There is no reliable guidance available in literature so far for the selection of a suitable technique for denoising Magnetic Resonance (MR) images. The performance of edge-preserving denoising schemes like Nonlocal Means, Bilateral, Total Variation, Anisotropic Diffusion, Kuwahara, wavelet denoising, Linear Minimum Mean Square Error, Smallest Univalue Segment Assimilating Nucleus and Beltrami filters on MR images are evaluated and compared in this paper. Performance evaluation is done on real-time MR Images, Shepp–Logan Phantom images and simulated MR images. Image Quality Analysis indices used for the evaluation are Structural Similarity Index Metric, Noise Quality Measure, Peak Signal to Noise Ratio, Edge Preservation Index, MetricQ, Anisotropic Quality Index, Blind Reference Image Quality Evaluator and computational time. It has been observed that the performance of each filter is completely different on Shepp–Logan, simulated MR and real-time MR images. It is critically sensitive to the strength of noise also. No filter which can offer good performance equally on Phantom, simulated MR image and real-time MR images, is available in the literature. Values of the objective indices are not in concordance with subjective quality ratings. Filter designs optimized on Phantom or simulated MR using maximum PSNR between denoised and ground-truth images as an objective function (minimum error sense in general) do not perform well on real-time MRI.

Introduction

Magnetic Resonance Imaging (MRI) is a modality extensively used in neuroimaging studies (Benou, Veksler, Friedman, & Raviv, 2017; Hermessi, Mourali, & Zagrouba, 2019). In neuroimaging, Magnetic Resonance (MR) images are helpful for both diagnosis and characterization of Multiple Sclerosis, Dementia, Alzheimer’s disease, infectious diseases, intra-cranial lesions etc. MR images are extensively used as assistive tools in image-guided stereotactic surgery and Radiation Treatment (RT) planning also. Compared to other imaging modalities, MR images contain more features and structural details which help the physicians for better diagnosis. The quality of the MR images is usually hindered by random noise (Rundo et al., 2019). Even though the image acquisition techniques have undergone tremendous development in hardware engineering, extenuation of noise via hardware modifications is remaining as an unreached objective in MRI. Noise reduces the visibility of low contrast anatomical structures, especially at low signal-to-noise ratio (SNR). Presence of noise adversely affects the performance of edge-based segmentation schemes used in software packages for computerized image analysis. The presence of noise intervenes with the accurate computation of radiation dosage in RT planning. As it is not trivial to address the issue of noise in MRI through design modifications of the MR equipment, postprocessing techniques have a significant role in improving MR image’s quality. The visual quality of MR images can be improved feasibly by denoising. The conventional neighbourhood averaging techniques do not preserve edges.