As retinal diseases become a growing global challenge, screening tools harnessing the latest in AI techniques can enable early detection and accurate diagnosis.
Retinal diseases stand as the leading cause of blindness globally. Alarmingly, half of those affected are unaware of their condition until it is too late, highlighting the need for early detection and diagnosis. According to the World Health Organization, in 2019, around 2.2 billion people worldwide were living with some form of visual impairment, with approximately one billion of those cases being preventable.
The integration of machine learning techniques into healthcare has seen remarkable advancements, especially in the medical fields related to vision and eyesight. Models are showing promising capabilities in screening major retinal diseases that can lead to severe visual impairment or blindness if left untreated.
A team of researchers including Khalifa University’s Dr. Bilal Hassan and Prof. Naoufel Werghi has developed a comprehensive review of existing models for retinal disease screening. Their work ǿմý how machine learning techniques and models can enhance early diagnosis and treatment by assisting clinicians in identifying retinal abnormalities through advanced imaging technologies. The research team included experts from Khalifa University in collaboration with Hina Raja and Siamak Yousefi, University of Tennessee; Taimur Hassan, Abu Dhabi University; Muhammad Usman Akram, National University of Sciences and Technology, Pakistan; Hira Raja, Margalla Institute of Health Sciences, Pakistan; and Alaa Abd-alrazaq, Weill Cornell Medicine-Qatar. They published their findings in, a top 1% journal.
“Machine learning is revolutionizing retinal disease screening. By enabling early detection and accurate diagnosis, these screening tools not only improve patient outcomes but also reduce the economic burden associated with advanced retinal diseases.”
— Prof. Naoufel Werghi, Professor of Computer Science, KU
“In the United States, for example, approximately 50 percent of people with retinal conditions are not diagnosed until the disease has progressed significantly,” Prof. Werghi explained. “This delay stems from the insidious nature of many retinal diseases, where symptoms often do not become apparent until the disease has caused irreversible damage. However, early diagnosis, facilitated by AI, can transform outcomes for patients by ensuring timely intervention.”
Machine learning models have emerged as a valuable tool in detecting retinal diseases early. They enable efficient and accurate screening of large populations, especially in regions with limited access to eye care professionals. These models can analyze retinal images, identifying structural abnormalities such as lesions and irregularities in the retinal layers, which can indicate the early onset of diseases like diabetic retinopathy and glaucoma.
Diabetic retinopathy (DR), for example, is caused by prolonged high blood sugar levels, leading to damage in the retinal blood vessels. If left untreated, it can result in complete vision loss. Models trained on vast datasets of retinal images have shown remarkable accuracy in detecting the earliest signs of DR even when the patient has no symptoms. Similarly, glaucoma, often dubbed the “silent thief of sight,” can be diagnosed early by models analyzing the retinal nerve fiber layers. In glaucoma, increased pressure in the eye damages the optic nerve. Assessing the thickness of these retinal layers and detecting early signs of damage can be achieved by these AI models, even before significant vision loss occurs.
“As machine learning techniques continue to evolve, their integration into healthcare presents endless possibilities,” Prof. Werghi said. “By enabling early detection and accurate diagnosis, these screening tools are revolutionizing the field of ophthalmology, offering innovative solutions for early diagnosis and management of retinal diseases.”
Jade Sterling
Science Writer
22 October
