Deep Neural Networks for fatty liver ultrasound images classification

Depth learning has been applied extensively in various fields of computer vision in recent year. Although a CNN-based network structure can obtain the ideal results in many image recognition, it is rarely used to classify the ultrasonic images of the fatty liver. This is principally because the fatty liver ultrasonic image has no obvious texture features and the low resolution. In this paper, we design the network structure for the characteristics of B-mode ultrasonic images, and utilize the CNN-based model to classify fatty liver ultrasound images. The experimental results show that we achieve a satisfactory classification effect through applying the proposed CNN network and this method is better than the traditional method for classifying fatty liver ultrasonic images.