This study’s findings suggest that surgical skin markings in dermoscopic images significantly interfered with the convolutional neural networks (CNN’s) correct diagnosis of nevi by increasing the melanoma probability scores and consequently the false-positive rate.
A predominance of skin markings in melanoma training images may have induced the CNN’s association of markings with a melanoma diagnosis.
https://jamanetwork.com/journals/jamadermatology/article-abstract/2740808
This article discusses a number of other applications of deep learning in healthcare. What is very exciting about deep learning is that a model trained on large publicly available data could be applied to an organizations private data using transfer learning approaches. In particular the use cases related to EHRs are quite relevant. https://www.nature.com/articles/s41591-018-0316-z
opengraphobject:[360477432225792 : https://jamanetwork.com/journals/jamadermatology/article-abstract/2740808 : title=“Association Between Surgical Skin Markings in Dermoscopic Images and Diagnostic Performance of a Deep Learning Convolutional Neural Network for Melanoma Recognition | Dermatology | JAMA Dermatology |…” : description=“This cross-sectional study investigates the association of surgical skin markings in dermoscopic images with the diagnostic performance of a deep learning convo”], opengraphobject:[360483256369152 : https://www.nature.com/articles/s41591-018-0316-z : title=“A guide to deep learning in healthcare | Nature Medicine” : description=“A primer for deep-learning techniques for healthcare, centering on deep learning in computer vision, natural language processing, reinforcement learning, and generalized methods.”]