Meeting Banner
Abstract #4712

Automated Localization and Segmentation of Locally-Advanced Rectal Cancer Based on T2, DWI and DCE Multi-Parametric MRI Using Deep Learning

Yang Zhang1, Liming Shi2, Xiaonan Sun2, Tianye Niu2, Ning Yue3, Peter Chang4, Daniel Chow1, Melissa Khy1, Tiffany Kwong1,3, Jeon-Hor Chen1, Min-Ying Su1, and Ke Nie3

1Department of Radiological Sciences, University of California, Irvine, CA, United States, 2Department of Radiation Oncology, Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, Hangzhou, China, 3Department of Radiation Oncology, Rutgers-The State University of New Jersey, New Brunswick, NJ, United States, 4Department of Radiology, University of California, San Francisco, CA, United States

A deep learning method using the convolutional neural network (CNN) was implemented to segment rectal cancer in 48 patients. Six sets of images (one T2, Two DWI, three DCE) were used as inputs. The Dice Similarity Coefficient (DSC) was used to evaluate results generated by the CNN algorithm compared to the manually outlined ground truth. When the search was done on the entire image the mean DSC was 0.64, and the errors were mainly from tissues outside the rectum. The rectum could be easily segmented, and when the search was confined within 1.5 times of rectal area, the DSC was improved to 0.75.

This abstract and the presentation materials are available to members only; a login is required.

Join Here