Scalar Parameters Optimization in PDE Based Medical Image Denoising by using Cellular Wave Computing

摘要

In order to help with biomedical images,a set of complex and effective mathematical models areavailable, based on the PDE (PDE - partial differentialequation). On one hand, effective implementation ofthese methods is difficult, due to the difficulty ofdetermining the scalar parameter values, on which theimage processing efficiency depends, while on theother hand, due to the considerable computing powerneeded in order to perform in real time. Currentlythere are no analytical and / or experimental methodsin the literature for the exact values determination ofthe scaled parameters to provide the best results for aspecific image processing.This paper proposes a method for optimizing thevalues of a scaling parameter set, which ensureeffective noise reduction of medical images by usingcellular wave computing. To assess the overallperformance of noise extraction, the error function(quantitative component) and direct visualization(qualitative component) are used at the same time.Moreover, by using this analysis, the degree towhich the CNN templates are robust against the rangeof values of the scalar parameter, is obtainable.