AsianScientist (Feb. 1, 2017) – Researchers in China have developed an artificial intelligence (AI) platform that can diagnose congenital cataracts as accurately as human doctors. These findings, published in Nature Biomedical Engineering, could extend to the diagnosis of other rare diseases.
Congenital cataract is a rare disease where babies are born with cloudy lenses that may obscure their vision. Although it is the leading cause of treatable childhood blindness, treatment is among the most difficult and expensive interventions in ophthalmology.
“An estimated 200,000 children are bilaterally blind from cataracts, and many more suffer from partial cataracts that progress and cause increasing visual difficulty as the child ages,” explained study corresponding author, Professor Lin Haotian of Sun Yat-Sen University.
To help make congenital cataract diagnosis more widely available, Lin and his team developed CC-Cruiser, a convolutional neural network-based AI platform that can identify, evaluate and suggest treatment for congenital cataracts.
Building on a database collected under the Childhood Cataract Program of the Chinese Ministry of Health, the researchers trained CC-Cruiser with a dataset of 476 images of normal eyes and 410 images of congenital cataracts, each independently labelled by two experienced ophthalmologists.
The researchers then tested CC-Cruiser in several complex, real-world settings, including a multihospital clinical trial, a website-based test and a comparative performance test between CC-Cruiser and individual human ophthalmologists.
“For the overall accuracy, CC-Cruiser’s performance was comparable to that of a qualified ophthalmologist. Notably, CC-Cruiser successfully diagnosed all potential patients among 50 cases while all of the human ophthalmologists incorrectly flagged a few cases,” Lin told Asian Scientist Magazine.
“For the treatment suggestions, CC-Cruiser provided accurate treatment suggestions for all the patients in need of surgery (no missed cases), with five false positives.”
The performance of CC-Cruiser could further be improved by having a larger dataset, Lin added. To this end, the researchers have built a collaborative cloud platform for data integration and are exploring the feasibility of implementing CC-Cruiser in more non-specialized hospitals.
Such cloud-based platforms could replace or complement the resource-intensive approach to addressing rare diseases. Currently, rare diseases are typically treated at specialized centers which are expensive to run and may be geographically scattered, making it difficult for patients to receive adequate care.
Despite the encouraging results, Lin believes that AI like CC-Cruiser is more likely to augment rather than replace human doctors.
“Machines have the advantages of automation, objectivity and precision, but the human ability to communicate and interact affectively is indispensable for medical treatment. We hold the view that deep learning results collaborating with human analysis will achieve a better health care quality and efficiency,” he concluded.
Copyright: Asian Scientist Magazine; Photo: Shutterstock.
Disclaimer: This article does not necessarily reflect the views of AsianScientist or its staff.