Spatiotemporal Maps for Dynamic MRI Reconstruction

Area of Science:

• Medical Imaging
• Biophysics
• Signal Processing

Background:

• Dynamic MRI reconstruction commonly uses the partially separable functions (PSF) model.
• The PSF model has limitations when voxels exhibit diverse spatiotemporal characteristics.
• Existing models struggle with spatially varying temporal and spectral properties in MRI data.

Purpose of the Study:

• To introduce a novel spatiotemporal maps (STM) model for dynamic MRI.
• To address the limitations of the PSF model in representing complex MRI signals.
• To enhance accelerated dynamic MRI reconstruction.

Main Methods:

• Developed the STM model, decomposing the MRI signal into spatially dependent spatial and temporal components.
• Leveraged autoregressive properties of (k, t)-space for signal decomposition.
• Utilized advanced signal processing and randomized linear algebra for efficient STM computation from autocalibration data.

Main Results:

• The STM model extends the PSF model by allowing spatial dependence in temporal functions.
• Efficient computation of STMs is demonstrated using signal processing and randomized linear algebra.
• STMs were successfully applied to reconstruct 2D animal gastrointestinal MRI and 3D human functional MRI data.

Conclusions:

• The proposed STM model offers a more robust representation of dynamic MRI signals compared to the PSF model.
• STMs can be integrated into existing reconstruction frameworks for accelerated dynamic MRI.
• The model's effectiveness is validated across different MRI modalities and anatomical regions.