Material decomposition for simulated dual-energy breast computed tomography via hybrid optimization method

Area of Science:

• Medical Imaging
• Computational Imaging
• Radiology

Background:

• Dual-energy breast CT reconstruction offers potential for separating microcalcifications from healthy breast tissue, aiding early breast cancer detection.
• Accurate separation is crucial for diagnostic accuracy and requires effective noise suppression algorithms.

Purpose of the Study:

• To investigate and validate a noise suppression algorithm for decomposing simulated breast phantoms into microcalcification and healthy breast components.
• To assess the performance of a novel hybrid optimization method (HOM) in noise reduction and image quality for dual-energy breast CT.

Main Methods:

• A hybrid optimization method (HOM) was developed, utilizing Simultaneous Algebraic Reconstruction Technique (SART) output as a prior image.
• Self-adaptive dictionary learning was employed, incorporating patch sparsity and non-local similarity for image regularization.
• Numerical phantoms with varying Gaussian noise levels were simulated to test the HOM's performance.

Main Results:

• The HOM achieved superior performance with high Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity (SSIM) values across different energy levels (35 kVp and 49 kVp).
• Significant improvements in PSNR were observed compared to existing methods like TWIST, SART, and Filtered Back Projection (FBP).
• The signal-to-noise ratio (SNR) for decomposed normal breast tissue was notably higher with HOM, demonstrating effective noise suppression.

Conclusions:

• Dual-energy reconstruction in breast CT, utilizing the proposed HOM, can effectively detect and separate microcalcifications from healthy tissues without noise amplification.
• The HOM demonstrates superior noise suppression capabilities for both reconstructed and decomposed images compared to competing methods.
• This technique holds promise for enhancing the accuracy and reliability of early breast cancer detection through improved image analysis.