Pre-study for facilitating the discovery of microfluidic properties in blood vessels using retinal fundus images

COVID-19 disease may cause alterations of microfluidic properties of blood circulation in the retinal tissue. For the pre-study of microfluidic blood physiology and transport, retinal fundus images are applied for clinical screening of abnormal vessels forms. However, fundus images captured by operators with various levels of experience have diversity in quality. Low-quality fundus images increase uncertainty in clinical observation and lead to the risk of misdiagnosis. Due to the optical beam of fundus image acquisition and vessel structure of the retina, natural image restoration methods cannot be applied straight to address this problem. The semi-automatic blind deblurring is a useful technique to restore the underlying sharp image given some assumed or known information about the cause of degradation. In this work, we propose a new hybrid algorithm for image restoration that does not require a priori knowledge of the noise distribution. The degraded image is first de-convoluted in Fourier space by parametric Wiener filtering; then, it is smoothed by using the anisotropic diffusion. The filtering model was tested on 177 fundus images. Experiment filtering results show the efficiency of our algorithm with a superlative performance (p-value < 0.05) when compared with state of the art methods.