ESTIMATION OF CONVERSION COEFFICIENTS FROM DOSE-AREA PRODUCT TO EFFECTIVE DOSE FOR BARIUM MEAL EXAMINATIONS FOR ADULT PATIENTS

The aim of the current study was to establish conversion coefficients (CCs) from dose-area product to effective dose for most common barium meal (BM) fluoroscopic examination. The study was based on data collection in two X-ray rooms in a major university hospital in St-Petersburg, Russia that allowed evaluating a structure of BM fluoroscopic examinations and developing a computed model of effective dose estimation using PCXMC 2.0 software. Results indicate that effective doses and the CCs were mainly influenced by the structure (contribution of different projections) and by the parameters (field size and energy characteristics of the X-ray beam) of the fluoroscopic examination. Resulting values of CCs estimated in the study were comparable with the published data for BM examinations.

Fluoroscopic examinations contribute significantly to the collective dose from medical exposure, both in Russia (7% in 2015) [1] and European countries (2-50%) [2]. The examinations of the upper gastrointestinal tract (UGIT) with barium contrast (barium meal, BM) are among the most common fluoroscopic examinations. These examinations are performed in a majority of hospitals both for adult and pediatric patients, corresponding to 38% contribution to the collective dose from fluoroscopic examinations in Russia [1]. Hence, it is important to justify and optimize fluoroscopic examinations. Besides that, according to the Russian Federal State law №3-FZ "On Radiation Safety of the Public" 1 , each patient should be informed about the dose and possible consequences (radiation detriment) from the medical exposure. That is fulfilled by using the effective dose (E, mSv). For the medical exposure of the patients, E is commonly calculated using a dedicated software (PCXMC 2.0, CALDoseX, EDEREX, etc) based on the measurable dose quantity: dose-area product (DAP, cGy×cm 2 ).
However, for the fluoroscopic examinations, the process of effective dose calculation is complicated due to necessity to simulate the moving X-ray irradiation field. Modelling of the irradiation for different anatomical regions in different projections can be influenced by variability of the conversion coefficients (CCs) 2 from dose-area product to effective dose within a single fluoroscopic examination. It complicates the Fluoroscopic examinations of the upper gastro-intestinal tract and, especially, barium meal examinations, are commonly performed in a majority of hospitals. These examinations are associated both with substantial individual patient doses and contribution to the collective dose from medical exposure. Effective dose estimation for this type of examinations is complicated due to: 1) the necessity to simulate the moving X-ray irradiation field; 2) differences in study structure for the individual patients; 3) subjectivity of the operators; and 4) differences in the X-ray equipment. The aim of the current study was to estimate conversion coefficients from dose-area product to effective dose for barium meal examinations for the over couch and under couch exposure conditions. The study was based on data collected in the X-ray unit of the surgical department of the St-Petersburg Mariinsky hospital. A model of patient exposure during barium meal examination was developed based on the collected data on fluoroscopy protocols and adult patient irradiation geometry. Conversion coefficients were calculated using PCXMC 2.0 software. Complete examinations were converted into a set of typical fluoroscopy phases and X-ray images, specified by the examined anatomical region and the projection of patient exposure. Conversion coefficients from dose-area product to effective dose were calculated for each phase of the examination and for the complete examination. The resulting values of the conversion coefficients are comparable with published data. Variations in the absolute values of the conversion coefficients can be explained by differences in clinical protocols, models for the estimation of the effective dose and parameters of barium meal examinations. The proposed approach for estimation of effective dose considers such important features of fluoroscopic examinations as: 1) non-uniform structure of examination, 2) significant movement of the X-ray tube within a single fluoroscopic phase, and 3) the variety of exposure geometries within complete examination.
estimation of a single CC for the fluoroscopic examination. A common practice is to simplify the fluoroscopy assuming that the patient was irradiated only in one projection [3], or in several projections, but only for a single anatomic region [4]. Additionally, only a limited set of CCs is currently available for certain exposure conditions [4,5]. Hence, using the existing CCs may lead to an incorrect estimation of the effective dose.
In Russian practice, CCs from DAP to E for BM examinations are presented in Methodical Guidelines "Assessment of effective dose to the patients undergoing X-ray examinations" 3 . They are provided only for posterioranterior (PA) projection, corresponding to the under couch position of the X-ray tube. However, in present time, more than 60% of the fluoroscopy X-ray units in Russia are remotely controlled, with the standard over couch position of the X-ray tube (see Figure 1).
Al with anti-scatter grid: 110 lines/inch, R=13:1, F = 180 cm. Imaging was performed using default vendor protocols with automated brightness control (ABC) without the digital image intensification. The X-ray unit was equipped with the DRK-1 clinical dosimeter (NPP "DOZA", Russia), calibrated using a reference ionization chamber prior to the study. Patient positioning, examination structure, fluoroscopy frame rate and total time of irradiation were selected by the radiologist (a resident with 5 years of experience) individually for each patient based on his personal preferences, patient condition and preliminary diagnosis.
Each examination was converted into a set of typical fluoroscopy phases and X-ray images, specified by the examined anatomical region and the projection of patient exposure. The following data was collected for each fluoroscopy phase and for each X-ray image taken for each patient: patient position (standing, supine, prone, recumbent), projection, total fluoroscopy time (s), fluoroscopy frame rate (frames×s −1 ), field size (cm×cm), average tube voltage (kV), total DAP (cGy×cm 2 ). Data was collected manually by the authors during the examination using dedicated spreadsheets. All examinations were exported from the PACS and digitally recorded in DICOM format; these records were used for modelling of the exposure of the patients with the PCXMC 2.0 and for verification of the collected data.

Development of a model for patient exposure for BM examinations
Each fluoroscopic phase was described by a set of discrete irradiation fields, corresponding to locations of the relevant organs and tissues. If there was no significant movement of the X-ray tube and if only a single organ was irradiated (i.e. fluoroscopy of the stomach and duodenum with contrast), the phase consisted of a single irradiation field. On the other hand, if different organs were exposed and if the tube movement was significant (i.e. survey fluoroscopy of the UGIT without contrast), the phase consisted of several irradiation fields, each corresponding to relevant anatomic location. Exposure parameters for each irradiation field within a single phase were considered to be constant. The number of irradiation fields and their locations for the specific fluoroscopic phases were selected in cooperation with the radiologists based on their experience and digital records of the examinations.
A total of eight projections were selected to describe the exposure of a patient (see Table 2). It was assumed that all oblique projections laid in a transverse plane and formed a 45° angle with the AP/PA axis [5]. Hence, it is necessary to update the existing CCs, since they do not reflect the actual exposure conditions of the patients.
The aim of the current study was to estimate conversion coefficients from DAP to E for the BM examinations based on data collection in a typical general practice hospital in St-Petersburg, Russia. That required the evaluation of the structure of the selected fluoroscopic examinations, to collect the relevant parameters of the examinations, to develop a model of patient exposure and to calculate CCs using the PCXMC 2.0 software [7].

Data collection
Data for the effective dose estimation was collected in the X-ray room belonging to surgical department in St-Petersburg "Urban Mariinsky hospital" for a sample of patients undergone BM examinations (40 patients in 2016-2017). Data on age and anthropometric characteristics of the patients is presented in Table 1.
The BM examinations were performed on the digital KRT-Electron (JSC "NIPK "Electron", Russia) X-ray unit. The KRT-Electron is a remotely guided X-ray unit with the over-couch X-ray tube and a 12' CCD-matrix detector, commonly used for fluoroscopic examinations. The following settings were used: focal-image distance 115 cm; total filtration of 5 mm Радиационная гигиена Том 11 № 1, 2018 Examples of coordinates of the centers of corresponding irradiation fields for selected fluoroscopic phases are presented in Table 3. These coordinates correspond to an arbitrary point inside the phantom, through which the central axis of the x-ray beam is directed. The origin of the phantom's coordinate system is located at the center of the bottom of the phantom trunk section. The positive z-axis is directed upwards, the positive y-axis to the back of the phantom, and the positive x-axis to the left-hand side of the phantom [7].
An example of the set of fields for a survey fluoroscopy of the UGIT is presented in Figure 2. For single X-ray images, it was assumed that the coordinates matched the coordinates of the last irradiation field for the corresponding fluoroscopic phase.

Calculation of conversion coefficients
CCs were calculated using standard adult (PCXMC 2.0 default, 178.6 cm height and 73.2 kg body mass) parameters both for the over couch and under couch irradiation geometries (see Fig. 1). For the latter, the study structure was kept the same, but the irradiation angles were inverted by 180°.   Table 3 for the respective field coordinates. The images correspond to a 28×28 cm field size and a 115 cm FID To estimate the CCs for the complete BM fluoroscopic examinations, the following method was used: • Calculation of the CCs for each fluoroscopic phase and X-ray image for each projection for each patient; • Estimation of DAP contribution of each projection into the total DAP for the complete examination for the whole patient sample; • Estimation of the weighted mean CC for the complete fluoroscopic examination using Eq. 1: where K 60 , 103 are the CC for the complete fluoroscopic examination estimated using tissue weighting coefficients from the ICRP Publications 60 and 103 [8], respectively; DAP projection is the DAP (cGy×cm 2 ) for fluoroscopic phases and X-ray images for the selected projection for the whole patient sample; DAP total is the total DAP (cGy×cm 2 ) for all fluoroscopic phases and X-ray images for the whole patient sample for the selected type of fluoroscopic examination; K 60 , 103 projection are the CC for single fluoroscopic phase or X-ray image, calculated using tissue weighting coefficients from the ICRP Publications 60 and 103 [8], respectively.
Statistical evaluation was performed using Statistica 10 software.
Differences were considered to be significant with p<0.05.

Results
Structure and main parameters of BM examinations are presented in Table 4. (1) where K 60,103 are the CC for the complete fluoroscopic examination estimated using tissue weighting coefficients from the ICRP Publications 60 and 103 [8], respectively; DAP projection is the DAP (cGy×cm 2 ) for fluoroscopic phases and X-ray images for the selected projection for the whole patient sample; DAP total is the total DAP (cGy×cm 2 ) for all fluoroscopic phases and X-ray images for the whole patient sample for the selected type of fluoroscopic examination; K 60,103 projection are the CC for single fluoroscopic phase or X-ray image, calculated using tissue weighting coefficients from the ICRP Publications 60 and 103 [8], respectively.
Statistical evaluation was performed using Statistica 10 software. Differences were considered to be significant with p<0.05.

Results
Structure and main parameters of BM examinations are presented in Table 4. Data on dose-area product for BM examinations is presented in Table 5. Data on the effective doses for the over couch and under couch irradiation geometries, estimated using tissue weighting coefficients from the ICRP Publications 60 and 103 is presented in Table 6.
Data on the contribution of different projections (see Table  2) into total DAP is presented in Table 7 for the whole patient sample.
The resulting values of the CCs for the complete BM examination for the under couch and over couch irradiation geometries, are presented in Table 8.

Discussion
The proposed approach for the estimation of the CCs considers important features of fluoroscopic examinations: non-uniform structure of examination, movement of the X-ray tube and the variety of exposure geometries. Segmentation of the fluoroscopic examination into a set of typical fluoroscopic phases allows evaluating the impact of the differences in CCs for individual phases on a resulting conversion coefficient for the complete examination. A similar approach was used in [9] for the barium swallow examinations.
The PCXMC 2.0software allows two approaches for setting the coordinates of the irradiation field: as a coordinate of the center of the relevant anatomic organ or as a coordinate of the corresponding point on the phantom surface. These two approaches had been compared prior to the study; the differences in the estimated organ and effective doses did not exceed 5-7%. Hence, the first approach of defining the irradiation field was used for the convenience of modelling.
Several approaches for describing the tube movement within a single fluoroscopic phase were evaluated, varying the number of irradiation fields per phase and their exact locations. The resulting sets of fields (see Table 3) were    selected as a compromise between the speed of calculation and a reproduction of real patient exposure. The major difference, compared to other available methods of effective dose estimation [4,5], is the inclusion of the multi-field phases of the survey fluoroscopy of the UGIT and fluoroscopy of the esophagus.
Patient data collection was designed to monitor the differences in fluoroscopic protocols due to the operator subjectivity in the same department. The distributions of the effective doses and conversion coefficients for individual patients for 2016 and 2017 patient samples were checked for normality (the Kolmogorov-Smirnov test) and then compared using the Mann-Whitney U-test. No significant differences were found between 2016 and 2017 samples (p<0.05). Comparison of the 2016 and 2017 distributions of the effective dose and conversion coefficients, using tissue weighting coefficients from the ICRP Publication 60 [8] for the over couch irradiation geometry is presented in Figure 4.
Comparison of the estimated CCs for the complete BM examination with the available literature data is presented in Table 9.
The results of the current study are comparable with other published CCs. The differences in the absolute values of the CCs can be explained by various factors. The most important is the difference in the clinical protocols between the countries and hospitals. Another factor is the difference between the methods used for effective dose estimation, mainly selection of specific anatomic regions and projections to be included into a model of BM examination. By definition, CCs depend on the patient irradiation geometry (anatomical region or organs of interest, projection, focal-image distance, irradiation field size) and the energy characteristics of the X-ray beam (tube voltage, total filtration). All of these factors are influenced by the operator subjectivity and the characteristics of the X-ray unit, requiring consideration for an accurate dose estimation in a specific X-ray room or medical facility. It can be seen from Table 9, that the existing CC for the BM examination in the PA projection matches the CC for the under couch X-ray tube position, estimated in the current study. However, the difference between the CCs for the over couch X-ray tube position is significant -26% (Mann-Whitney U-test, p<0.05). That derived CC would be included in the updated version of the methodical guidelines on effective dose estimation, allowing more accurate patient dose estimation.