Researchers have used an artificial intelligence (AI) and machine learning tool to speed up the screening process for patients with retinal diseases, potentially cutting time to diagnoses and treatment.
Scientists at Shiley Eye Institute at UC San Diego Health and University of California San Diego School of Medicine, with colleagues in China, Germany and Texas developed the solution.
It uses an AI-based convolutional neural network—a class of deep machine learning used to analyse visual imagery—to review more than 200,000 eye scans. The scans were conducted with optical coherence tomography, a non-invasive technology that bounces light off the retina to create two and three-dimensional representations of tissue.
The researchers then used a technique called transfer learning, in which knowledge gained solving one problem is stored and applied to another, before using occlusion testing to identify the areas in each image that are of greatest interest.
“Machine learning is often like a black box where we don’t know exactly what is happening,” said senior author Kang Zhang, MD, PhD, professor of ophthalmology at Shiley Eye Institute and founding director of the Institute for Genomic Medicine at UC San Diego School of Medicine.
“With occlusion testing, the computer can tell us where it is looking in an image to arrive at a diagnosis, so we can figure out why the system got the result it did. This makes the system more transparent and increases our trust in the diagnosis.”
Current computational approaches are laborious and expensive, requiring millions of images to train an AI system.
With simple training the researchers found the machine performed similarly to a well-trained ophthalmologist. It decided whether a patient should be referred for treatment within 30 seconds and with over 95% accuracy.
“Artificial intelligence has huge potential to revolutionise disease diagnosis and management by doing analyses and classifications involving immense amounts of data that are difficult for human experts—and doing them rapidly,” said Zhang.
The scientists also applied AI technology to the diagnosis of childhood pneumonia, based on machine analyses of chest X-rays. They found that the computer was able to differentiate between viral and bacterial pneumonia with more than 90% accuracy.
Zhang also believes that AI technology has many more potential medical applications, such as discerning between benign and malignant lesions detected on scans.
The findings are published in the 22 February issue of Cell.