Technology & Media

AI Program Of Google: Developing Better Algorithms For Identifying Eye Disease

The capability of AI to assist in diagnosing patients for an ordinary diabetic eye disease has gained momentum as a result of a study conducted by Lily Peng and team at the Google AI research group. Often, individuals don’t perceive alterations in their vision in the early phases of the disease. However, as it develops, diabetic retinopathy generally results in vision loss that in several instances cannot be overturned. This makes it vital that individuals with diabetes should have annual screenings.

Neural networks were used by Dr Peng and team to identify diabetic retinopathy in previous study. Numerous retinal scans were fed into these neural networks to train them to “perceive” small hemorrhages and other abrasions that are retinopathy’s early warning indications. Dr Peng demonstrated the software functioned nearly as fine as human experts.

However, Dr Peng is fascinated in crafting a method that would be befitting for her grandmother. Thus, to enhance the software’s accuracy, she embraced the input of ophthalmologists, retina specialists who specialize in retinal diseases.

To figure out how this can be executed, Dr Peng evaluated the original algorithm’s performance against manual image grading by either by a consensus grading by 3 retinal specialists or a bulk verdict of 3 general ophthalmologists.

The retina specialists rated the pictures discretely and then functioned collectively to solve any differences. Their evaluation and following consensus diagnosis provided substantial perspective into the grading method, assisting to rectify faults such as artifacts resulted from dust stains, differentiating between diverse sorts of hemorrhages, and making more specific descriptions for “gray regions” that make it tricky to make an ultimate analysis.

At the conclusion of the method, the retina specialists specified that the accuracy utilized in the decision course was above that normally utilized in daily clinical practice. Making use of these specialist-graded pictures, Dr Peng can then modify the software that enhanced the performance of their model as well as disease detection.

Vipin Shettigar

With more than 5 Years of experience in the field of energy and resources, Vipin also is a source of Industry news from technology and media domains. With his jolly nature and humor, Vipin always cheers up the team when in stress and load. Surely, he has turned out to be an essential factor of the team. He likes hanging out with his friends in his spare time.

Related Articles