Iterative variational mode decomposition based automated detection of glaucoma using fundus images

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer-Aided Diagnosis

Background:

• Glaucoma is a primary cause of irreversible blindness worldwide.
• Current diagnostic methods are manual, time-consuming, and require expert supervision.
• There is a critical need for automated systems for widespread glaucoma screening.

Purpose of the Study:

• To develop and validate a novel automated system for glaucoma diagnosis.
• To utilize digital fundus images for accurate and efficient screening.
• To provide a tool for ophthalmologists to support clinical decision-making.

Main Methods:

• Employing Variational Mode Decomposition (VMD) for iterative image decomposition.
• Extracting key features such as entropy measures (Kapoor, Renyi, Yager) and fractal dimensions from VMD components.
• Utilizing the ReliefF algorithm for feature selection and Least Squares Support Vector Machine (LS-SVM) for classification.

Main Results:

• Achieved high classification accuracies of 95.19% (three-fold cross-validation) and 94.79% (ten-fold cross-validation).
• Demonstrated the effectiveness of VMD and selected features in distinguishing glaucoma from normal cases.
• Validated the system's performance through rigorous cross-validation strategies.

Conclusions:

• The proposed automated system shows significant promise for accurate glaucoma diagnosis.
• This method can facilitate mass screening, improving early detection rates.
• The system serves as a valuable adjunct tool for ophthalmologists in interpreting fundus images.