Room: Room 202
Purpose: Recording dose from each CT scan is becoming a standard practice across healthcare providers. This risk management is best achieved using rapid, accurate, and patient-specific organ dose estimates. The purpose of this study was to design and implement a comprehensive organ dose estimation method applicable to clinical CT imaging of adult, pediatric, and pregnant patients.
Methods: This dose estimation framework consists of both patient-specific anatomy and scanner/protocol-specific exposure modeling. Patients are matched to human models from an library of XCAT phantoms using landmarks segmented from patient localizer images (DoseWatch, GE), gender, and age (gestational age for pregnant patients). The XCAT library consisted of 58 adult (18-78 y.o.; 52-117 kg; 23 F 35 M), 56 pediatric (2-18 y.o.; 12-88.4 kg; 31 F 25 M), 50 pregnant (3-38 weeks gestational ages; 56-117 kg), and 14 ICRP reference models. Organ doses were computed a priori for all XCAT phantoms for 10 body and 3 head-and-neck protocols using a Monte Carlo method and formatted as lookup tables. For each new patient, matched to a specific protocol and phantom, a convolution-based technique was applied to accommodate the applied tube current modulation. A round robin technique was applied to ascertain the bias and uncertainty in the organ dose estimates. The organ dose estimates were validated against full Monte Carlo estimation (referred to as true dose).
Results: True dose was in general within the confidence interval of the predicted dose. For most organs within the irradiation field, the predicted and true dose values showed average absolute error per CTDIvol below 15%. The method provided a near real-time method of patient-informed dose estimation.
Conclusion: This study involved the implementation of a comprehensive organ dose estimation framework. This framework demonstrates adequate robustness, speed, and accuracy for clinical implementation enabling patient-informed monitoring of organ doses in clinical CT operation.
Funding Support, Disclosures, and Conflict of Interest: E.S. received funding from GE and SIEMENS.
Monte Carlo, Anatomical Models, Radiation Dosimetry