Denoising and Quality Improvement of Mri Images Using Hybrid Filters and Deep Neural Network Based Northern Goshawks Optimization Algorithm

Abstract

Purpose: The main purpose of this work is to enhance the quality of the MRI images by removing the noises. This provides details about the diseases without any doubt and improves the robustness.

Theoretical Structure: In the study of musculoskeletal magnetic resonance imaging, picture denoising and image quality improvement are essential. Electric current is transmitted and received through an electromagnetic coiled wire by radio and magnet communication devices used in magnetic resonance imaging (MRI) systems. Loud noises are produced when the flow of current causes coils to expand. This could have led to the acquisition of noise in the MRI pictures. These days, studies have shown that non-local means (NLM) approaches appear to be far more effective and dependable than traditional local statistical filters such as average filters when Rician noise is introduced.

Methodology: In order to denoise the MRI images for this investigation, we combined and applied a non-local means filter with a Non-subsampled Shearlet transform (NLM-NSST) model. To further improve the appearance of MRI imagery, have implemented the Northern Goshawks optimization (NGO) technique with a deep neural network (DNN).

Findings: The MATLAB software manages the execution of the work on the MRI picture database. When compared to earlier investigations, the suggested study successfully denoises the MRI pictures and enhances the image quality.

Research and Practical Implications: Through this work, we found that it will provide minute details about the disease to physicians and thus improvise the clinical findings.

Originality/Value: This work help in providing the details about improving the MRI scan images for analyzing the diseases with the parameters such as SNR, PSNR, accuracy, MSE, and time consumption.

Purpose: The main purpose of this work is to enhance the quality of the MRI images by removing the noises. This provides details about the diseases without any doubt and improves the robustness.

Theoretical Structure: In the study of musculoskeletal magnetic resonance imaging, picture denoising and image quality improvement are essential. Electric current is transmitted and received through an electromagnetic coiled wire by radio and magnet communication devices used in magnetic resonance imaging (MRI) systems. Loud noises are produced when the flow of current causes coils to expand. This could have led to the acquisition of noise in the MRI pictures. These days, studies have shown that non-local means (NLM) approaches appear to be far more effective and dependable than traditional local statistical filters such as average filters when Rician noise is introduced.

Methodology: In order to denoise the MRI images for this investigation, we combined and applied a non-local means filter with a Non-subsampled Shearlet transform (NLM-NSST) model. To further improve the appearance of MRI imagery, have implemented the Northern Goshawks optimization (NGO) technique with a deep neural network (DNN).

Findings: The MATLAB software manages the execution of the work on the MRI picture database. When compared to earlier investigations, the suggested study successfully denoises the MRI pictures and enhances the image quality.

Research and Practical Implications: Through this work, we found that it will provide minute details about the disease to physicians and thus improvise the clinical findings.

Originality/Value: This work help in providing the details about improving the MRI scan images for analyzing the diseases with the parameters such as SNR, PSNR, accuracy, MSE, and time consumption.