Denoising and Multiple Tissue Compartment Visualization of Multi-b-Valued Breast Diffusion MRI.

Area of Science:

• Radiology
• Medical Imaging
• Diffusion MRI

Background:

• Multi-shell diffusion MRI offers valuable breast imaging metrics but faces challenges with low signal-to-noise ratio and extended scan times.
• Addressing these limitations is crucial for enhancing the clinical utility of diffusion MRI in breast cancer assessment.

Purpose of the Study:

• To evaluate the impact of model-based denoising on multi-shell breast diffusion MRI without compromising spatial resolution.
• To assess the effects of data downsampling on multi-shell diffusion MRI metrics.
• To quantify these effects in multi-b-value acquisitions for breast cancer imaging.

Main Methods:

• Utilized prospective multi-shell and retrospective multi-b-value diffusion MRI datasets from healthy subjects and breast cancer patients.
• Analyzed diffusion metrics including mean diffusivity, axial/radial diffusivity, and fractional anisotropy (FA).
• Applied model-based denoising and assessed its impact on image noise, tumor conspicuity, and metric accuracy using statistical tests.

Main Results:

• Denoising effectively reduced fractional anisotropy (FA) by 45-78% and introduced minor biases in mean diffusivity.
• In multi-b-value data, denoising minimally affected apparent diffusion coefficient (ADC) metrics but significantly reduced their coefficient of variation (-1% to -24%).
• Denoising improved the differentiation between tumor and normal tissues and enhanced subjective image quality, particularly in fast ADC maps.

Conclusions:

• Model-based denoising is effective in suppressing spurious high FA values and improving the accuracy of diffusivity metrics in breast diffusion MRI.
• This denoising approach enhances image quality and aids in distinguishing tumor from normal tissue, offering potential for improved breast cancer diagnosis.