Scale- and rotation-invariant fundus tracking system for noninvasive OCT retinal micro-capillary network visualization and retinal disease progression OCT measurements

Abstract

Purpose: To introduce the capability of robust scale- and rotation-invariant fundus tracking system based on the images acquired using Scanning Laser Opthalmoscope (SLO) and exploit it to enhance the recently developed method of retinal micro-capillary network visualization using Optical Coherence Tomography (OCT). To validate the applicability of scale- and rotation-invariant fundus tracking for monitoring of microangiopathic changes using OCT.

Methods: The retina displacements Δψ between the reference and subsequent SLO frames were computed using the Fast Fourier Transform algorithm /FFT/ within the frames of the Fourier shift theorem. The latter, applied to image processing, states that the displacement between the two images corresponds to their phase difference Δφ in Fourier domain. In order to compute the rotations and scale-changes between two SLO images we transformed the images to polar coordinates (r,θ) prior Δφ computation. For the high-quality, accurate and robust tracking system, we developed advanced filtering techniques and applied them to SLO data before phase difference computation. Furthermore, the retina displacements on the SLO image are determined with a sub-pixel accuracy. The values of Δψ are updated during OCT measurement according to the SLO preview and exploited to adjust the OCT trajectory such that only the interested retina area is examined.

Results: The results were obtained in the eyes of 5 healthy volunteers and 10 patients with diabetic retinopathy, branch retinal vein occlusion and central retinal vein occlusion. We show the precision and speed results of our tracking system. Furthermore, we present how the introduced tracking procedure can improve the visualization of microangiopatic changes in small retinal capillary network.

Conclusions: The presented fundus tracking can improve the results of intensity-based optical coherence tomography for visualization of 3D retinal capillary network in the healthy and pathologic eyes. Furthermore, the scale- and rotation-invariant fundus tracking is well suited for disease progression measurements. To this end, only the single SLO image from previous examination is required. The developed tracking system will automatically adjust the scanning path to new patient’s position.

Keywords: 688 retina • 549 image processing