[P040] Variability of mammographic exam doses as a result of positioning and technique

Purpose Automatic collection of dosimetric data from PACS (Picture Archiving and Communication System) allows comparisons of exam doses and establishment of diagnostic reference levels (DRLs) based on large datasets. In an oncology institute, dosimetric data pertaining to mammography will include breasts previously submitted to surgery and radiotherapy, in addition to data from untreated breasts. It seems timely to compare dosimetric data for treated and untreated breasts, to determine if the two should be considered separately. The aim of this work is to establish the baseline of normal variation attributable to positioning and technique in mammography, for comparison with the variation between treated and untreated breasts. Methods 10 patients were randomly selected among those who had multiple mammographic exams archived in PACS and no previous history of surgery or radiotherapy. These patients were submitted to between 4 and 9 exams, performed in the same mammographic unit (GE Senograph DS), in standard automatic exposure mode, over a period of 8 years. The anonymized images were downloaded from PACS, and a script was written in Python to extract the necessary information: compression force, compressed breast thickness, entrance surface air kerma (ESAK), mean glandular dose (MGD), area and maximum perpendicular width of the imaged region. The area and maximum width were obtained using thresholding, while the other parameters were extracted from the DICOM header. Results The most variable parameter proved to be the compression force, which for the same patient (same view, same side) had typical variability between exams around 19–23% of the mean value, reaching more than 30% in some patients. The maximum width and the area of the imaged region had the least variability (3% and 4% respectively). Visual comparison of the contours for different years confirms the good reproducibility of positioning, despite small translational variations. Variability of MGD and ESAK was 6–7% and 9–10% respectively, for repeated views of the same untreated breast. Conclusions Variability of exam doses with positioning and technique for the same mammographic unit was found to be small, despite large variations in the compression force. This is probably related to the remarkably consistent positioning.