An efficient wavelet and curvelet-based PET image denoising technique

Area of Science:

• Medical Imaging
• Signal Processing
• Computational Science

Background:

• Positron Emission Tomography (PET) imaging is crucial for disease diagnosis but suffers from noise and low resolution.
• Existing denoising methods like wavelet and curvelet transforms have limitations in handling both isotropic and anisotropic features.
• Noise in PET images hinders accurate segmentation and disease screening.

Purpose of the Study:

• To develop an efficient PET image denoising technique.
• To overcome the limitations of individual wavelet and curvelet transform-based denoising methods.
• To improve the accuracy of medical image analysis tasks such as segmentation and tumor identification.

Main Methods:

• A novel PET image denoising technique combining wavelet and curvelet transforms.
• Implementation of a new adaptive threshold selection for wavelet coefficients, integrating BayesShrink and neighborhood window concepts.
• Validation on simulated phantom and clinical PET datasets.

Main Results:

• The proposed method outperforms existing techniques (VisuShrink, BayesShrink, NeighShrink, ModineighShrink, curvelet, wavelet-curvelet) in denoising performance.
• Superior results demonstrated across various metrics including Mean Squared Error (MSE), Signal-to-Noise Ratio (SNR), Peak Signal-to-Noise Ratio (PSNR), and Image Quality Index (IQI).
• Significant improvements observed in medical applications like gray matter segmentation and precise tumor region identification.

Conclusions:

• The combined wavelet-curvelet transform with adaptive thresholding offers superior PET image denoising.
• This method enhances the reliability of PET imaging for clinical diagnosis and research.
• The technique provides a robust solution for improving image quality and analytical accuracy in medical imaging.