Wavelet-based noise-model driven denoising algorithm for differential phase contrast mammography.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Radiology

Background:

• Traditional mammography has limitations in detecting subtle breast tissue differences.
• Phase contrast and scattering X-ray imaging offer complementary information based on electron density and scattering power.
• Grating-based X-ray interferometry provides absorption, differential phase contrast (DPC), and scattering signals simultaneously.

Purpose of the Study:

• To develop and evaluate a wavelet-based denoising algorithm for improving image quality in grating-based X-ray interferometry for breast cancer screening.
• To address the significant noise in DPC and scattering signals obtained from whole native breast samples.
• To enhance visualization and diagnostic potential of novel X-ray imaging techniques.

Main Methods:

• Investigated noise models for absorption, DPC, and scattering signals from grating-based X-ray interferometry.
• Developed a wavelet-based denoising algorithm utilizing computed noise variance.
• Evaluated the algorithm using both simulated and experimental mammography data from whole native breast samples.

Main Results:

• The proposed wavelet-based denoising algorithm effectively reduced noise in DPC and scattering signals.
• The method demonstrated good denoising quality while preserving crucial edges and structural features in the images.
• The algorithm proved superior to conventional mammography in preliminary evaluations.

Conclusions:

• Wavelet-based denoising significantly improves image quality from advanced X-ray imaging techniques for breast cancer screening.
• The developed algorithm aids in better visualization and diagnosis by reducing noise in DPC and scattering signals.
• This technique supports the implementation of further post-processing, such as signal fusion, for enhanced diagnostic accuracy.