snapMRF

snapMRF accelerates magnetic resonance fingerprinting (MRF) reconstruction by performing GPU-accelerated dictionary generation and signal matching using extended phase graph (EPG) formalism and Bloch equation simulation to enable quantitative T1 and T2 parameter mapping.

Key Features:

• GPU Acceleration: Performs parallel processing on GPUs to accelerate dictionary generation (reported 10–1000× speedups) and signal matching (reported 10–100× speedups).
• Extended Phase Graph (EPG) and Bloch Simulation: Uses EPG formalism alongside Bloch equation simulation for signal generation modeling to capture complex MRI signal behavior.
• Dictionary Generation and Signal Matching: Implements GPU-parallelized simulation and matching workflows for MRF dictionary creation and fingerprint matching.
• Unit Tests and Validation: Includes unit tests to verify computational correctness and has been validated with calibration phantoms and in vivo brain studies.
• Support for Variable Sequences: Supports variable sequences for flexible parameter retrieval across different imaging scenarios.
• Real-Time Parameter Map Potential: Achieves sufficient speed to enable potential real-time generation of parameter maps, including with small dictionaries.

Scientific Applications:

• Clinical Imaging: Enables rapid quantitative T1 and T2 mapping for tissue characterization in diagnostic MRI workflows.
• Research and Development: Facilitates exploration of variable MRF sequences to study tissue properties and disease mechanisms and to develop new acquisition strategies.

Methodology:

Parallelizes simulation and matching across GPU threads and uses extended phase graph (EPG) formalism with Bloch equation simulation for signal generation modeling; computational correctness is assessed via unit tests and validation with calibration phantoms and in vivo brain studies.