Magnetic resonance imaging enhancement using prior knowledge and a denoising scheme that combines total variation and histogram matching techniques

Introduction Brain perfusion-weighted images obtained through dynamic contrast studies play a critical and clinical role in diagnosis treatment decisions. However, due to the patient's limited exposure radiation, computed magnetic resonance imaging (MRI) suffers from low contrast-to-noise ratios (CNRs). Denoising MRI is task many e-health applications for disease detection. The challenge this research field define novel algorithms strategies capable of improving accuracy performance terms image vision quality computational cost process data. Using statistical information, authors present method by combining total variation-based denoising algorithm with histogram matching (HM) techniques. Methods variation Rudin–Osher–Fatemi (TV-ROF) minimization approach, TV-L2, using isotropic TV setting bounded (BV) component. dual-stage approach tested against two implementations TV-L2: split Bregman (SB) fixed-point (FP) iterations scheme. In HM, study explores approximate exact Coltuc. Results As measured structural similarity index (SIMM), results indicate that more realistic scenarios, FP an HM pairing one best options, improvement up 12.2% over without HM. Discussion findings can be used evaluate investigate advanced machine learning-based approaches developing infer information ad hoc histograms. proposed methods are adapted medical since they account preference expert: single parameter balance preservation (expert-dependent) relevant details degree noise reduction.