Similarity regularized sparse group lasso for cup to disc ratio computation

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Glaucoma detection relies on accurate cup-to-disc ratio (CDR) measurement.
• Automated CDR computation from color fundus images is a promising area for glaucoma diagnosis.
• Numerous algorithms have been developed over the last decade for this task.

Purpose of the Study:

• To review recent advancements in automated CDR computation.
• To present a novel similarity-regularized sparse group lasso method for automated CDR estimation.
• To evaluate the performance of the proposed method against existing techniques.

Main Methods:

• A novel similarity-regularized sparse group lasso method was developed.
• The method reconstructs testing disc images using reference images and sparse group lasso constraints.
• Reconstruction coefficients are utilized for CDR estimation.

Main Results:

• The proposed method was validated on 650 images with manual CDR annotations.
• It achieved an average CDR error of 0.0616 and a correlation coefficient of 0.7.
• Diagnostic test areas under the curve reached 0.843 (manual segmentation) and 0.837 (automatic segmentation), surpassing other methods.

Conclusions:

• The novel similarity-regularized sparse group lasso method demonstrates superior performance for automated CDR estimation.
• This technique shows significant potential for improving glaucoma detection accuracy.
• The method outperforms existing approaches in both CDR error and diagnostic capability.