The research team of Professor Soo-Jung Song of Gangbuk Samsung Hospital revealed that they recently developed a deep learning algorithm that automatically diagnoses eye diseases commonly found in medical examination by reading a fundus picture. In particular, the research team explained that it is of great significance in that it automatically reads various diagnosis names beyond finding abnormal findings on a simple fundus photograph.
Fundus photography is currently the most widely used test in ophthalmology and is widely used to diagnose abnormalities in ophthalmic diseases such as retina and glaucoma. This study was developed based on about 40,000 fundus photographs taken at the health examination center and ophthalmology of Gangbuk Samsung Hospital, and is the result obtained from detailed annotations and readings by ophthalmologists and a long-term study.
The fundus photo automatic diagnosis algorithm developed so far is limited to telling only one diagnosis or finding abnormal findings in the fundus photo, such as diabetic retinopathy. It is significant in that it is the closest role to an ophthalmologist.
Professor Song Soo-Jung of the Department of Ophthalmology at Kangbuk Samsung Hospital said, “Since the algorithm developed through this study can automatically diagnose various diseases like an actual ophthalmologist, it is expected that it can be used for screening purposes in medical examination.”
He added, “However, it has not yet replaced the role of an ophthalmologist in clinical practice, and it is expected that it will be able to expand its function in the future, starting with assisting the role of a doctor.”
Meanwhile, the research results were published in the January issue of the electronics journal.
|
