Computer giant uses deep learning to diagnose eye disease
IBM team harnesses visual analytics to classify the severity of diabetic retinopathy with an accuracy of 86%
International Business Machines Corporation (IBM) has harnessed deep learning and visual analytics to aid the early detection of diabetic eye disease.
A research team at the multinational technology giant were able to classify the severity of diabetic retinopathy across an internationally recognised scale with an accuracy of 86%.
Findings from the research were presented at the International Symposium on Biomedical Imaging in Melbourne in April.
Using a new method across 35,000 eye images, the IBM technology was trained to identify lesions indicating damage to the retina.
The novel method for both diagnosing the presence of diabetic retinopathy and classifying its severity combines deep learning techniques and dictionary-based data to incorporate pathologies that are specific to the disease.
IBM research scientists hope to further advance the system to enhance its understanding of diabetic retinopathy.
Dr Peter van Wijngaarden, a principal investigator at the Centre for Eye Research Australia, told OT that “alarming projections” of the number of patients with diabetic retinopathy had major implications for the health system
“To substantially reduce the number of people unnecessarily losing vision from diabetic eye disease, there is a real need for innovation to improve effective screening,” he highlighted.