Classification of focal liver disease in egyptian patients using ultrasound images and convolutional neural networks

Recently, <span>computer-aided diagnostic systems for various diseases have received great attention. One of the latest technologies used is deep learning architectures analyzing and classifying medical images. In this paper, a new system that uses to classify three focal in liver besides normal proposed. A pre-trained convolutional neural network utilized. Two types networks are used, ResNet50 AlexNet with fully connected (FCNs). After extracting features using learning, FCNs can input images different states disease, such as Normal, Hem, HCC, Cyst. Dataset obtained from Egyptian Liver Research Institute. classifiers utilized, first includes two classes (Normal/Cyst, Normal/Hem, Normal/HCC, HCC/Cyst, HCC/Hem, Cyst/Hem) second contains four (Normal/Cyst/ HCC/Hem) distinguish Using performance criteria, it has been shown two-category given better results than four-class classifier, accordingly hybrid classifier was suggested merge weighted probabilities by each singular classifier. Experimental achieved an accuracy 96.1% which means be assistive method classification disease.</span>