Varicocele detection in ultrasound images using deep learning

Varicocele is a disease exhibited by an abnormal dilation of the scrotal venous pampiniform plexus. It is a common cause of infertility and in some cases may cause pain or discomfort. This paper presents a new approach to automatic varicocele classification by employing convolutional neural networks with deep learning. The available dataset consists of images converted into different color modes. Each color mode dataset is partitioned, augmented, and trained by employing a deep learning network. Experiments were run with all possible combinations of four different pre-trained models and over three color modes to determine the best combination that achieves the highest performance with and without augmentation. The implementation results of training and testing were evaluated with several metrics. The analysis of the results demonstrated the efficacy of the proposed system as it identified and classified varicocele with a relatively high accuracy that surpassed the previous works with a significant improvement.