Room: Room 202
Purpose: MRI has shown promise in identifying prostate tumors with high sensitivity and specificity for the detection of prostate cancer. Accurate prostate segmentation plays a key role various tasks: to accurately localize prostate boundaries for biopsy needle placement and radiotherapy, to initialize multi-modal registration algorithms or to obtain the region of interest for computer-aided detection of prostate cancer. However, manual segmentation during biopsy or radiation therapy can be time-consuming and subject to inter- and intra-observer variation. This studyâ€™s purpose is to develop an automated method to address this technical challenge.
Methods: A 3D supervision strategy was incorporated into the hidden layers of fully convolutional neural network (FCN) to deal with the optimization difficulties when training a deep network with limited training data. Through up-scaling low, middle and high-level feature volumes at each forwarding path of the hidden network using additional deconvolutional layers, the softmax function was employed on these full-sized feature volumes to obtain dense predictions. Then, the negative log-likelihood loss and the batch-based Dice loss of the dense predictions were introduced into the whole loss function for deep supervision training. During the segmentation stage, the patches were extracted from newly acquired MR image as the input of the well-trained network and the well-trained network adaptively identified the prostate tissue.
Results: This segmentation technique was validated with a clinical study of 11 patients. The accuracy of our approach was assessed using the manual segmentation (gold standard). The mean Dice Overlap coefficient was 88.14Â±2.23% between our and manual segmentation, which indicates that the automatic segmentation method works well and could be used for 3D MRI-guided prostate intervention.
Conclusion: We have developed a novel, and clinically validated, MRI prostate segmentation approach based on the 3D deeply supervised FCN framework. This could be a useful tool for image-guided interventions in prostate cancer.
Not Applicable / None Entered.