Automatic Quantification of Anterior Lamina Cribrosa Structures in Optical Coherence Tomography Using a Two-Stage CNN...

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Accurate quantification of lamina cribrosa (LC) morphological parameters is crucial for diagnosing optic nerve head diseases.
• Current methods for LC parameter measurement can be time-consuming and subjective.
• Optical coherence tomography (OCT) provides high-resolution cross-sectional images of the optic nerve head.

Purpose of the Study:

• To develop and validate an intelligent system for automatic quantification of LC morphological parameters from OCT images.
• To assess the accuracy and reliability of the proposed deep learning-based system.

Main Methods:

• A two-stage deep learning (DL) model comprising detection (YOLOv3) and segmentation (Attention U-Net) models was developed.
• Post-processing using polynomial regression was employed to refine anterior LC curve boundary information.
• An image processing algorithm was used for final quantification of morphological parameters.

Main Results:

• The system achieved high detection accuracy for Bruch's membrane opening (BMO) (99.92%) and LC (99.18%).
• A strong correlation (R² = 0.96) was observed between predicted and ground-truth values for morphological parameters.
• The system demonstrated accurate and fully automatic quantification of BMO and LC morphological parameters.

Conclusions:

• The proposed intelligent system offers a robust and accurate solution for automated LC morphological parameter quantification.
• This DL-based approach has the potential to improve the efficiency and objectivity of ophthalmic diagnostic procedures.
• The system's high performance validates its utility in clinical settings for optic nerve head analysis.