Automated central macular fluid volume quantification provides a higher diagnostic accuracy than central macular thickness measurement for diabetic macular edema

Abstract

Purpose : To assess the diagnostic accuracy of automated central macular fluid volume (CMFV) quantification for diabetic macular edema (DME).

Methods : All participants underwent comprehensive clinical examinations, 6x6-mm horizontal 19-line macular structural optical coherence tomography (OCT) raster scans and 6x6-mm macular OCT angiography (OCTA) volumetric scans. Two retinal fellows reviewed structural raster OCT scans independently and diagnosed DME if intraretinal or subretinal fluid was detected in central 1-mm region. A retinal specialist arbitrated cases with discrepancy. Mean central macular thickness (CMT) within central 1mm circle was measured on structural OCT scans. A deep-learning based algorithm automatically detected CMFV within central 1mm circle on 6x6-mm OCTA scans.

Results : We enrolled 1 eye each of 202 diabetic patients (90 men) with a mean age of 60 years. DME was diagnosed in 88 eyes. (Table1) The intraclass correlation for DME diagnosis of the two fellows was 0.888. The area under the receiver operating characteristic curve (AROC) of CMFV for diagnosis of DME was 0.904, significantly larger (P=0.03) than AROC of CMT 0.837. (Fig 1.) The sensitivity at 95% specificity of CMFV for detecting DME was 79.1%, significantly higher than 56.0% of CMT (P=0.002). Those DME cases missed by CMT but detected by CMFV had a significantly higher proportion of previous macular focal laser treatment (44% vs. 19%, P=0.016). The Pearson correlation coefficient between CMFV and the best-corrected visual acuity (BCVA) was -0.321, similar to -0.327 between CMT and BCVA.

Conclusions : Automated CMFV quantification provides a significant higher diagnostic accuracy than CMT for DME. CMFV is a useful biomarker for assessment of DME, particularly in previously focal laser treated eyes.

This is a 2020 Imaging in the Eye Conference abstract.