Abstract
Purpose :
To validate the performance of an artificial-intelligence (AI) based software for the detection and topographic quantification of intraretinal (IRF) and subretinal fluid (SRF) volumes in eyes affected by diabetic macular edema (DME).
Methods :
After adequately training of the deep learning algorithm using normal and diabetic OCT images, a specific tool was developed for the automatic detection and quantification of IRF and SRF volumes in the 6 mm macular area centered onto the fovea. The topographic distribution of IRF in the central 1mm circle and the 3mm inner and 6mm outer ring of the ETDRS map was also automatically quantified. Eyes affected by DME were then enrolled from four different retina referring centers. Full map scan of each eye was analyzed by the AI automatic quantification software and by clinical evaluation, performed by one masked medical retinal expert. For each eye, the agreement between the software and clinical evaluation was calculated for SRF, and the accuracy of volumes quantification for both IRF and SRF was evaluated.
Results :
Three hundred and three eyes were enrolled in the four centers. Mean central subfield thickness was 386.5 ± 130.2 micron. The mean volume of IRF detected by the software was 0.898 ± 1.367 mm3, with a mean of :13.9±18.0%, 34.4±21.9%, 51.3±30.2% of IRF in the 1, 3, 6 ETDRS mm, respectively, and mean IRF density (%/relative surface area) of 0.088±0.114, 0.047±0.068 and 0.025±0.043 in the 1, 3, 6 ETDRS mm, respectively. SRF was defined as clinically relevant when ≥ 0.002 mm3 and was detected in 43 eyes by the software, with a mean volume of 0.111±0.191 mm3. The agreement for SRF detection was almost perfect (k=0.831, AC1=0.948). The automatic quantification was defined clinically accurate in 289 (95.38%) eyes for IRF and 287 (94.72) for SRF. No error in foveal identification and automatic retinal layers segmentation was identified in any of the analyzed linear scan.
Conclusions :
Accurate, repeatable location and quantification of fluid volumes at OCT in DME are currently mandatory to adequately diagnose, prognosticate and follow treatment response. A fully validated AI based tool, as reported, allows the clinicians to routinely identify and quantify these clinical parameters offering an objective way of precisely diagnosing and following, in a fully quantitative way, DME eyes.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.