Identification of diabetic glaucoma using the colorimetric method Laguna ONhE and OCT Spectralis

Abstract

Purpose : To verify, in a prospective sample, the retrospective results previously obtained in 173,661 diabetic patients (1) whose advanced stages seem to favor the incidence of glaucoma.

Methods : The Laguna ONhE application analyzes optic nerve color photographs, using Deep Learning, to quantify the topographic distribution of hemoglobin (1) and has been fully automated (2). It provides one index (GDF) useful as a disease identifier and another index (GIP) with more quantitative perfusion information. A total of 717 eyes of 478 diabetics without significant retinopathy were analyzed. 221 eyes (Group 1, G1) were considered normal in the retrospective study (GDF>0), 279 (G2) with moderate-high risk of glaucoma (only one eye with GDF<-15) and 217 (G3) with high risk (both eyes with GDF<-15). The results were compared with the BMO-MRW and RNFL indexes of the Spectralis OCT (Heidelberg, Germany).

Results : The frequency distribution of the results of the three groups (Figure 1) shows a higher differentiation for GDF than for MRW. Figure 2 shows GIP having a curvilinear correlation with MRW results (y=0.005142x2+1.0043x+262.2305, R=0.7065, P<0.0001). Normalizing the data, by the mean value and standard deviation of the normal group, the results for other two groups were: GDF mean=-3.16, SD=1.06, and MRW mean=-1.86 and SD=0.96 (P<0.0001). 14.93% of the normal cases for Laguna ONhE had one of the OCT indices below the 5% percentile of the normal population. This was also true for 44.29% of those considered at moderate risk and 81.57% of those considered at high risk. Most of the glaucomas detected were "low-tension glaucomas", associated with the diabetic microangiopathy, which was detected in the nerve vessels (4).

Conclusions : Laguna ONhE estimates perfusion, achieving a wider dynamic range than OCT which lower limit is limited by remnant tissue.

References:
Gonzalez de la Rosa M et al. IOVS 54: 482-489,2013
Gonzalez-Hernandez M et al. J Clin Med 10:323,2021
Gonzalez-Hernandez M et al. J Clin Med 11:7294,2022
Gonzalez de la Rosa M et al. BMJ Open Ophthal 7:e001100,2022

This abstract was presented at the 2023 ARVO Imaging in the Eye Conference, held in New Orleans, LA, April 21-22, 2023.