New Deep Learning Model Helps Automated Screening of Common Eye Disorders

A new deep learning (DL) model that identifies disease-related features from images of a group of Tohoku University researchers. This ‘lightweight’ DL model can be trained with a small number of images, even ones with a high degree of noise, and is resource-efficient, meaning it is deployable on mobile devices.

Published in the journal Scientific Reports on May 20, 2022.

With many societies aging and limited medical personnel, DL model reliant self-monitored and tele-screening diseases are becoming more routine. Yet, deep learning algorithms are often task specific, and identify or detect general objects such as humans, animals, or road signs.

Identifying diseases, on the other hand, demands precise measurement of tumors, tissue volume, or abnormalities of other sorts. To do so requires a model to look at in separate images and mark boundaries as a process known as segmentation. But the more accurate the prediction, the greater the computational output, rendering them difficult to deploy on mobile devices.

“There is always a trade-off between accuracy, speed and computational resources when it comes to DL models,” says Toru Nakazawa, co-author of the study and professor at Tohoku University’s Department of Ophthalmology. “Our developed model has better segmentation accuracy and enhanced model training reproducibility, even with fewer parameters – making it more efficient and more lightweight when compared to other commercial software.”

Eye for the AI. The developed lightweight model accurately and rapidly detects the image of abnormalities related to diseases. The model is expected to provide accurate analysis on mobile devices / low CPU-GPU resource single-board computers using standalone self-monitoring devices. © Sharma et al.

Professor Nakazawa, Associate Professor Parmanand Sharma, Dr Takahiro Ninomiya, and students from the Department of Ophthalmology worked with Professor Takayuki Okatani from Tohoku University’s Graduate School of Information Sciences to produce the model.

Using low-resource devices, they obtained measurements of the foveal avascular zone, a region with a fovea centralis at the center of the retina, and enhanced screening for glaucoma.

“Our model is also capable of detecting / segmenting optic discs and hemorrhages with high precision,” added Nakazawa.

In the future, the group is a lightweight model of deploying hopefuls for other common eye disorders and other diseases.

Publication Details:

Title: Ophthalmic images of automatic segmentation and analysis for a lightweight deep learning model

Authors: Parmanand Sharma, Takahiro Ninomiya, Kazuko Omodaka, Naoki Takahashi, Takehiro Miya, Noriko Himori, Takayuki Okatani & Toru Nakazawa

Journal: Scientific Reports

DOI: 10.1038 / s41598-022-12486-w

/ Public Release. From this material the originating organization / author (s) may have a point-in-time nature, edited for clarity, style and length. The views and opinions expressed are those of the author (s) .View in full here.

Leave a Reply

Your email address will not be published.

Back to top button
AGADIR-GROUP