Development and validation of a high-resolution hyperspectral imaging system for the retina

Area of Science:

• Ophthalmology
• Biomedical Engineering
• Medical Imaging

Background:

• Early detection of retinal diseases like diabetic retinopathy and macular degeneration is crucial for effective patient care.
• Advanced retinal imaging technologies are needed to improve diagnostic accuracy and treatment outcomes.

Purpose of the Study:

• To develop a high-resolution hyperspectral imaging (HSI) system for the retina.
• To simultaneously extract retinal vessel diameter and oxygenation rate from HSI data.

Main Methods:

• A hyperspectral retinal imaging system was constructed using a snapshot hyperspectral camera, RGB camera, beamsplitter, and imaging endoscope.
• Deep learning-based pansharpening algorithms were developed to generate high-resolution HSI data.
• The system was validated using retina phantoms and in vivo mouse models.

Main Results:

• The deep learning algorithm achieved high accuracy in spatial resolution and spectral fidelity.
• Average arteriole and venule lumen diameters were measured at 31.5 ± 8.7 μm and 45.7 ± 13.6 μm, respectively.
• Average arteriole and venule oxygen saturation (sO2) were measured at 98% and 58%, respectively.

Conclusions:

• A high-resolution hyperspectral imaging system for retinal analysis has been successfully developed and validated.
• The system enables precise measurement of retinal blood vessel dimensions and oxygen saturation.
• This technology holds promise for advancing the early detection and management of retinal diseases.