University of Pittsburgh researchers have developed a novel image processing technique, virtual averaging (VA), to enhance for Optical Coherence Tomography (OCT) images. Using this approach, nonframe-averaged OCT images collected during a simplified OCT scan can be improved to match the quality of more advanced and time-consuming active eye-tracking frame-averaged OCT images. This novel method could reduce OCT scanning time by as much as 80% and improve the accessibility of OCT scans in patients who struggle to control eye movements for long periods of time.
Description
OCT is commonly used in ophthalmology to produce high resolution, cross-sectional images of the tissues in the eye. The images collected using this non-contact, non-invasive technique can be used to assess tissue structures and support clinical decision-making. To produce clinically useful images, eye and patient movement needs to be minimized, which is not feasible for all patients. A clinical need exists to develop image processing techniques for faster OCT scanning without loss of image quality, particularly for these underserved patients. VA can enhance the images collected over shorter periods of time and could improve accessibility eye health screening, applications of OCT, and patient outcomes.
Applications
- Glaucoma
• Diabetic retinopathy
• Central serous chorioretinopathy (Central serous retinopathy)
• Other retinal diseases
Advantages
Eye motion and blinking impact OCT image quality and traditional image smoothing algorithms blur the boundaries of different anatomical structures, limiting clinical applications. To overcome this issue, some OCT devices use active eye-tracking systems to capture multiple images until no eye movement or blinking is detected, and frame averaging to produce clear retinal images. While these techniques can improve the quality of images, it requires patients to sit still for extended periods which can be difficult for some patients, particularly the elderly or patients with specific eye pathologies.
VA is designed to overcome these challenges. By sampling voxels from OCT signals, adding random Gaussian deviations and averaging results, this approach can dramatically reduce the time required to perform an OCT scan while producing similar qualitative and quantitative images to eye-tracking frame-averaged OCT images.
Invention Readiness
Using 21 healthy volunteers, non-eye-tracking nonframe-averaged OCT images were processed using VA over 15 iterations and compared to active eye-tracking 9-frame-averaged (AETFA) OCT. VA reduced scanning time by 80% while producing clinically meaningful images. These images are comparable to AETFA OCT and could improve clinical applications of OCT.
IP Status
https://patents.google.com/patent/US11170480B2
