Denoising in Magnetic Resonance Images using Improved Gaussian Smoothing Technique
Beshiba Wilson1, Julia Punitha Malar Dhas2
1Beshiba Wilson, Research Scholar, Department of Computer Science and Engineering, Noorul Islam Centre for Higher Education, Kanyakumari district, India.
2Dr. Julia Punitha Malar Dhas, Professor and Head, Department of Computer Science and Engineering, Noorul Islam Centre for Higher Education, Kanyakumari district, India.
Manuscript received on 09 March 2019 | Revised Manuscript received on 15 March 2019 | Manuscript published on 30 July 2019 | PP: 3693-3696 | Volume-8 Issue-2, July 2019 | Retrieval Number: B2859078219/19©BEIESP | DOI: 10.35940/ijrte.B2859.078219
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Magnetic Resonance Images (MRI) are usually prone to noise like Rician and Gaussian noise. It is very difficult to perform image processing functions with the presence of noise. The objective of our work is to investigate the best method for denoising the MRI images. This study included 25 MRI subjects selected from the Open Access Series of Imaging Studies (OASIS) – 3 database. The 25 brain image subjects includes cases of both men and women aged 60 to 80. The input RGB image is first converted to gray scale image in which the contrast, sharpness, shadow and structure of the color of image are preserved. The proposed work uses an improved Gaussian smoothing technique for denoising Magnetic Resonance Images by constructing a modified mask for Gaussian smoothing. The performance of the proposed technique has been compared with various filters like median filter, Gaussian filter and Gabor filter. The performance evaluation was carried out by metrics like Peak Signal to Noise Ratio (PSNR), Mean Square Error (MSE) and Structural Similarity (SSIM) index. The experimental results show that the Improved Gaussian Smoothing Technique (IGST) performs better than other methods. All experiments were conducted using Scikit Learn version 0.20 and Scikit Image version 0.14.1 under Python version 3.6.7.
Index Terms: Denoising, Gaussian Noise, Improved Gaussian Smoothing Technique (IGST), Magnetic Resonance Images (MRI), Rician Noise, Susceptibility Weighted Images (SWI).
Scope of the Article: Image Processing and Pattern Recognition