Abstract
Purpose :
Purpose: To assess the utility of various enhancement techniques for retinal images generated on near-infrared scanning laser ophthalmoscopy (SLO) and its' implication on the management of common retinal diseases such as diabetic retinopathy.
Hypothesis: Enhanced contrast, brightness, and sharpness result in better visualization and assessment of retinal pathology such as diabetic retinopathy, as compared with standard images from near-infrared SLO. Enhanced images can guide clinicians in making more informed management decisions for common retinal diseases.
Methods :
Sixty near-infrared SLO images with various retinal findings were previously acquired with Spectralis OCT (Heidelberg engineering) and extracted as standard images with no enhancement (Group 0). The same images were enhanced using 3 different methods. In Group 1 each image was manually enhanced on an open-source raster graphics editor GIMP. In Group 2 images were manually enhanced using the proprietary Heidelberg software. In group 3 images were processed using automatic enhancement software custom developed by our group. Two independent, masked clinicians compared the images to determine superior contrast, brightness, and sharpness under standardized viewing conditions. First, Group 0 was compared with Groups 1,2, and 3. A secondary comparison was performed amongst images from groups 1, 2, and 3.
Results :
Enhanced images were noted to have superior contrast and resolution, particularly those processed using Heidelberg software and automatic enhancement software where the enhancement was more uniform. There was near 100% inter- and intra-grader agreement noted on all standard and test image combinations, indicating that enhanced images were easily favored for resolution and contrast.
Conclusions :
Extracted near-infrared SLO imaging enhanced by altering gain and contrast was reported as better quality imaging for assessing retinal pathology. There was a strong physician/grader agreement reported. The use of automatic software to enhance images was favored. This allowed for faster, and superior, operator-independent visualization of retinal pathological features which may not have been visualized on non-enhanced imaging. This will enable clinicians to achieve better clinical assessments for diagnosis and management.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.