Abstract #2394

Artificial Intelligence Analysis on Prostate DCE-MRI to Distinguish Prostate Cancer and Benign Prostatic Hyperplasia

Yang Zhang¹, Weikang Li², Zhao Zhang², Yingnan Xue², Peter Chang¹, Daniel Chow¹, Min-Ying Su¹, and Qiong Ye^2,3

¹Department of Radiological Sciences, University of California, Irvine, Irvine, CA, United States, ²The Department of Radiology, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, China, ³High Magnetic Field Laboratory, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, China

Three convolutional neural network architectures were applied to differentiate prostate cancer from benign prostate hyperplasia based on DCE-MRI: (1) VGG serial convolutional neural network; (2) one-directional Convolutional Long Short Term Memory (CLSTM) network; (3) bi-directional CLSTM network. A total of 104 patients were analyzed, including 67 prostate cancer and 37 benign prostatic hyperplasia. Upon 10-fold cross-validation, the differentiation accuracy was 0.64-0.77 (mean 0.68) using VGG, 0.75-0.87 (mean 0.81) using the CLSTM, and 0.73-0.89 (mean 0.84) using bi-directional CLSTM. The radiomics model built by SVM using histogram and texture features extracted from the manually-drawn tumor ROI yielded accuracy of 0.81.

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

Join Here