Radiomics-Based Fundus Photography Analysis in Diabetic Retinopathy.

Area of Science:

• Ophthalmology
• Medical Imaging
• Computational Biology

Background:

• Diabetic retinopathy (DR) is a leading cause of vision loss.
• Accurate staging of DR is crucial for timely intervention and management.
• Current diagnostic methods can be subjective and time-consuming.

Purpose of the Study:

• To apply radiomics feature extraction from fundus photographs.
• To automatically identify biomarkers distinguishing different stages of diabetic retinopathy (DR).
• To assess the potential for automated DR classification and screening.

Main Methods:

• Extracted 52 radiomic features from fundus photographs across five DR stages: diabetes without DR, mild, moderate, severe non-proliferative DR (NPDR), and proliferative DR (PDR).
• Analyzed 68 images from 34 eyes with NPDR over a 1-year follow-up.
• Utilized statistical methods including bootstrap confidence intervals, Wilcoxon rank-sum tests, linear regression, and multi-class logistic regression.

Main Results:

• Significant differences in radiomic features were found between eyes without DR and NPDR stages (P < 0.01).
• Several features showed a linear trend with increasing DR severity, with PDR exhibiting a distinct pattern.
• Overall, 34 of 52 features significantly distinguished no DR from NPDR, and 17 features discriminated across all stages.

Conclusions:

• Radiomic features from fundus photographs show potential for distinguishing eyes without DR from NPDR.
• These findings support the use of radiomics for automated DR classification and screening applications.
• Radiomic-derived biomarkers offer automated, objective support for DR screening and staging.