Intraretinal fluid identification via enhanced maps using optical coherence tomography images

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Age-related macular degeneration (AMD) and diabetic macular edema (DME) are leading causes of blindness.
• Both conditions involve cystoid fluid in retinal layers.
• Optical coherence tomography (OCT) is crucial for detecting this intraretinal fluid.

Purpose of the Study:

• To develop a novel methodology for identifying and characterizing intraretinal fluid accumulations in OCT images.
• To generate binary maps for direct representation and heat maps for region confidence.
• To evaluate the system's performance and independence from specific OCT devices.

Main Methods:

• Utilized 312 intensity and texture-based features.
• Employed Sequential Forward Selection (SFS) for feature selection.
• Tested LDC, SVM, and Parzen window classifiers, with the best used for map generation.

Main Results:

• Achieved a mean test accuracy higher than 94% across tested classifiers.
• Binary maps correctly classified over 91.33% of OCT images.
• Heat maps correctly classified 93.50% of OCT images, demonstrating device independence.

Conclusions:

• The proposed methodology effectively identifies intraretinal fluid using binary and heat maps.
• The system demonstrates high accuracy and robustness across different OCT devices.
• This approach shows significant potential for screening AMD and DME, aiding early diagnosis and monitoring.