Staging of Hepatocellular Carcinoma Using Deep Feature in Contrast-Enhanced MR Images
Qiyao Wang, Dashun Que · 2017
Clinical stage of hepatocellular carcinoma (HCC) is of great significance for prognosis.Texture features of HCC in Contrast-enhanced MR images have been effective for predictions of staging.However, texture features are low-level features, which are usually insufficient to capture the complicated characteristics of HCCs.Recently, some studies have been dedicated to learning features in a data driven way for predictions.In this study, we use deep learning that can extract highlevel features in order to more accurately staging HCCs.Experimental results demonstrate that deep feature outperforms traditional texture features for HCC staging, and ensembles of deep features derived from multiview observations of HCCs yield best results.