1RRPPS, Birmingham Medical Physics Service, Birmingham, West Midlands, ... Over 1500 patients undergoing chest radiography in the West Midlands have ...
1995, The British Journal of Radiology, 68, 1221-1229
Optimization of radiographic technique for chest radiography * 1 H M WARREN-FORWARD, BSc and t 2 J S MILLAR, FRCR 1 RRPPS, Birmingham Medical Physics Service, Birmingham, West Midlands, and department of Radiology, Queen Elizabeth Medical Centre, Birmingham, West Midlands, UK
Abstract
Over 1500 patients undergoing chest radiography in the West Midlands have been monitored for entrance surface doses using lithium borate thermoluminescent dosemeters. In total 63 X-ray tubes were monitored from 30 hospital departments. The mean patient entrance surface dose is 0.15 mGy, and the 75th percentile entrance surface dose is 0.18 mGy. A reference level of 0.18 mGy has been recommended for PA chest radiography in the West Midlands. Image quality has been assessed on patient radiographs. Departmental radiologists were responsible for assessing radiographs taken within their hospital. Independent analysis was performed by a control radiologist. Film-screen processor sensitivity has been assessed on 48 film-screen processor combinations. Significant differences were observed between the nominally quoted sensitivities and the measured sensitivities. Only 26% of systems produced measured sensitivities within 10% of the nominal values. A four variable regression model, explaining 78% of the variance, provided the best description for the variation in patient dose. These variables were actual sensitivity, applied potential, generator waveform and radiographic quality. Four recommendations have been made to lower patient doses; these are: (1) an increase in applied potential to a minimum of 90kVp; (2) a film-screen sensitivity of 400; (3) optimization of processor performance and (4) regular radiological audits to reduce repeat rates to a level of 5%. If all of these recommendations are followed, an estimated overall entrance surface dose saving of 53% would result. Changing the applied potential alone will see the variation in the mean entrance surface dose from non-gridded systems reduce from a factor of 4 to a factor of 2. The effective use of ionizing radiation in diagnostic radiology involves the interplay of three factors: image quality, radiographic technique and patient dose. A good radiographic technique should produce an imag^ containing all the information essential to a diagnosis and should result in the minimum possible dose to the patient. The quality of the resulting radiograph will also depend on the condition of the patient and the degree of co-operation given by the patient. The amount of information that can be obtained by the attenuation of the X-rays within the body depends on the composition of the anatomical or pathological structures, the radiation quality and the transfer into an image suitable for interpretation. A number of national and regional dose surveys [1-4], have revealed large patient dose variations, for patients undergoing the same type of diagnostic X-ray examination. A previous dose survey carried out by the RRPPS [5] showed that a similar situation occurs in
the West Midlands Region. These studies imply that all exposures are not being kept as low as possible. The publication of such findings has prompted a recognition of the importance of developing appropriate strategies for reducing patient doses in diagnostic radiology. Many of the measured doses in the upper end of the dose range may represent radiation levels that are unnecessarily high. It may also be true that many doses in the lower end may result in poor image quality. It is therefore important to assess levels of image quality before recommendations on dose reduction are issued. A study group of the Radiation Protection Programme of the Commission of the European Communities has produced guidelines [6] for radiographic factors and techniques that might be employed to achieve radiographic images of good quality and acceptable dose. A pilot study has shown low correlation between patient entrance surface dose (ESD) and nominal filmscreen sensitivity [7]. This is thought to be due to variations between actual and nominalfilm-screensensitivitReceived 7 March 1995 and in revised form 18 May 1995, ies. The film-screen sensitivity classification (nominal accepted 11 July 1995. sensitivity) used by manufacturers is usually quoted at *Current address: Department of Radiology, Robert an applied potential of 80kVp and is known to be Jones & Agnes Hunt Orthopaedic Hospital, Oswestry, dependent on beam quality. The actual sensitivity of Shropshire, UK. systems in use by an X-ray department depends not only fCurrent address: Wessex Neurosurgery Unit, on the choice offilmsand screens but also on the proSouthampton General Hospital, Southampton, UK. cessing techniques used. Evidence is given in this current Vol. 68, No. 815
1221
H M Warren-Forward and J S Millar
paper to demonstrate the importance of using filmscreen processor sensitivity to investigate the relationship with patient dose.
borate was selected since, when doped with magnesium, it has a maximum variation of + 5 % in the photon energy range 10-1000 keV; this value was confirmed using a method traceable to the primary standard.
Materials and methods Participants An open invitation was given to all X-ray departments in the West Midlands to participate in a patient dosimetry study. All departments were sent a questionnaire, on which they had to nominate up to four X-ray tubes in their department. There was a 9 1 % response rate, in which only two hospitals said it was inappropriate to be involved in the study. It has been stated that nonresponse rates higher than 15-20% may throw doubts on any conclusion drawn from a particular study [ 8 ] , so the representativeness of X-ray tubes in this study should not affect the results. The hospitals were then chosen at random until 30 X-ray tubes had been selected. It was thought that this number distributed throughout the region would provide the full range of techniques employed, including those tubes with average performance as well as tubes showing extreme values with respect to the radiation dose, filmscreen sensitivity and performance in image quality. Based on information found in the literature and on the results from the analysis of a previous survey carried out at the RRPPS [ 5 ] , an initial protocol for dose measurement was planned. 20 tubes were chosen for the pilot study to test the data collection procedures in order to discover snags, ambiguities or errors in design of the data sheets, to provide preliminary estimates on the distribution of some variables and to indicate a reasonable time scale needed to collect data on 30 patients. The results of the pilot study have been reported elsewhere [ 7 ] .
Patient dosimetry Entrance surface dose measurements were made by attaching the TLDs to the patient's skin on the central axis of the X-ray beam. The estimated height and weight were recorded for each patient, together with an indication of the patient's posteroanterior (PA) thickness at the centre of the X-ray field, thereby allowing the size and shape of the patient to be taken into account. The questionnaire accompanying the patient exposure recorded the exposure factors in terms of applied potential, current-time product, film size, focus-to-film distance (FFD) and the film processor used. The total number of films taken per patient was recorded, each duplication was regarded as a "repeat" film. That is, if the patient undergoes two PA radiographs, the second is seen as a repeated film, although a PA penetrated film following a PA is not classed as a repeat radiograph. The reason for the repeat was also given. This information was then used to assess the repeat rate for chest radiography between different hospitals and different X-ray rooms in the same hospital. The radiographic techniques seen in the study could broadly be divided into two techniques: a low kVp technique, using applied potentials in the range 55-98 kVp with no antiscatter grid; and a high kVp technique using 108-135 kVp with antiscatter grid. Out of the 63 units investigated only four employed the high kVp technique.
Radiographic equipment Quality assurance surveys were carried out on the majority of X-ray tubes, in accordance with Part I of HPA document TGR 32 [ 9 ] . The results provided information on the radiation output, total filtration, target angle, generator waveform and the type of film-screen combination used. Another tube (Philips Super Rotalix SRT 2550) was added to the programme for the measurement of film-screen processor sensitivity. The tube had a total filtration of 2.55 mm aluminium, the voltages were within + 2 kV; the variation in radiation output did not exceed + 3 % and the tube output was found to vary linearly with the timer setting to within + 3 % . Lithium borate thermoluminescent dosemeters (TLDs) were used for the assessment of ESD. The overall random uncertainty at the 95% confidence level for a dose of 0.1 mGy was calculated to be 16%. The accuracy of the measurement has been estimated to be within 10% at the 95% confidence level. These values have been calculated using the method outlined in the National Protocol for Patient Dose Measurements in Diagnostic Radiology [10]. The overall uncertainty in the measurement at the dose level of 0.1 mGy was 18%, which reduced to 13% at doses around 0.3 mGy. Lithium 1222
Film-screen sensitivity One loaded cassette per processor used in the study was collected from each centre. Each cassette was then exposed with the same X-ray tube, (Philips Super Rotalix SRT 2550). A 28 mm thick aluminium absorber was placed at the light beam diaphragm to simulate a patient and to produce realistic doses to the cassette front. The air kerma was measured in the plane of the film-screen cassette, with a MDH 2025 series monitor and 20X5-3 ion chamber. This was known to have a flat energy response to within ± 5 % over the energies covered. Each film was used to assess the variation in sensitivity at four applied potentials: 60 kVp, 80 kVp, 90 kVp and 110 kVp were chosen to cover the range used for chest radiography in the West Midlands. The current was kept constant at 25 mA and the timer was adjusted so that for a nominal film-screen sensitivity of 400, each exposure produced an air kerma of approximately 1.0 uGy at the cassette front. Higher values of air kerma to the cassette front were used for film-screen sensitivities of lower nominal values. One lead mask of thickness 2 mm was used to cover three quarters of the cassette and another mask (which was moved 20 mm between exposures) was used to produce six steps at the given applied potential in the remaining quarter. After exposure the film was processed in its associated departmental processor. After processing, the density of the steps and the base plus fog was measured using a The British Journal of Radiology, November 1995
Radiographic technique for chest radiography
Parry transmission densitometer, model DTI405, calibrated with a standard strip. A characteristic curve of dose (mGy) with optical density was plotted, and the air kerma needed to produce a net density of 1.0 (1.0 +base plus fog) was determined. The sensitivity was calculated as 1 mGy multiplied by the reciprocal of this air kerma. The consistency of the method was assessed by measuring the sensitivity of the standard film-screen combination over the four applied potentials for 10 films measured over a period of 8 weeks using a Fuji FPM 2100 processor at temperature 32.5 °C. Image quality For PA chest radiography, the Commission of the European Communities (CEC) advocates a radiographic voltage in the range 100-150 kVp and the use of an antiscatter grid with a lateral AEC chamber selected [ 6 ] . Previous studies [11] have shown that this method of high voltage radiography will produce higher ESDs. It has been said that the use of an antiscatter grid enables the radiologist to obtain an improved image quality compared with systems without a grid [ 6 ] , however, the introduction of a grid can increase radiation dose by a factor of 2 to 3. Since the aim of this present study is the optimization of chest radiography, assessment of image quality is needed at the lower levels of radiographic voltages used in the West Midlands. It cannot be assumed that radiographic quality is lower when no antiscatter grid is used, where the lower doses are produced. The CEC criterion forms the basis of the image quality assessment in the present study. A number of local radiologists were invited to comment on the important characteristic features of "normal" posteroanterior and lateral chest radiographs. Each radiologist then checked the CEC quality criteria and commented on the relevance of each statement. In view of their comments several minor changes and additions were made. A note of interest was that most radiologists agreed with the relevance of all parts of the CEC criteria even though completely differing techniques are employed to produce the resulting radiograph. The regional questionnaire was broken down into four areas: the examination, patient details, patient positioning and radiographic image. A section was added about the purpose of the examination and whether any previous radiographs had been consulted prior to the X-ray examination. Concerning patient positioning, several features were added for completeness. Assessment of patient rotation was included, since an abnormal projection of the heart or mediastinum may simulate pathology. Each anatomical feature was identified singularly, since there might be some differences in responses if several features appeared in the same statement; a yes answer might indicate that the viewer sees all structures, or it might indicate that the viewer sees only one of them. For the lateral projection assessment of patient rotation was included by asking about the superimposition of the posterior ribs. The only statement included by the CEC on the radiographic image was on the sharp reproduction of the posterior border of the heart. In the present questionnaire,
Vol. 68, No. 815
statements were added about the reproduction of the trachea, mediastinum, aorta, diaphragm, sternum, thoracic spine and posterior costo-phrenic angles. Using the questionnaire, a clinical assessment of each patient radiograph was made at each hospital by the department radiologists. The viewer was asked to tick appropriate statements yes or no. If any features were obscured due to abnormal pathology they were marked with an asterisk with the results of these films being removed from any further analysis. Each centre was asked to assess the radiation dose for up to 30 patients, and hence image quality for 30 radiographs. From these, 10 radiographs were randomly selected and copied, with the originals being sent to the RRPPS. In order to provide an independent analysis, clinical assessment was made on these 10 radiographs by a control radiologist, using the questionnaires. The summation of all the statements in the questionnaire leads to the assessment of radiographic quality of the image, and represents features that are dependent on the radiographic technique used for the examination, for example the radiographic voltage. When assessing some relationships with image quality the statements in the category's radiographer positioning and patient respiration are not included, since applied potential and dose are independent of these variables. Since variation in the responses was expected between the control radiologist and the individual departmental radiologists, intraobserver variation of the control radiologist was also assessed. One way to check for intraobserver variation was to have the control radiologist assess each radiograph twice; indeed questionnaire scores of films scored by the control within his department were in excellent agreement with the scores that he gave for the same films at the RRPPS. This method was considered too time-consuming, so another method of assessment was used. In a daily routine, when the radiologist assesses a radiograph to establish a diagnosis, this is established by a visual impression and not by a thorough analysis of all anatomical features. In keeping with this approach all of the 10 films sent to the RRPPS from each X-ray tube were reviewed by the control radiologist. All films were reviewed in a single sitting, where they were given a score of 1 to 6. A score of 6 was allocated for a perfect film, where the control radiologist considered the optical density to be optimum, patient positioning to be excellent and where all the important features were visible. A high score here should correspond with a high score in the questionnaire since, if consistent, the optical density, patient positioning and good contrast would have scored highly in the questionnaire. The radiograph was given 5 if it was very good, 4 for satisfactory, 3 for poor, 2 for a repeat if the patient was still in the department and 1 where it was thought appropriate to recall the patient back to the department to repeat the film. Results Patient dosimetry Over 1500 patients have been monitored from 30 hospitals incorporating a total of 63 X-ray tubes. All of the 1223
H M Warren-Forward and J S Millar
evaluated ESDs were associated with radiographs that were reported upon by radiologists, i.e. no rejected films. All values of ESD are expressed in terms of absorbed dose to ICRU muscle to allow comparative analysis with the results of the NRPB. The majority of examinations (87%) were PA projections and the analysis of data concentrates on this high frequency examination. The histogram of ESDs (Figure la) is represented by a wide positively skewed distribution, with the mean ESD being 0.15 mGy and the 75th percentile being 0.18 mGy. The range factor of ESDs was found to be 46 (0.02-0.92 mGy). The average film reject rate was found to be 11% (0%-36%); with 80% of all repeats being due to incorrect exposure settings and 7% due to positioning, patient movement or processor problems. The NRPB has indicated that the patient size is unlikely to account for maximum-minimum ESD ratios of more than 10. Most hospitals (65%) were able to produce range factors less than this, while others (17%) exceed 20. The maximum ratio was observed to be 34. 250 200
150
The results have been reduced to the mean values of each tube, in this way we are assessing the average exposure factors used for each X-ray tube when examining the average patient. Significant differences are apparent between the mean ESDs obtained between the tubes. The range of mean ESDs, 0.07-0.59 mGy (Figure lb), indicates that some hospitals are using lower dose techniques than others. A local reference level for the West Midlands has been set at the 75th percentile value (0.18 mGy); this is in accordance with the setting of the national reference level (0.30 mGy). The four X-ray units using the high kVp technique produced mean ESDs of 0.14 mGy, 0.14 mGy, 0.18 mGy and 0.59 mGy. Three of these indicate that it is possible to use this technique and obtain ESDs lower than the national reference level. Factors contributing to the ESD variations include film-screen sensitivity, filtration, generator waveform, applied potential and focus-to-film distance. Table I summarises the mean radiographic techniques used in the study, comparison has also been made with techniques advocated by the CEC. It is disappointing to record that only 23% of departments are using film-screen sensitivity of 400, after its much publicised effect in the reduction of patient doses. Even though the cost of new screens may contribute to this low level, increases in overall sensitivity may be obtained by changes in the film used with particular screens, without detriment to image quality.
2 100
IIIIII
50
o
—
.
— o
o
o
Mean Entrance Surface Dose (mGy) (a)
Film-screen sensitivity The results of the consistency measurements showed that at an applied potential of 80 kVp, the maximum variation from the mean was 7%. 48 systems were assessed for film-screen processor sensitivity. The nominal speeds quoted by the manufacturer are usually quoted at an applied potential of 80 kVp. A summary of the results for each X-ray system at 80 kVp are presented Table I. Range of mean radiographic techniques used at each department Variable
Regional Reference Level = 0.18 mGy
X-Ray Tubes (b) Figure 1. Distribution of entrance surface doses for PA chest radiography, (a) shows the distribution of entrance surface doses for all patients, while (b) represents the distribution of mean entrance surface dose for the individual X-ray tubes in the survey.
1224
Minimum Maximum CEC mean mean value value
0.07 Entrance surface dose (mGy) Applied potential 55 (kVp) Current-time 1.46 product (mAs) Focus-film-distance 150 (cm) Filtration 1.8 (mm Al) Film-screen 185 sensitivity (mGy" 1 )
0.26 135 14.10 305 4.3 400
