Meeting Banner
Abstract #3919

Noninvasive prediction of tumor-fibrosis using texture analysis of multiparametric MRI in pancreatic cancer model

Dae Chul Jung1, Ravneet Vohra2, Seon Young Lee3, Kyunghwa Han1, Helen Hong3, and Donghoon Lee2

1Radiology, Yonsei University, Seoul, Republic of Korea, 2Radiology, University of Washington, Seattle, WA, United States, 3Software Convergence, Seoul Women’s University, Seoul, Republic of Korea

Authors want to evaluate the correlations between texture features of tumor on multi-parametric MRI (mp-MRI) and tumor-fibrosis in animal model of pancreatic cancer. mp-MRI was performed in a genetically engineered mice model of human pancreatic cancer. Texture features of tumors were extracted from each parametric map using texture analysis. Linear regression with LASSO method was used to evaluate the correlations between the texture features and percentage of fibrosis on histologic slides. Several texture features were correlated with tumor fibrosis. Statistical learning showed preliminary prediction model. Texture analysis of mp-MRI is helpful for predicting and monitoring tumor-fibrosis in pancreatic cancer model.

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

Join Here