Anisotropic Diffusion-based Multiscale Medical Image Analysis Technique for COVID-19 Detection

ABSTRACT

The main objective of this work is to develop a novel multi-scale analysis framework for coronavirus disease diagnosis from Chest X-Ray images. A scale-space representation is created by applying the numerical approximation algorithm that solves a new well-posed nonlinear anisotropic diffusion-based model. A 2D circular filter-based feature extraction is performed at each scale and the feature vectors determined at multiple scales are then combined into the final medical image descriptor. Next, a supervised binary classification process is performed on these final feature vectors, by using a training medical image set, its output representing the COVID-19 detection result. Some numerical simulations representing coronavirus detection tests are finally discussed in this paper.