WHOCARES

WHOCARES retrieves and resolves voxel-wise cardiac pulsation signals from highly accelerated simultaneous multislice (SMS) fMRI time-series without external physiological recordings or ad hoc modeling, enabling retrospective removal of cardiac noise that confounds neural activity measurement.

Key Features:

• Spatio-temporal cardiac resolution: Resolves cardiac pulsations across space and time in fMRI time-series, including signals near blood vessels.
• fMRI hyper-sampling with SMS: Integrates fMRI hyper-sampling with simultaneous multislice (SMS) imaging to increase effective temporal sampling.
• Aliasing mitigation: Overcomes aliasing between heart rate and acquisition sampling frequency (1/TR) to recover cardiac dynamics even when frequencies overlap with neural activation ranges.
• Voxel-wise cardiac regressors: Retrieves voxel-wise cardiac signal regressors without requiring external physiological measurements or ad hoc modeling assumptions.
• Validation and benchmarking: Validated on 695 healthy subjects from the Human Connectome Project and benchmarked against RETROICOR, HAPPY, and pulse oximeter-based regression, demonstrating superior cardiac signal retrieval.

Scientific Applications:

• Retrospective physiological denoising: Enables removal of cardiac-related noise from fMRI datasets that lack physiological recordings.
• Improved neural signal inference: Improves accuracy and reliability of brain activity mapping by reducing cardiac confounds in BOLD signals.
• Cardiac activity mapping in fMRI: Facilitates mapping of cardiac pulsation patterns throughout the brain to study physiological contributions to fMRI measurements.

Methodology:

WHOCARES uses fMRI hyper-sampling combined with simultaneous multislice (SMS) acquisitions to retrieve voxel-wise cardiac signal regressors without external physiological recordings and was validated on 695 Human Connectome Project subjects with comparisons to RETROICOR, HAPPY, and pulse oximeter regression.