Deep learning in radiology: an overview of the concepts and a survey of the state of the art.

摘要: Deep learning is a branch of artificial intelligence where networks simple interconnected units are used to extract patterns from data in order solve complex problems. algorithms have shown groundbreaking performance variety sophisticated tasks, especially those related images. They often matched or exceeded human performance. Since the medical field radiology mostly relies on extracting useful information images, it very natural application area for deep learning, and research this has rapidly grown recent years. In article, we review clinical reality discuss opportunities algorithms. We also introduce basic concepts including convolutional neural networks. Then, present survey applied radiology. organize studies by types specific tasks that they attempt broad range utilized Finally, briefly challenges incorporating practice future.