Grid-free interactive and automated data processing for MR chemical shift imaging data

Abstract

Purpose: Today's available chemical shift imaging (CSI) analysis tools are based on Fourier transform of the entire data set prior to interactive display. This strategy is associated with limitations particularly when arbitrary voxel positions within a 3D spatial volume are needed by the user. In this work, we propose and demonstrate a processing-resource-efficient alternative strategy for both interactive and automated CSI data processing up to three spatial dimensions.

Methods: This approach uses real-time voxel-shift by first-order phase manipulation as a basis and therefore allows grid-free voxel positioning within the 3D volume. The corresponding spectrum is extracted from the 4D data (3D spatial/1D spectral) at each time a voxel position is selected. The spatial response function and hence the exact voxel size and shape are calculated in parallel including the same processing parameters. Using this mechanism sequentially along with AMARES time-domain modeling, we also implemented automated quantitative and B (0)-insensitive metabolite mapping.

Results: Metabolite maps of N-acetyl aspartate, choline and creatine were generated using (1)H-CSI data from the brain of healthy volunteers and patients with tumor and epilepsy. (31)P-3D-CSI of the heart of a healthy volunteer is also shown.

Conclusion: The calculated metabolite maps demonstrate good stability and accuracy of the algorithm in all situations tested. The suggested algorithm constitutes therefore an attractive alternative to existing CSI processing strategies.