Abstract #0101

Prediction of Neoadjuvant Chemoradiation Therapy Response in Rectal Cancer Using Radiomics Compared to Deep Learning Based on Pre-Treatment and mid-RT MRI

Yang Zhang¹, Liming Shi², Ke Nie³, Xiaonan Sun², Tianye Niu², Ning Yue³, Tiffany Kwong^1,3, Peter Chang¹, Daniel Chow¹, Jeon-Hor Chen^1,4, and Min-Ying Lydia Su¹

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, E-Da Hospital and I-Shou University, Kaohsiung, Taiwan

The capability to predict patients’ response to neoadjuvant chemoradiation therapy is important for improving their management. The multi-parametric MRI (T2, DWI, DCE) performed before treatment and after 3-4 weeks of radiation were analyzed to predict final pathological response. Quantitative radiomics was performed using GLCM texture and histogram parameters, and also ROI and deep learning using convolutional neural network (CNN) were performed. Combining quantitative radiomics features with tumor volume and diffusion coefficient could achieve accuracy of 0.86 for pCR vs. non-pCR and 0.93 for GR vs. non-GR, and adding follow-up to pre-treatment MRI could improve accuracy, especially for CNN analysis.

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

Join Here