New Tool to Detect Glaucoma in Eyes With High Axial Myopia
By Lynda Seminara
Selected by Richard K. Parrish II, MD
Journal Highlights
American Journal of Ophthalmology, October 2022
Download PDF
Bowd et al. looked at the diagnostic accuracy of a new en face OCT method aimed at differentiating glaucomatous from nonglaucomatous eyes among adults with high axial myopia. Their comparison of eyes with primary open-angle glaucoma (POAG) and control eyes showed that the new method, which requires segmentation of only one retinal layer, can distinguish glaucomatous eyes from unaffected eyes. This method also was diagnostically superior to standard OCT thickness measurements.
For this research, the authors recruited participants of the Diagnostic Innovations in Glaucoma Study, enrolling adults with high axial length myopia (axial length >26 mm), BCVA of 20/40 or better, and open anterior chamber angles. They explained that the en face method was developed from the SALSA-Texture model and produces a visual pattern reflecting spatial arrangement of an image’s pixel intensities; it can capture the granularity and repetitive patterns of object surfaces. In the study, they generated OCT en face images from 70-μm slabs just below the vitreal border of the inner limiting membrane. The data used for the comparative analysis included areas under the receiver operating characteristic curves (AUROCs) and areas under the precision recall curves (AUPRCs), which were adjusted for both eyes, participant age, axial length, disc area, and image quality.
There were 92 eyes in the POAG cohort and 44 in the control group. The best parameter-adjusted AUROC for differentiating glaucomatous from nonglaucomatous eyes in patients with high myopia was .92 for en face texture images, compared with 0.88 for thickness of the retinal nerve fiber layer (RNFL), .87 for thickness of the ganglion cell–inner plexiform layer, and .87 for thickness of the ganglion cell complex. In eyes with highly advanced myopia (axial length ≥27 mm), the best parameter-adjusted AUROC also was .92 for en face images, whereas thickness-measurement AUROCs were .84 for the RNFL and .86 for each of the other layers. AUPRC values also were superior for the new method.
The authors believe that this novel texture-based analysis can exceed the diagnostic accuracy of standard OCT measurements and avoid the challenges of multiple-layer segmentation.
The original article can be found here.