Automated Macular Fluid Volume As a Treatment Indicator for Diabetic Macular Edema

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Diabetic macular edema (DME) is a leading cause of vision loss.
• Accurate assessment of DME severity is crucial for timely treatment.
• Current methods like central subfield thickness (CST) may not fully capture disease activity.

Purpose of the Study:

• To evaluate the diagnostic accuracy of automatically quantified macular fluid volume (MFV) for identifying treatment-required diabetic macular edema (DME).
• To compare the performance of MFV against CST and visual acuity (VA) in predicting treatment decisions for DME.

Main Methods:

• Retrospective analysis of optical coherence tomography (OCT) and OCT angiography scans from 139 eyes with DME.
• A custom deep-learning algorithm automatically quantified MFV.
• Retina specialists made treatment decisions based on standard clinical and OCT findings, without MFV data.

Main Results:

• Macular fluid volume (MFV) showed a higher area under the receiver operating characteristic curve (AUROC) (0.81) than CST (0.67) in predicting treatment decisions (P=.0048).
• MFV, not CST, was significantly associated with treatment decisions in multivariate analysis (P=.0008).
• Untreated eyes meeting MFV treatment thresholds had better visual acuity than treated eyes (P=.0053).

Conclusions:

• Automatically quantified MFV is a more accurate predictor of treatment needs in DME compared to CST.
• MFV may serve as a valuable tool for the ongoing management and monitoring of diabetic macular edema.
• AI-driven MFV quantification offers potential for improved DME patient care.