Evaluation of Convolution Neural Network Models Using Clinical Datasets