Radiologists traditionally don't pay much attention to the retina, but maybe we should.

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.

orainformatics

New Member
5+ Year Member
Joined
Mar 17, 2017
Messages
10
Reaction score
2
Usable medical image analysis seems far fetched but starting small has shown great potential.

Using machine learning to evaluate the vessels of the retina demonstrated promise to diagnose diabetic retinopathy. Why can't we do the same with ovarian or liver lesions? Thyroid nodules? #ML #radiology

Accuracy of a Deep Learning Algorithm for Detection of Diabetic Retinopathy

Members don't see this ad.
 
As a retina specialist I think that this article is overstating what a machine can do. It's not hard to identify any diabetic retinopathy. You just have the computer look for red spots (dot blot hemorrhages) and differentiate them from the natural blood vessels. I'll bet it can't differentiate diabetic retinopathy from the many other disease that cause hemorrhages.
 
Top