We propose a hybrid deep learning-based method, which includes a cycle consistent generative adversarial network(CycleGAN) and deep attention fully convolution network implemented by a U-Net (DAUnet), to perform volumetricmulti-organ segmentation for pelvic computed tomography (CT). The proposed method first utilized CycleGAN togenerate synthetic MRI (sMRI) to provide superior soft tissue contrast. Then, the proposed method fed the sMRI into theDAUnet to obtain the volumetric segmentation of bladder, prostate and rectum, simultaneously, via a multi-channeloutput. The deep attention strategy was introduced to retrieve the most relevant features to identify organ boundaries.Deep supervision was incorporated into the DAUnet to enhance the features’ discriminative ability. Segmented contoursof a patient were obtained by feeding the CT image into the trained CycleGAN to generate sMRI, which was then fed tothe trained DAUnet to generate the organ contours. A retrospective studied was performed with data sets from 45 patientswith prostate cancer. The Dice similarity coefficient and mean surface distance indices for bladder, prostate, and rectumcontours were 0.94, 0.47 mm; 0.86, 0.78 mm; and 0.89, 0.85 mm, respectively. The proposed network provides accurateand consistent prostate, bladder and rectum segmentation without the need of additional MRIs. With further evaluationand clinical implementation, this method has the potential to facilitate routine prostate-cancer radiotherapy treatmentplanning.
展开▼