Professional Documents
Culture Documents
A Comparison of Image Reject Rates When Using Film PDF
A Comparison of Image Reject Rates When Using Film PDF
net/publication/12690614
A comparison of image reject rates when using film, hard copy computed
radiography and soft copy images on picture archiving and communication
systems (PACS) workstations
CITATIONS READS
40 421
3 authors, including:
Stirling Bryan
University of British Columbia - Vancouver
283 PUBLICATIONS 9,872 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
The Use of Economic Evaluation in Health Care Decision Making View project
All content following this page was uploaded by Stirling Bryan on 30 May 2014.
Abstract. A comparison has been made of the reject rates of plain images for three separate
periods when ®lm, computed radiography (CR) and PACS systems were in operation throughout
the Hammersmith Hospital, London. There was a statistically signi®cant reduction in the overall
percentage reject rate across all examinations from 9.9% to 8.1% when the hospital changed from
using a conventional ®lm based system to a CR system. There was a further reduction in the
reject rate to 7.3% when the hospital moved to a hospital-wide PACS system, but this change was
not statistically signi®cant. Using estimations of the total number of images used, the percentage
reject rates were 6.6% for ®lm, 5.5% for CR and 5.5% for PACS. Thus, if the radiation dose for
each image is unchanged, and the same types of images are used for the examination of each body
area, a move from conventional ®lm imaging to phosphor plate imaging provides the potential to
reduce the patient population dose.
``Reject analysis'' is described as ``the critical [12]. For images of body areas where there are
evaluation of radiographs which are used as part large differences in density and thickness within
of the imaging service but do not play a useful the patient, or there is unexpected pathology
part in the diagnostic process'' [1]. The analysis of which causes the area to be either particularly
rejected ®lms gives an indication of the sources of radiolucent or dense, manipulation of soft copy
radiographic errors and highlights areas where images should allow most areas of the images to
improvements can be made. Reject rates vary be visualized [13]. Thus, when CR systems are
between 2% and 13% of all ®lms taken [2±9]. In incorporated into a picture archiving and com-
1990, the average reject rate in the UK was 10% munication system (PACS), which includes
of all ®lms [10]. This implies that, nationally, manipulation facilities such as grey scale and
rejected ®lms are responsible for an unnecessary density variation, the reject rates should be lower
increase in the radiation dose to the patient than those for both ®lm and CR. A General
population [11] and for additional staff, ®lm and Electric (GE) hospital-wide PACS is now in
processing costs. operation at the Hammersmith Hospital,
The introduction of computed radiography London and has been the subject of an indepen-
(CR) systems is expected to reduce reject rates dent evaluation by the Health Economics
because the phosphor plates used for acquiring Research Group (HERG) at Brunel University
the images have a wider latitude than conven- [14]. As part of this evaluation it was possible to
tional ®lm and repeats resulting from incorrect compare the results of reject analyses when each
exposure factors are therefore largely eliminated of the three types of radiographic images were
used: ®rst with Film, second with CR, and third
with PACS soft copy images.
Two hypotheses were tested in this study:
Received 26 November 1998 and in revised form 17
March 1999, accepted 24 March 1999. (1) the reject rate of images would be reduced
This study was one component of the evaluation of a after the introduction of phosphor plate
hospital-wide picture archiving and communication computed radiography; and
system which was undertaken by the Health
Economics Research Group and funded by a grant (2) the reject rate would be reduced further after
from the Policy Research Programme of the the introduction of PACS workstations with
Department of Health. manipulation facilities for soft copy images.
box in the ®lm viewing area. At the end of the CR image is required, the ®lm has to be
study all rejected ®lms were viewed in order to transported via a roller system to a standard
identify the body areas examined and the reason chemical processor. Faults in this system are
for rejection of the image. On the few occasions similar to processing faults of ®lm as described
when a label was missing, one of the authors previously in section E.
(MW), in consultation with a radiographer from
the department, coded the ®lm for the reason for
rejection.
Soft copy PACS images
Hard copy CR images This study took place during November and
December 1995, 18 months after the department
This part of the study was undertaken in began to use PACS and at a time when the staff
September 1994 at a time when the CR system had adapted to using the new system. The method
had been in use in the hospital for 18 months and used here was different from the previous two
radiology staff were generally satis®ed with the rounds because of changes which had occurred
system. The method used was essentially the same with the use of PACS relating to the rejected
as for ®lm but the rejected images were sorted, images and the examination request forms. After
and on the rare occasions when a code was PACS was fully operational, no hard copy images
missing, coded by another author (GW) in were produced routinely. It was not possible to
consultation with departmental radiographers. delete from the system an image which was
In addition to the rejection codes indicated identi®ed on the workstation as unsatisfactory,
above, the following CR-speci®c codes were used: but it was transferred to the electronic reject
H CR technique/scatter/position of image on folder by the radiographer. The reject folder
cassette/irradiated area too small. A pattern contained a permanent work list which retained
recognizer for iris of exposure ®eld (PRIEF) is details of all patients and their examinations from
an algorithm used by the image plate reader to which images had been rejected. It did not
detect the irradiated area on the plate. Each provide details of how many images or which
PRIEF has speci®c exposure precautions views were rejected or the reason for their
relating to irradiated ®eld positioning, size rejection. However, the reject process included a
and shape. There are ®ve PRIEFs, one of compulsory ®eld in which the reason for rejecting
which is pre-selected by the radiographer for the image had to be given. The images were then
each individual examination. Data obtained annotated with the reason for rejection for the
outside the irradiated ®eld, for example, purpose of audit, using the codes listed above.
scattered radiation, can adversely affect the Images could only be viewed in the reject folder
data analysis and the resultant image. The for 8 days and, owing to logistic dif®culties, data
positioning of the irradiated ®eld within the on the reason for rejection were not collected on
imaging plate area and the limitation of scatter 36 (7%) rejected images, although the body areas
is therefore very important. were known.
Table 1. Comparison of reject rates when the calculations are based on the total number of examinations
examinations as the denominator, care should be copy compared with ®lm (6.6% for ®lm and 5.5%
taken in attempting to relate these results to local for hard copy CR), but no further reduction when
situations elsewhere. In an attempt to help readers PACS soft copy images were used (Table 2).
extrapolate the results to local settings, the PACS
database was interrogated to obtain annual mean
values of the number of images used for each Reject rates by body area
body area. Since these values and the examination The reject rates by body areas are shown in
protocols remained virtually unchanged for the 3 Table 3. Examinations of the chest accounted for
years after this study, the values provided for the majority of rejects during all periods. This was
1995 were applied to all three phases of the study expected since this is the body area most
to provide a broad estimate of the reject rates frequently examined. When ®lm was used, the
based on the total number of images used. body area with the highest reject rate over all
examinations was the thoracic spine and this rate
dropped noticeably in the CR and PACS periods.
Results For examinations of the hip and the skull, the
Rejects expressed as rates across all reject rates in the CR and PACS periods were
examinations undertaken higher than in the ®lm period, and there was an
increase in rejects owing to errors in patient
There was a decrease in reject rates across all positioning and radiographic technique (Tables 4±
examinations from 9.9% when ®lm was used to 6). For all other body areas the reject rates were
8.1% when hard copy CR was used and 7.3% very similar for all three periods.
when PACS was used (Table 1). The reductions in
the reject rates per examination were statistically
signi®cant for both CR (p,0.01) and PACS Reject rates by reason for rejection
(p,0.01) compared with ®lm but there was no The reasons for rejection of images are shown
statistically signi®cant reduction when comparing in Table 7. Positioning and radiographic techni-
PACS with CR. que errors were the major reasons for rejects
throughout all periods and contributed to a
greater proportion of all rejects when CR and
Rejects expressed as rates across all images
PACS were used than when ®lm was used. The
undertaken
rejects which were expected to be completely
The reject rates across all images undertaken eliminated when ®lm was no longer used were
followed a similar trend. There was a statistically rejects associated with ®lm processing and fogging
signi®cant decrease in the reject rates for CR hard of ®lms, which accounted for 3.4% of all rejects.
Table 2. Comparison of reject rates when the calculations are based on the total number of images used
Chest 2148 161 (7.5) 2413 105 (4.4) 3653 235 (8.9)
Abdomen 231 20 (8.7) 307 26 (8.5) 386 40 (10.4)
Skull 132 34 (25.8) 146 53 (36.3) 203 47 (23.2)
Cervical spine 126 30 (23.8) 134 32 (23.9) 209 41 (19.6)
Thoracic spine 57 29 (50.9) 43 3 (7.0) 79 9 (11.4)
Lumbar spine 161 26 (16.1) 223 22 (9.9) 284 38 (13.4)
Pelvis 122 17 (13.9) 178 23 (12.9) 219 10 (4.6)
Hip 71 8 (11.3) 78 29 (37.2) 137 29 (21.2)
Upper limbs & 403 33 (8.2) 485 45 (9.3) 677 45 (6.7)
shoulder girdle
Lower limbs 453 27 (6.0) 495 27 (5.5) 706 31 (4.4)
excluding hips
Total 3904 385 4502 365 6617 525a
a
This total includes 36 rejects which did not involve irradiation of the patient.
These rejects were eliminated but new CR-speci®c and 5.5% for CR and PACS) falling within this
reasons for rejection of images emerged which range.
were responsible for 6.3% of all CR rejects and The reject rates reported here are from the same
7.4% of all PACS rejects. hospital and calculated by the same method and it
is therefore the change in reject rate rather than
the actual value of the reject rate which is of
central importance. This shows that the reject rate
Discussion
was reduced signi®cantly when the CR system was
The reject rates over all examinations under- introduced in the hospital, but that there was no
taken for ®lm (9.9%), CR (8.1%) and PACS further signi®cant reduction when the PACS
(7.3%) can be compared with a recent study from system was used. If the radiation dose for each
Nottingham [15] which reports its ®lm reject rates image is unchanged, and the same types of images
over all examinations as 8%. The rates calculated are used for the examination of each body area,
across all images undertaken can be directly there is therefore the potential to reduce the
compared with other published results which are patient population dose by moving from ®lm
based on reject rates calculated as a percentage of imaging to phosphor plate imaging [16].
all images taken. The published reject rates [2±10] These results must be interpreted with caution
vary between departments and range from 2% to as the process of rejection is subjective and the
13%, the rates found in this study (6.6% for ®lm changes following the introduction of CR and
A, patient position, radiographic technique; B, patient movement; C, incorrect exposure; D, X-ray equipment fault; E, processing,
fogging; F, not required; G, miscellaneous.
Table 5. Reason for rejecting image (number (%) of rejects for body area): CR
Chest 88 (83.8) 0 (0) 2 (1.9) 0 (0) 2 (1.9) 2 (1.9) 4 (3.8) 2 (1.9) 0 (0) 5 (4.8) 105
Abdomen 21 (80.8) 1 (3.8) 0 (0) 0 (0) 3 (11.5) 0 (0) 0 (0) 0 (0) 0 (0) 1 (3.8) 26
Skull 48 (90.1) 0 (0) 0 (0) 0 (0) 0 (0) 1 (1.9) 0 (0) 0 (0) 0 (0) 4 (7.5) 53
Cervical spine 23 (71.9) 0 (0) 1 (3.1) 0 (0) 1 (3.1) 0 (0) 0 (0) 0 (0) 0 (0) 7 (21.9) 32
Thoracic spine 2 (66.7) 0 (0) 1 (33.3) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 3
Lumbar spine 17 (77.3) 0 (0) 2 (9.0) 0 (0) 0 (0) 1 (4.5) 0 (0) 0 (0) 0 (0) 2 (9.0) 22
Pelvis 19 (82.6) 0 (0) 0 (0) 0 (0) 1 (4.3) 0 (0) 0 (0) 0 (0) 0 (0) 3 (13.0) 23
Hip 27 (93.1) 0 (0) 0 (0) 0 (0) 1 (3.5) 0 (0) 0 (0) 0 (0) 0 (0) 1 (3.5) 29
Upper limbs & 34 (75.5) 2 (4.4) 0 (0) 0 (0) 4 (8.9 1 (2.2) 0 (0) 0 (0) 0 (0) 4 (8.9) 45
shoulder girdle
Lower limbs 25 (92.6) 0 (0) 0 (0) 0 (0) 1 (3.7) 1 (3.7) 0 (0) 0 (0) 0 (0) 0 (0) 27
excluding hips
Total5365
A, patient position, radiographic technique; B, patient movement; C, incorrect exposure; D, X-ray equipment fault; H, CR technique;
J, incorrect organ code; K, Digiscan fault; L, hard copy processor fault; F, not required; G, miscellaneous.
Code E not applicable.
PACS could be a result of changes in the that manipulation of the soft copy image would
threshold of acceptance of images by the current allow this area to be visualized through the mass
staff or indeed by changes in staff over the period of the shoulders. Overall, the reject rates for the
of the study. cervical spine remained a reasonably constant
As was anticipated, the introduction of CR proportion of all rejects throughout all three
reduced the high percentage of thoracic spine studies (®lm 23.8%, CR 23.9% and PACS 19.6%).
rejects which were caused by incorrect exposure However, whereas the reject rates caused by
factors. It was expected that the use of PACS incorrect exposure factors decreased, the reject
workstations might reduce some of the rejects of rates resulting from incorrect positioning and
cervical spine images resulting from errors in radiographic technique increased. The expected
positioning and radiographic technique. This was overall reduction in cervical spine rejects was
because many of the rejects were caused by therefore not achieved.
inadequate demonstration of the cervicothoracic It was expected that when PACS was fully
junction in the lateral view and it was anticipated operational, the rejects caused by wet processing
Table 6. Reason for rejecting image (number (%) of rejects for body area): PACS
Chest 178 (75.7) 4 (1.7) 1 (0.4) 5 (2.1) 7 (3.0) 2 (0.8) 12 (5.6) 5 (2.1) 21 235
Abdomen 29 (72.5) 1 (2.5) 1 (2.5) 4 (10.0) 0 (0) 1 (2.5) 4 (10.0) 0 (0) 0 40
Skull 41 (87.2) 1 (2.1) 1 (2.1) 0 (0) 0 (0) 0 (0) 0 (0) 2 (4.2) 2 47
Cervical spine 28 (68.3) 0 (0) 1 (2.4) 0 (0) 4 (9.8) 0 (0) 1 (2.4) 1 (2.4) 5 40
Thoracic spine 6 (66.7) 0 (0) 1 (11.1) 0 (0) 0 (0) 0 (0) 2 (22.2) 0 (0) 0 9
Lumbar spine 25 (65.8) 0 (0) 0 (0) 0 (0) 3 (7.9) 1 (2.6) 8 (21.1) 0 (0) 1 38
Pelvis 7 (80.0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 1 (12.5) 0 (0) 2 10
Hip 20 (69.0) 0 (0) 3 (10.3) 1 (3.5) 4 (13.8) 0 (0) 0 (0) 0 (0) 2 30
Upper limbs & 22 (48.9) 0 (0) 10 (23.2) 0 (0) 8 (17.8) 1 (2.2) 2 (4.4) 0 (0) 2 45
shoulder girdle
Lower limbs 23 (76.7) 0 (0) 0 (0) 1 (3.2) 1 (3.2) 1 (3.2) 1 (3.2) 0 (0) 4 31
excluding hips
Total5525
a
These rejects do not involve irradiation of the patient.
b
Images could not be fetched and viewed, so reason for rejection unknown.
A, patient position, radiographic technique; B, patient movement; C, incorrect exposure; D, X-ray equipment fault; H, CR technique;
K, Digiscan fault; F, not required; G, miscellaneous.
Codes E, J and L not applicable.
14. Bryan S, Weatherburn G, Buxton M, Watkins J, 16. Weatherburn GC, Bryan S. The effect of a picture
Keen J, Muris N. Evaluation of a hospital picture archiving and communication system (PACS) on
archiving and communication system. Journal of patient radiation doses for examination of the
Health Services Research and Policy (forthcoming). lateral lumbar spine. Br J Radiol 1999;72:534±545.
15. Dunn MA, Rogers AT. X-Ray ®lm reject analysis
as a quality indicator. Radiography 1998;4:29±31.
Book review
Modelling in Clinical Radiobiology Based on an Practical Requirements of Clinical Radiotherapy
International Workshop in WuÈrzburg. Ed. by (R G Dale); this includes a discussion on where
K Baier, D Baltas, pp. 345, 1997 (Albert- the linear-quadratic (LQ) model should be applied
Ludwigs-University Freiburg Publishing, in the clinic, factors involved in moving from low
Germany) dose rate (LDR) to high dose rate (HDR)
ISBN 3980413901 brachytherapy, combining treatment modalities
Radiobiology continues to be an area of much and hypofractionation for palliative treatment.
activity and interest in radiotherapy treatment. It Pulsed Brachytherapy Radiobiological Aspects
combines the efforts of clinicians, physicists and (J F Fowler); dependence on dose-per-pulse and
biologists in deriving accurate parameters for use frequency of pulse delivery and the reliability of
in modelling the biological effects of radiation on radiobiological parameters.
both individual patients and on populations in Determination of the Relative Seriality of a
order to fully optimize radiotherapy treatment. As Tissue from its Response to Non-uniform Dose
more reliable patient data is collected, greater Delivery (A A Ê gren, P KaÈllmann and A Brahme);
accuracy in predicting responses will be achiev- a mathematical treatment of tissues behaving
able and increasingly sophisticated models will either as a serial (e.g. lung, liver) or parallel (e.g.
become accessible. spinal cord) structure in terms of NTCP.
This monograph reports on the papers pre- Volume Effect and Normal Tissue Responses
sented at an International Workshop in (D Baltas); discussion and comparison of various
WuÈrzburg, Germany in 1993, updated in 1995. dose-response models for NTCP.
A total of 31 papers are presented in the form of Extraction of Average Values for Tumour
chapters, covering a broad range of topics within Parameters from Clinical Local Tumour Control
clinical radiobiology from both theoretical and Probability Data (J van de Geijn, H Xie, J Chen);
clinical perspectives. Approximately two-thirds of application of a 6-parameter TCP model using
the contributions are from the host country, clinical data.
additional contributions being from researchers in Radiobiologically Based Treatment Plan
the USA, UK, Sweden, Canada, France and Optimisation (A Brahme); current and emerging
Greece. technology for diagnosis and treatment, inverse
The monograph begins with a review by J F planning, molecular biology and methods for
Fowler on the history of modelling radiobiology, delivering non-uniform dose distributions.
its successes and failures. The papers that follow As is common with conference proceedings, a
consider the latest in current thinking and future small number of misprints or errors have
developments in a broad range of areas. remained in some contributions. However, on
Contributions from physicists, clinicians and the positive side, a useful index is provided and
biologists are presented side-by-side. In this way each chapter is supported by a large number of
the monograph provides the reader with a balance references.
of new theoretical models and results of clinical In summary, these proceedings provide a
studies. This combination of disciplines also valuable source of reference for anyone beginning
serves to remind the reader that any applications or currently undertaking research in radiobiology,
of the models considered must be tempered by whether from a medical or scienti®c perspective,
clinical judgement. covering as they do a full review and the status of
Amongst the many interesting chapters are the research in clinical radiobiology up to 1995.
following notable contributions: Modelling the C R BAKER