Abstract #0829

Machine Learning for Prediction of Chemoradiation Therapy Response in Patients with Locally-Advanced Rectal Cancer (LARC) Using Pre- and Early-Treatment Follow-up Multiparametric MRI

Yang Zhang¹, Liming Shi², Xiaonan Sun², Tianye Niu², Ning Yue³, Tiffany Kwong^1,3, Peter Chang⁴, Melissa Khy¹, Daniel Chow¹, 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 convolutional neural network (CNN) was implemented to predict the response of LARC patients receiving neoadjuvant chemoradiation therapy. The pre-treatment MRI, and the early-treatment follow-up MRI done at 2-3 weeks after the initiation of radiation were used. The MRI protocol included T2, DWI and DCE. A total of 41 patients were studied, with 8 pCR, 27 Tumor Regression Grade 1, and 9 TRG 2+3. The prediction accuracy was 0.71-0.89 for pCR vs. non-pCR; 0.70-0.77 for TRG(0+1) vs. TRG(2+3), not very good due to the limitations of a relatively small dataset. Using manually extracted tumor features in conjunction with neural network classifiers may achieve a higher accuracy.

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

Join Here