Local Complexity Estimation Based Filtering Method in Wavelet Domain for Magnetic Resonance Imaging Denoising

Area of Science:

• Medical Imaging
• Signal Processing
• Image Analysis

Background:

• Magnetic Resonance Imaging (MRI) is crucial for medical diagnosis.
• Image noise, particularly Rician and Additive White Gaussian Noise (AWGN), degrades MRI quality.
• Effective denoising is essential for accurate interpretation and diagnosis.

Purpose of the Study:

• To develop an advanced MRI denoising method using wavelet domain filtering.
• To improve the balance between noise suppression and preservation of fine details in MRI.
• To enhance image contrast for better delineation of anatomical structures.

Main Methods:

• Proposed a local complexity estimation based filtering method in the wavelet domain.
• Developed a threshold selection methodology based on pixel-wise local complexity.
• Compared performance using various wavelets and evaluated against other denoising techniques.

Main Results:

• The proposed method effectively suppresses Rician and AWGN.
• Achieved superior performance in noise suppression while preserving image details compared to existing methods.
• Demonstrated enhanced contrast, facilitating better delineation of regions of interest.
• The algorithm showed consistent results on both simulated and real MRI data.

Conclusions:

• The proposed local complexity estimation method offers efficient and feasible MRI denoising.
• It successfully balances noise removal and detail preservation, outperforming other methods.
• The algorithm's robustness to different noise types (Rician, AWGN) is a significant advantage.
• The method enhances image quality for improved diagnostic accuracy.