Brain Tissue Differentiation Using Diffusion-Relaxation Correlation Spectrum Imaging with a Spectral Weighting-Based Clustering Method

Motivation: Diffusion-Relaxation Correlation Spectroscopic Imaging (DR-CSI) holds promise for differentiating microscopic tissue components, but challenges remain in accurately resolving spectral peaks and distinguishing specific tissue components. Goal(s): To assess the feasibility of a spectral weighting-based clustering method for partitioning DR-CSI spectra and generating fractional maps for each identified component, improving differentiation of sub-voxel tissue components. Approach: The method was evaluated on a digital phantom simulation across varying signal-to-noise ratios and validated with data from healthy volunteers and brain tumor patients to assess clinical applicability. Results: Our method effectively partitioned components and generated corresponding fractional maps in both phantom and human data. Impact: The integration of DR-CSI spectra with spectral weighting-based clustering enables automated and precise differentiation of tissue components and sub-voxel tissue characterization. This approach shows promise in distinguishing challenging tumor types, potentially transforming brain tumor diagnostics and treatment planning.