Adult patient-specific CT organ dose estimations using automated segmentations and Monte Carlo simulations

Abstract

We aimed to determine feasibility in calculating patient-specific organ doses for abdominal computed tomography (CT) exams using an automated segmentation technique dedicated to abdominal organs combined with Monte Carlo simulation of a clinical CT scanner. We conducted the automated segmentation of five major abdominal organs (left and right kidneys, pancreas, spleen, and liver) for ten adult patients and calculated organ-specific doses for each patient. We observed significant variability (Coefficient of Variation up to 32%) in organ mass across the ten patients, which was up to two-fold greater or smaller than the reference organ mass for the ICRP reference adult male and female. Comparison of patient-specific organ dose per CTDIvol with those from the ICRP reference phantoms confirmed that reference phantom-based dose reporting programs cannot capture inter-patient dose variability, and dosimetric errors can go up to nearly 40%. We demonstrated an automated method for patient-specific organ dose calculations, which took about 45 min per patient. When the automatic segmentation method is extended to more organs and faster Monte Carlo calculation technique is employed, our method should be useful for patient-specific dose monitoring at the organ level and for epidemiological investigations of health risks in CT patients.