Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning

Journal: Cell

Published: 2018-02-22

DOI: 10.1016/j.cell.2018.02.010

Affiliations: 7

Authors: 20

Go to article

Research Highlight

Harnessing artificial intelligence for diagnosis

©Photographer is my life/Getty

©Photographer is my life/Getty

A tool that uses the power of neural networks to aid the interpretation of biomedical images could help doctors diagnose and treat diseases, according to a study in Cell.

The use of artificial intelligence for analyzing medical images, such as X-ray and magnetic resonance imaging (MRI) scans, to support diagnostic decision making has faced accuracy and interpretation challenges.

Now, an international team of scientists, including researchers from Sichuan University in China, has used neural networks to assist clinicians in diagnosing diabetic retinopathy, a condition in which an excess of blood sugar damages blood vessels in the back of the eye.

A neural network analyses images of a patient’s eye obtained using a special retina camera to determine the presence of macular degeneration and diabetic retinopathy. This is the first artificial intelligence diagnostic device to be approved by the US Food and Drug Administration.

Supported content

  1. Cell 172, 1122–1131 (2018). doi: 10.1016/j.cell.2018.02.010
Institutions Share
Guangzhou Women and Children's Medical Center, China 0.21
UC San Diego Health System, United States of America (USA) 0.20
University of California, San Diego (UC San Diego), United States of America (USA) 0.20
UC San Diego Health Sciences, United States of America (USA) 0.20
National Collaborative Innovation Center for Biotherapy, China 0.10
Heidelberg Engineering, Germany 0.05
Guangzhou KangRui Biological Pharmaceutical Technology Company, China 0.03