AI Image Analysis Used to Diagnose Glaucoma

AI Image Analysis Used to Diagnose Glaucoma

Artificial Intelligence (AI) systems have shown tremendous potential for use in the diagnosis of disease, with AI image analysis one of the most exciting uses of AI in medicine.

Radiographers, opthalmologists and physicians can look at medical images and identify the signs of disease, but even with extensive training and experience, mistakes can be made. Medical professionals do not have a truly encyclopedic knowledge of every symptom of every disease and it is not possible for any one individual to read all of the scientific research papers related to every medical condition.

Even when medical images are taken, diseases can go undiagnosed for long periods. Patients respond best to treatments if they are administered in the early stages of disease. Any delay in diagnosis could therefore reduce the likelihood of a full recovery being made, or at the very least, involve the patients suffering for longer than necessary.

To improve the speed and accuracy of medical diagnoses, biomedical researchers are looking at AI systems, particularly AI image analysis to ensure medical diagnoses are made more quickly. IBM is investing heavily in the use of AI in medicine.

In 2015, a team of IBM researchers based in Melbourne trained their Watson AI system to conduct AI image analyses of retinal scans to look for abnormalities that could be indicators of glaucoma.

Glaucoma covers a range of conditions that cause pressure in the eyes to increase. As pressure increases, the optic nerve is compressed causing severe pain. Eventually the condition causes blindness. Early diagnosis of the condition is essential to reduce suffering and improve outcomes. Unfortunately, many individuals are unaware that they have glaucoma because the condition does not cause any symptoms in the early stages. However, the condition can be identified from retinal scans.

IBM researchers have now announced the results of their findings. They have confirmed that they were able to train Watson to identify the signs of glaucoma. Watson was fed left and right eye retinal scans of 88,000 patients for the study and was trained to look for the signs of glaucoma. Watson was able to differentiate between left and right eye retinal scans, assess the quality of those scans, measure the optic cup-to-disc ratio, and search for indicators that the patients had glaucoma. IBM says Watson was able to diagnose glaucoma with 95% accuracy

According to IBM Research Australia’s vice president Dr. Joanna Batstone, “This kind of medical image analysis had the capacity to change the delivery of healthcare services.” She went on to say, “[AI] Cognitive technology holds immense promise for confirming the accuracy, reproducibility and efficiency of clinicians’ analyses during the diagnostic workflow.”

The team is now looking at using Watson to perform an AI image analysis to identify other eye disorders such as age-related macular degeneration and diabetic retinopathy.

Leave a Reply