posterior
segment of eye, nor can it accurately determine the size and loca-
tion of uveal melanoma which computed tomography is capable of.
However, IR thermal imaging does not emit ionizing radiation as
computed tomography does, or generate high-frequency acoustic
wave, expose part of body to powerful magnetic field and infrared
light. It passively receives IR radiation emitted by subjects, and
does not cause any discomfort either through contact with instru-
ment or expose the eye to any electromagnetic and acoustic wave.
IR thermal imaging is able to detect pathological changes in
eyes which are obscured under examination by other imaging
techniques and methods. It was reported that, for mild inflamma-
tion it may not be observed under examination of slit lamp biomi-
croscope since the changes were too subtle but sufficient to cause a
measurable increase in OST [55].
The IR thermal imaging is non-invasiveness and is an excellent
diagnostic tool. As an example, the diagnosis of dry eye syndrome
mainly based on measurement of tear film production, such as
Schirmer test, tear breakup time method, and Rose Bengal stain-
ing method. However, Schirmer test requires contact with eye
and the use of topical anesthesia, tear breakup time method
requires patients to hold their eye for a long time, and often it
is not easy to do so, and Rose Bengal staining method is an inva-
sive method, patients can feel uncomfortable. The diagnosis of dry
eye syndrome using IR thermal imaging has none of the above
problems.
IR Thermography can be extensively used as an effective non-
invasive ocular temperature measuring tool. The advancement in
this area offers a great potential to study and measure OST with
greater ease and precision. This enables researchers to look into
various aspects in the field of ophthalmology, and understand ocu-
lar physiology and diseases. However, to establish IR thermogra-
phy as a tool to diagnose ocular diseases requires more robust
automatic cornea localization and the use of sophisticated tools,
such as image analysis, and statistical analysis.
107
References
[1] E.F.J. Ring, Progress in the measurement of human body temperature, IEEE
Engineering in Medicine Biology Magazine 17 (1998) 19–24.
[2] R.N. Lawson, Implications of surface temperature in the diagnosis of breast
cancer, Canadian Medical Association Journal 75 (1956) 309–310.
[3] R.N. Lawson, Thermography: a new tool in the investigation of breast lesions,
Canadian Services Medical Journal 13 (1957) 517.
[4] C. Meola, G.M. Carlomagno, Recent advances in the use of infrared
thermography, Measurement Science and Technology 15 (2004) R27–R58.
[5] M. Wisniewski, S. Lindow, E. Ashworth, Observations of ice nucleation and
propagation in plants using infrared video thermography, Plant Physiology 113
(1997) 327–334.
[6] M. Cehlin, B. Moshfegh, M. Sandberg, Visualization and measuring of air
temperatures based on infrared thermography, in: Proceedings of the 7th
International Conference on Air Distribution in Rooms ROOMVENT, Reading,
UK, 2000.
[7] I. Pavlidis, J. Levine, Thermal image analysis for polygraph testing, IEEE
Engineering in Medicine and Biology Magazine 21 (2002) 56–64.
[8] W. Ruddock. Advanced infrared resources. <http://www.infraredthermography.
com>, 2008 (accessed 5.05.09).
[9] H. Hooshmand, M. Hashmi, E.M. Phillips, Infrared thermal imaging as a tool in pain
management an 11 year study: I, Thermology International 11 (2001) 53–65.
[10] H. Hooshmand, M. Hashmi, E.M. Phillips, Infrared thermal imaging as a tool in
pain management an 11 year study: II Clinical applications, Thermology
International 11 (2001) 119–129.
[11] G. Stillwagon, L.K. Stillwagon, Vertebral subluxation correction and its affect
on thermographic readings: a description of the advent of the visi-therm as
applied to chiropractic patient assessment, Journal of Vertebral Subluxation
Research 2 (1998) 1–4.
[12] E.Y.K. Ng, Y. Chen, Segmentation of breast thermogram: improved boundary
detection with modified snake algorithm, Journal of Mechanics in Medicine
and Biology 6 (2006) 123–136.
[13] E.Y.K. Ng, A review of thermography as promising non-invasive detection
modality for breast tumor, International Journal of Thermal Sciences 48 (2008)
849–859.
[14] T.K. Kalili, B.M. Gratt, Electronic thermography for the assessment of acute
temporomandibular joint pain, Compendium of Continuing Education in
Dentistry 10 (1996) 979–983.
[15] E.Y.K. Ng, U.R. Acharya, A review of remote-sensing infrared thermography for
indoor mass blind fever screening in containing an epidemic, IEEE Engineering
in Medicine and Biology Magazine 28 (2008) 76–83.
[16] A. Merla, G.L. Romani, A. Tangherlini, S.D. Romualdo, M. Proietti, E. Rosato, A.
Aversa, F. Salsano, Penile cutaneous temperature in systemic sclerosis: a
thermal imaging study, International Journal of Immunopathology and
Pharmacology 20 (2007) 139–144.
[17] A. Helmy, M. Holdmann, M. Rizkalla, Application of thermography for non-
invasive diagnosis of thyroid gland disease, IEEE Transactions on Biomedical
Engineering 55 (2008) 1168–1175.
[18] P.B. Stefanie. Thermotopography shows ’enormous promise’ for diagnosis and
treatment of eye diseases. <http://www.escrs.org/eurotimes/March2003/
thermo.asp>, 2009 (accessed 5.05.09).
[19] T. Rosenstock, P. Chart, J.J. Hurwitz, Inflammation of the lacrimal drainage
system-assessment by thermography, Ophthalmic Surgery 14 (1983) 229–
237.
[20] P.B. Morgan, A.B. Tullo, N. Efron, Ocular surface cooling in dry eye: a pilot
study, Journal of British Contact Lens Association 19 (1996) 7–10.
[21] P.B. Morgan, J.V. Smyth, A.B. Tullo, N. Efron, Ocular temperature in carotid
artery stenosis, Optometry and Vision Science 72 (1999) 850–854.
[22] S. Tokuoka, M. Nakajima, J. Nishikawa, H. Kuroda, I. Azuma, Cold response of
skin and eye temperature in glaucoma, Folia Ophthalmology Japan 41 (1990)
1159–1165.
[23] P. Bourjat, M. Gautherie, Unilateral exophthalmos investigated by infrared
thermography, Modern Problems in Ophthalmology 14 (1975) 278–285.
[24] J. Hannerz, Pain characteristics of painful ophthalmoplegia (the Tolosa–Hunt
syndrome), Cephalalgia 5 (1985) 103–106.
[25] G. Cardona, P.B. Morgan, N. Efron, A.B. Tullo, Ocular and skin temperature in
ophthalmic postherpetic neuralgia, Pain Clinic 9 (1996) 145–150.
[26] I.B. Bogdasarov, O.P. Lenskaia, B.M. Belkina, Thermography in the diagnosis of
retinoblastoma in children, Meditsinskaia Radiologiia 30 (1985) 19–21.
[27] A.I. Eremenko, Thermography in the diagnosis of vascular neuritis of the optic
nerve, Oftalmologicheskii Zhurnal 4 (1990) 235–239.
[28] R.F. Rosenbluth, I. Fatt, Temperature measurements in the eye, Experimental
Eye Research 25 (1977) 325–341.
[29] J.L. Miller, Principles of Infrared Technology: A Practical Guide to the State of
the Art, Van Nostrand Reinhold, 1994.
[30] J.L. Miller, E. Friedman, Photonics Rules of Thumb: Optics, Electro-Optics, Fiber
Optics and Lasers, McGraw-Hill, 2004.
[31] H. Hartridge, A.V. Hill, The transmission of infra-red rays by the media of the
eye and the transmission of radiant energy by Crookes and other glasses,
Proceedings of the Royal Society 89 (1915) 58–76.
[32] R. Mapstone, Measurement of Corneal Temperature, Experimental Eye
Research 7 (1968) 237–243.
[33] E.K. Plyler, N. Aquista, IR absorption of liquid water from 2 to 42 microns,
Journal of the Optical Society of America 44 (1954) 505–508.
108 J.-H. Tan et al. / Infrared Physics & Technology 52 (2009) 97–108
[34] H. Hamano, S. Minami, Y. Sugimori, Experiments in thermometry of the
anterior portion of the eye wearing a contact lens by means of infra-red
thermometer, Contact Lens and Anterior Eye 13 (1969) 12–22.
[35] J.I. Prydal, P. Artal, H. Woon, F.W. Campbell, Study of human precorneal tear
film thickness and structure using laser interferometry, Investigative
Ophthalmology and Visual Science 33 (1992) 2006–2011.
[36] P.B. Morgan, M.P. Soh, N. Efron, A.B. Tullo, Potential Applications of Ocular
Thermography, Optometry Vision Science 70 (1993) 568–576.
[37] I. Fatt, J. Chaston, Temperature of a contact lens on the eye, International
Contact Lens Clinic 7 (1980) 195–198.
[38] J.A. Scott, A finite element model of heat transport in the human eye, Physics in
Medicine and Biology 33 (1988) 227–241.
[39] C. Purslow, J.S. Wolffsohn, The relation between physical properties of the
anterior eye and ocular surface temperature, Optometry and Vision Science 84
(2007) 197–201.
[40] R. Mapstone, Ocular thermography, British Journal of Ophthalmology 54
(1970) 751–754.
[41] L. Wachtmeister, Thermography in the diagnosis of diseases of the eye and the
appraisal of therapeutic effects: a preliminary report, Acta Ophthalmologica 48
(1970) 945–958.
[42] P.J. Murphy, P.B. Morgan, S. Patel, J. Marshall, Corneal surface temperature
change as the mode of stimulation of the non-contact corneal aesthesiometer,
Cornea 18 (1999) 333–342.
[43] A.R. Fielder, A.F. Winder, G.A.K. Sheraidah, E.D. Cooke, Problems with corneal
arcus, Transactions of the Ophthalmological Societies of the United Kingdom
101 (1981) 22–26.
[44] N. Efron, G. Young, N. Brennan, Ocular surface temperature, Current Eye
Research 8 (1989) 901–906.
[45] P.B. Morgan, A.B. Tullo, N. Efron, Infrared thermography of the tear film in dry
eye, Eye 9 (1995) 615–618.
[46] F. Galassi, B. Giambene, A. Corvi, G. Falaschi, Evaluation of ocular surface
temperature and retrobulbar haemodynamics by infrared thermography and
colour Doppler imaging in patients with glaucoma, British Journal of
Ophthalmology 91 (2007) 878–881.
[47] F. Galassi, B. Giambene, A. Corvi, G. Falaschi, U. Menchini, Retrobulbar
hemodynamics and corneal surface temperature in glaucoma surgery,
International Ophthalmology 28 (2008) 399–405.
[48] A.A. Sodi, B.A.D. Giambene, G.B. Falaschi, R.C. Caputo, B.B. Innocenti, A.B. Corvi,
U.A. Menchini, Ocular surface temperature in central retinal vein occlusion:
preliminary data, European Journal of Ophthalmology 17 (2007) 755–759.
[49] S. Betney, P.B. Morgan, S.J. Doyle, N. Efron, Corneal temperature changes
during photorefractive keratectomy, Cornea 16 (1997) 158–161.
[50] A. Mori, Y. Oguchi, Y. Okusawa, M. Ono, H. Fujishima, K. Tsubota, Use of high-
speed, high-resolution thermography to evaluate the tear film layer, American
Journal of Ophthalmology 124 (1997) 729–735.
[51] H.K. Chiang, C.Y. Chen, H.Y. Cheng, K.-H. Chen, D.O. Chang, Development of
infrared thermal imager for dry eye diagnosis, in: Proceedings of SPIE – The
International Society for Optical Engineering, San Diego, CA, USA, 2006.
[52] E.Y.K. Ng, J.H. Tan, E.H. Ooi, C. Chee, U.R. Acharya, Variations of Ocular Surface
Temperature with Different Age Groups, Image Modeling of Human Eye,
Artech House, 2008.
[53] J.P. Craig, I. Singh, A. Tomlinson, et al., The role of tear physiology in ocular
surface temperature, Eye 14 (2000) 635–641.
[54] C. Purslow, J.S. Wolffsohn, S. Santodomingo-Rubido, The effect of contact lens
wear on dynamic ocular surface temperature, Contact Lens and Anterior Eye
28 (2005) 29–36.
[55] L. Tan, Z.-Q. Cai, N.-S. Lai, Accuracy and sensitivity of the dynamic ocular
thermography and inter-subjects ocular surface temperature (OST) in Chinese
young adults, Contact Lens and Anterior Eye 32 (2009) 78–83.
[56] T.-C. Chang, Y.-L. Hsiao, S.-L. Liao, Application of digital infrared thermal
imaging in determining inflammatory state and follow-up effect of
methylprednisolone pulse therapy in patients with Graves’ ophthalmopathy,
Graefe’s Archive for Clinical and Experimental Ophthalmology 246 (2008) 45–
49.
[57] U.R. Acharya, E.Y.K. Ng, C.Y. Gerk, J.H. Tan, Analysis of normal human eye with
different age groups using infrared images, Journal of Medical System 33
(2008) 207–213.
[58] J.H. Tan, E.Y.K. Ng, A.U. Rajendra, Automated detection of eye and cornea on
infrared thermogram using snake and target tracing function coupled with
genetic algorithm. Quantitative InfraRed Thermography International Journal,
in press.
[59] J.H. Tan, E.Y.K. Ng, U.R. Acharya, Detection of eye and cornea on IR thermogram
using genetic snake algorithm, in: 9th International Conference on
Quantitative Infrared Thermography, Krakow, Poland, 2008, pp. 143–150.
[60] M. Kass, A. Witkin, D. Terzopoulos, Snakes: active contour models,
International Journal of Computer Vision 1 (1988) 321–331.
[61] C. Xu, J.L. Prince, Snakes, shapes, and gradient vector flow, IEEE Transactions on
Image Processing 7 (1998) 359–369.
[62] A. Del Bimbo, P. Pala, Visual image retrieval by elastic matching of user
sketches, IEEE Transactions on Pattern Analysis and Machine Intelligence 19
(1997) 121–132.
[63] D. Cremers, C. SchnörrJ, J. Weickert, Diffusion-snakes: combining statistical
shape knowledge and image information in a variational framework, in:
Workshop on Variational and Level-set Methods in Computer Vision,
Vancouver, BC, Canada, 2001, pp. 137–144.
[64] T. Cootes, C.J. Taylor, D.H. Cooper, J. Graham, Active shape models-their
training and application, Computer Vision Image Understanding 61 (1995) 38–
59.
[65] K. Schindler, D. Suter, Object detection by global contour shape, Pattern
Recognition 41 (2008) 3736–3748.
[66] A.L. Yuille, D.S. Cohen, P.W. Hallinan, Feature Extraction from Faces Using
Deformable Templates, in: Proceedings of CVPR ’89, IEEE Computer Society
Conference, San Diego, CA, USA, 1989, pp. 104–109.
[67] P. Rysä, J. Savaranta, Corneal temperature in man and rabbit. Observations
made using an infra-red camera and a cold chamber, Acta Ophthalmologica 52
(1974) 810–816.
[68] R. Mapstone, Determinants of corneal temperature, British Journal of
Ophthalmology 52 (1968) 729–741.
[69] P.B. Morgan, Ocular thermography in health and disease, Ph.D. Thesis,
University of Manchester, Manchester, UK, 1994.
[70] R.D. Freeman, I. Fatt, Environmental influences on ocular temperature,
Investigative Ophthalmology and Visual Science 12 (1973) 596–602.
[71] R. Mapstone, Corneal thermal patterns in anterior uveitis, British Journal of
Ophthalmology 52 (1968) 917–921.
[72] I. Horven, Corneal temperature in normal subjects and arterial occlusive
disease, Acta Ophthalmologica (Copenhagen) 53 (1975) 863–874.
[73] N. Efron, N. Brennan, J. Hore, et al., Temperature of the hyperaemic bulbar
conjunctiva, Current Eye Research 7 (1988) 615–618.
[74] J. Alio, M. Padron, Normal variations in the thermographic pattern of the
orbitoocular regionocular region, Diagnostic Imaging 51 (1982) 93–98.
[75] P. Cho, B. Brown, I. Chan, R. Conway, M. Yap, Reliability of the tear break-up
time technique of assessing tear stability and the locations of the tear break-up
in Hong Kong Chinese, Optometry and vision science 69 (1992) 879–885.
[76] P. Cho, B. Brown, Review of the TBUT technique and a closer look at the TBUT
of HK-Chinese, Optometry and vision science 70 (1993) 30–38.
[77] P. Cho, M. Yap, Age, gender, and tear break-up time, Optometry and vision
science 70 (1993) 828–831.
[78] S. Patel, S. Laidlaw, L. Mathewson, L. McCallum, C. Nicholson, Iris colour and
the influence of local anaesthetics on precorneal tear film stability, Acta
Ophthalmol (Kbh) 69 (1991) 387–392.
[79] P.B. Morgan, M.P. Soh, N. Efron, Corneal surface temperature decrease with
age, Contact Lens and Anterior Eye 22 (1999) 11–13.
[80] G.T. Raflo, P. Chart, J.J. Hurwitz, Thermographic evaluation of the human
lacrimal drainage system, Ophthalmic Surgery 13 (1982) 119–124.
[81] A.B. Tullo, G. Cardona, P.B. Morgan, N. Efron, Ocular and facial thermography in
herpes zoster ophthalmicus and post-herpetic neuralgia, Investigative
Ophthalmology and Visual Science 37 (1996) S49 (ID code: 11628).
[82] M. Lewis, M. Dubin, V. Aandahl, Physical properties of bovine corneal collagen,
Experimental Eye Research 6 (1967) 57–69.
[83] W.D. Mathers, G. Binarao, M. Petroll, Ocular water evaporation and the dry
eye: a new measuring device, Cornea 12 (1993) 335–340.
[84] M. Rolando, M. Refojo, K. Kenyon, Increased tear evaporation in eyes with
keratoconjunctivitis sicca, Archives of Ophthalmology 101 (1983) 557–558.
[85] H. Fujishima, I. Toda, M. Yamada, N. Sato, K. Tsubota, Corneal temperature in
patients with dry eye evaluated by infrared radiation thermometry, British
Journal of Ophthalmology 80 (1996) 29–32.
[86] J. Nepp, K. Tsubota, E. Goto, J. Schauersberger, G. Schild, K. Jandrasits, C. Abela,
A. Wedrich, The effect of acupuncture on the temperature of the ocular surface
in conjunctivitis sicca measured by non-contact thermography: preliminary
results, Advances in Experimental Medicine and Biology 506 (2002) 723–726.
[87] C.R. Auker, L.M. Parver, S. Doyle, et al., Choroidal blood flow. I. Ocular tissue
temperature as a measure of flow, Archives of Ophthalmology 100 (1982)
1323–1326.
[88] K. Gugleta, S. Orgül, J. Flammer, Is corneal temperature correlated with blood-
flow velocity in the ophthalmic artery?, Current Eye Research 19 (1999) 496–
501
[89] H. Fujishima, I. Toda, Y. Yagi, K. Tsubota, Quantitative evaluation of
postsurgical inflammation by infrared radiation thermometer and laser flare-
cell meter, Journal of Cataract and Refractive Surgery 20 (1994) 451–454.
[90] M.D. Olson, K.M. Miller, In-air thermal imaging comparison of Legacy
Advantec, millennium, and Sovereign Whitestar phacoemulsification
systems, Journal of Cataract and Refractive Surgery 31 (2005) 1640–1647.
[91] A. Corvi, B. Innocenti, R. Mencucci, Thermography used for analysis and
comparison of different cataract surgery procedures based on
phacoemulsification, Physiological Measurement 27 (2006) 371–384.
[92] A.D. Rose, V. Kanade, Thermal imaging study comparing phacoemulsification
with the Sovereign with WhiteStar system to the Legacy with AdvanTec and
NeoSoniX system, American Journal of Ophthalmology 141 (2006) 322–326.
[93] E. Zeiss, Über Wärmestrahlungsmessungen an der lebenden menschlichen
Hornhaut, Arch Augenheilkd 102 (1930) 523–550.
[94] E.Y.K. Ng, E.H. Ooi, Ocular surface temperature: a 3D FEM prediction using
bioheat equation, Computers in Biology and Medicine 37 (2007) 829–835.