Deep Learning Lends a Hand to Pediatric Radiology.
Radiology. 2018 Apr;287(1):323-325. doi: 10.1148/radiol.2018172898. Summers RM1.
Machine learning in radiology is a hot topic. A part of computer science, machine learning is a field in which systems can be designed and trained to learn concepts from data to make predictions. Machine learning, and in particular a subtype called deep learning, has shown high accuracy in performing difficult tasks, such as object recognition in images and speech recognition, and is now of great interest for medical image analysis (1).
Machine learning can be used for a number of applications in radiology, including automated detection of disease, segmentation of lesions, and quantitation. Radiologists are anxious to learn whether and how machine learning will affect their practices. In diverse fields of medical image analysis, including nonradiologic tasks such as diagnosis of skin lesion and retinal photographs (2,3), evidence indicates that machine learning can diagnose disease on images at a level comparable to that of skilled physicians. There are few diagnostic applications in which machine learning performs comparably to board-certified radiologists. In this issue of Radiology, Larson et al (4) present one such example.