Circular symmetric laplacian mixture model in wavelet diffusion for dental image denoising

Area of Science:

• Medical Imaging
• Signal Processing
• Computational Biology

Background:

• Dental imaging is crucial for diagnosis, but noise degrades image quality.
• Effective noise reduction is essential for accurate interpretation and detection of pathologies like cavities.
• Existing methods may struggle with the complex noise structures in dental scans.

Purpose of the Study:

• To develop and evaluate a novel noise removal technique for dental images.
• To optimize wavelet shrinkage and nonlinear diffusion parameters for speckle noise reduction.
• To assess the method's efficacy in preserving diagnostic information, such as dental cavities.

Main Methods:

• A combination of wavelet shrinkage and nonlinear diffusion was employed for noise removal.
• Novel models for speckle-related modulus were proposed to enhance automatic threshold selection.
• Circular symmetric Laplacian mixture models were evaluated and selected for their suitability with wavelet coefficients.
• Contrast-to-noise ratio (CNR) was numerically evaluated across various dental image types.

Main Results:

• Significant improvements in CNR were observed across different dental image modalities.
• CNR increased from 2.9149 to 38.8813 in anterior-posterior images.
• CNR improved from 41.6131 to 86.3141 in cephal-lateral images.
• The method demonstrated effectiveness in retaining information about natural and artificial cavities.

Conclusions:

• The proposed wavelet shrinkage and nonlinear diffusion method offers substantial noise reduction in dental images.
• The optimized thresholding models improve the accuracy and reliability of the filtering process.
• The technique successfully preserves critical diagnostic features, aiding in cavity detection.