Restoration of retinal images with space-variant blur

Abstract

Abstract. Retinal images are essential clinical resources for the diagnosis of retinopathy and many other ocular diseases. Because of improper acquisition conditions or inherent optical aberrations in the eye, the images are often degraded with blur. In many common cases, the blur varies across the field of view. Most image deblurring algorithms assume a space-invariant blur, which fails in the presence of space-variant (SV) blur. In this work, we propose an innovative strategy for the restoration of retinal images in which we consider the blur to be both unknown and SV. We model the blur by a linear operation interpreted as a convolution with a point-spread function (PSF) that changes with the position in the image. To achieve an artifact-free restoration, we propose a framework for a robust estimation of the SV PSF based on an eye-domain knowledge strategy. The restoration method was tested on artificially and naturally degraded retinal images. The results show an important enhancement, significant enough to leverage the images' clinical use.

Retinal Image Analysis: Image Processing and Feature Extraction Oriented to the Clinical Task

“…Moreover, to achieve an artifact free restoration we used our strategy for detecting the non-valid local PSFs and replacing them with a corrected one [26,27,29].…”

“…In reality we know that the PSF is indeed spatially variant [25], in such a way that in certain cases the space-invariant approach may fail. For that reason, we proposed a space-variant approach [26,27]. To address this limitation we now model the blurred retinal image by a general linear operator…”

Retinal Image Analysis Oriented to the Clinical Task

“…The SI approach may fail because the blur changes across the field of view. For that reason, we proposed a SV BD approach [7]. Because the blur changes smoothly we assume the SV PSF to be locally SI.…”

“…We have dealt with and proposed solutions to problems that arise in retinal image acquisition and longitudinal monitoring of retinal disease evolution. Specifically, non-uniform illumination compensation [2], poor image quality [3], automated focusing [4], image segmentation [5], change detection [6], space-invariant (SI) [6] and space-variant (SV) [7] blind deconvolution (BD). Digital retinal image analysis can be effective and cost-efficient for disease management, computer-aided diagnosis, screening and telemedicine and applicable to a variety of disorders such as glaucoma, macular degeneration, and retinopathy [8,9].…”

In vivo blind‐deconvolution photoacoustic ophthalmoscopy with total variation regularization

“…Alternatively, the blurred imaging can be restored through deconvolution algorithm without physically modifying optical system. To date, the computational approach is adopted for enhancing in vivo photoacoustic microscopy (PAM) . Most deconvolution methods, such as Richardson‐Lucy (RL) deconvolution and Wiener Filtering, require a prior knowledge of the point spread function (PSF) of the imaging system, which is always impractical for in vivo microscopic imaging.…”