Classification of Mammogram using deep convolutional neural network with small dataset

December 4, 2019

The mammogram is one of the most difficult medical images to analyze. We present a performance evaluation of five top-notch deep learning models. We implement a layer-wise discriminative fine-tuning to optimally train these models within the smallest possible number of epoch without overfitting the small dataset. Likewise, the mixed-precision training was implemented to account for high computation by these deep learning models.