Multi-organ segmentation in clinical computed tomography for patient-specific image quality and dose metrology

The purpose of this study was to develop a robust and automated multi-organ segmentation model for clinical CT and implement the model as part of a patient-specific safety and quality monitoring system. 3D convolutional neural network (Unet) models were set up to segment 23 different organs and structures. For each organ, about 200 manually-labeled cases were used. The dataset was randomly shuffled, and divided with 60% used for training, 20% for validation, and the remaining 20% for testing. The model was deployed to automatically segment about 1000 clinical CT images as a demonstration of the utility of the method. Based on the segmentation masks, each case was made into a patient-specific computational phantom. The organ doses were then estimated using a validated Monte Carlo package. The segmented organ information was likewise used to assess contrast within each organ. The neural network segmentation model showed dice similarity coefficients (DSCs) above 0.85 for the majority of organs. Notably, the lungs and liver showed a DSC of 0.95 and 0.94, respectively. The segmentation results produced patient-specific dose and quality values across the tested 1000 patients with the histogram distributions. The results can help derive meaningful relationships between image quality, organ doses, and patient attributes.