A new view: Algorithms diagnose glaucoma

Integrating patients’ personal data, medical retinal image, and genome information through AGLAIA-MIII architecture sets a holistic precedent for automatic objective glaucoma diagnosis and screening, according to a study published online March 28 in the Journal of the American Medical Informatics Association.

Glaucoma is the second leading cause of blindness, with more than 50 percent of cases left undiagnosed because of the lack of an effective, standard screening process. Oftentimes, early treatment is not possible for the asymptomatic disease, which is commonly known as the “silent thief of sight.”

“In view of this, early detection of glaucomatous changes is crucial for timely treatment before the onset of permanent functional visual loss,” wrote Jiang Liu, MD, of Singapore’s Institute for Infocomm Research, and colleagues.

The researchers aimed to design an automatic glaucoma diagnosis through medical imaging informatics that combines personal patient data, medical retinal fundus image, and patient’s genome for screening.

Liu and colleagues piloted AGLIA-MIII screening architecture to diagnosis glaucoma in favor of analyzing intraocular pressure (IOP), which is the traditional risk factor associated with the disease. The researchers used AGLIA-MIII to evaluate data from the Singapore Malay Eye Study (SiMES) database. Comprised of 3,280 subjects, the population based study is from 2004-2007. The researchers used a final data set of 2,258 subjects, 400 of whom had glaucoma.

Before applying learning algorithms to the data, each feature dimension was normalized to range [0 1] to avoid magnitude differences and bias. A Support Vector Machine (SVM) based MKL framework was used to train the classifier for glaucoma assessment. Findings indicated the area under curve (AUC), which is an overall measure of diagnostic strength of a test, was 0.866. The architecture was, therefore, a substantial improvement over the current glaucoma screening approach.

Liu and colleagues discovered that a combination of any two data sources resulted in better performance than individually. The addition of genome information to any other source had the most beneficial effect on diagnosis, as the accuracy of retinal image features was boosted by 18.4 percent and personal data by 57.93 percent. A majority of the results outperformed the current practice of using IOP for glaucoma screening, with a more than two-fold increase in performance over IOP.

“The promising results demonstrated in this work raise the possibility of using a clinical decision support system such as AGLAIA-MII to run in parallel with existing clinical workflows to offer an objective, evidence-based diagnosis to clinicians as a second opinion,” wrote the authors. “From the public healthcare perspective, a carefully designed glaucoma screening programme based on AGLAIA-MII can provide a faster, more cost-effective and more accurate detection of the disease.”