Deep Learning for Medical Image Recognition: Open Issues and a Way to Forward

Abstract

In the recent decade, deep learning has taken lead over available analysis techniques. Today’s deep learning is used in diversified sectors like health care, traffic management, agriculture, etc. The expectations of researchers or people concerned with deep learning are very high. On another side, healthcare sector is totally different from other industry. It required serious attention, care of people and services (regardless of cost) towards patients; here, issue is a matter of patient’s life. Also, this sector requires a high budget and many people to work in parallel, to provide efficient services to each patient. The interpretations of data in the medical field are vested in the hands of medical experts, and this proves to be quite restrained because of the intervening sophistication, wide ranging varieties spread across a number of compilers, etc. The great victory of deep learning in real-time applications has led to the creation of striking results with extreme accuracy and precision, hence paving way to the spotlight in futuristic health sector applications. In this paper, we discussed state-of-the-art deep learning architecture and its optimization used for medical image segmentation and classification. Also, this sector discusses several useful components like open issues, challenges deep learning-based methods for medical imaging and future research directions.