Related Experiment Video
Updated: Jul 5, 2025
07:23
Retinal Vascular Reactivity as Assessed by Optical Coherence Tomography Angiography
Published on: March 26, 2020
7.6K
Automatic Identification and Severity Classification of Retinal Biomarkers in SD-OCT Using Dilated Depthwise
Adithiya S V1, Dharani Bai G1, Rajiv Raman2
1School of Electronics Engineering, Vellore Institute of Technology, Vellore, Tamil Nadu, India.
Current Eye Research
|January 22, 2024
Summary
This study introduces a hybrid AI model to detect and grade biomarkers in Uveitic Macular Edema (UME) from OCT scans. The model accurately identifies key indicators, improving early diagnosis and treatment for vision-threatening conditions.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Uveitic Macular Edema (UME) diagnosis via Spectral Domain OCT (SD-OCT) is crucial for monitoring visual impairment.
- Identifying biomarkers like intraretinal cysts (IRC), hyperreflective foci (HRF), hard exudates (HE), and neurosensory detachment (NSD) in OCT B-scans is challenging and time-consuming.
Purpose of the Study:
- To develop an automated image classification hybrid framework for UME biomarker detection and severity prediction.
- To improve the efficiency and accuracy of UME biomarker analysis in OCT B-scans.
Main Methods:
- A dataset of 10880 B-scans from 85 Uveitic patients was utilized.
- A novel Dilated Depthwise Separable Convolution ResNet (DDSC-RN) with SVM classifier was developed for biomarker identification and severity grading.
- The framework was designed for network compression and capturing multi-level features without compromising accuracy.
Main Results:
- The hybrid model achieved 98.64% accuracy in identifying biomarkers, comparable to state-of-the-art models.
- The SVM classifier predicted biomarker severity with an overall accuracy of 89.3%.
- The model demonstrated superior performance in identifying multiple biomarkers in complex OCT B-scans.
Conclusions:
- A novel hybrid model accurately identifies and grades four key retinal biomarkers in OCT B-scans for UME.
- The developed model enhances the effectiveness of clinical screening for UME, potentially leading to improved patient treatment outcomes.
- This AI-driven approach offers a more efficient method for analyzing complex OCT data in ophthalmology.
