ROI Selection to Improve Prediction Performance of CNN for Cirrhosis Detection Using Ultrasound Images

Cirrhosis is a liver disease and computer-aided diagnostic systems using image processing and machine learning techniques have been developed to help physicians detect cirrhosis as a second opinion. In this paper, we propose applying Reliability Evaluation Method to Region of Interests (ROIs) extracted automatically to make use of ROIs possible for training and classification with CNN models. Our experimental results show the effectiveness of ROI selection for improvement of prediction performance.