Image classification using deep learning algorithm for thyroid imaging

We conduct image differentiation between benignancy and malignancy for ultrasonography image of thyroid, and also classification of false positive reduction from true positive mass of mammogram images, via convolutional neural networks. For thyroid images we have differentiation accuracy over 76%. For mammogram image classification, we obtained over 80% of accuracy for test datasets. We present the numerical result and corresponding convolutional neural network(CNN) architectures.