Biexponential fitting to filter CSF-rich voxels and denoise diffusion-weighted image data: improving MADI quantification

Motivation: MADI is a promising new diffusion-based MR method that produces quantitative maps of physiologic parameters: mean cell volume, cell (number) density, and oxidative metabolic activity. Quantitative maps display metabolic activity and uniquely characterize tumors. Accuracy of MADI maps is compromised by CSF-rich voxel contamination. Goal(s): To improve the accuracy of MADI parameter maps by removing CSF-rich voxels and denoising diffusion-weighted images prior to quantification. Approach: Develop and evaluate 1) a filter to remove CSF-rich voxels and 2) denoising of diffusion-weighted images using biexponential fitting. Results: CSF-rich voxels were successfully removed, improving MADI quantitative maps. Denoising remarkably reduced noise in the diffusion-weighted images. Impact: MADI produces quantitative maps of physiologic cellular parameters: cell volume, cell density, and oxidative metabolism. It uniquely characterizes tumor metabolism and treatment response. This work improves MADI accuracy by robust filtering of CSF-rich voxels.