Computer-aided diabetic retinopathy detection using trace transforms on digital fundus images.

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Science

Background:

• Diabetic retinopathy (DR) is a primary cause of vision loss in diabetic individuals.
• Early DR detection is crucial for effective treatment, but symptoms often appear late.
• Regular screening is vital, yet resource-intensive due to a shortage of skilled professionals.

Purpose of the Study:

• To develop an automated system for early diabetic retinopathy screening.
• To improve the efficiency and accuracy of DR detection in medical images.
• To reduce the burden on medical professionals by automating the screening process.

Main Methods:

• Utilized trace transforms to model the human visual system for image analysis.
• Employed Support Vector Machine (SVM) with various kernels (quadratic, polynomial, RBF) for feature classification.
• Integrated Probabilistic Neural Network (PNN) and Genetic Algorithm (GA) for parameter optimization and classification.

Main Results:

• Achieved high classification accuracy, with PNN-GA reaching 99.41% and SVM quadratic kernel reaching 99.12%.
• Demonstrated the effectiveness of trace transforms in modeling human visual perception for DR detection.
• Validated the performance of machine learning classifiers in an automated DR screening context.

Conclusions:

• The developed automated system shows significant promise for accurate and efficient diabetic retinopathy screening.
• The integration of trace transforms and machine learning offers a viable solution to overcome limitations in manual screening.
• This approach can aid in early diagnosis, potentially preventing vision loss among diabetic populations.