Multi-shell diffusion signal recovery from sparse measurements.

Area of Science:

• Medical Imaging
• Biophysics
• Signal Processing

Background:

• Traditional multi-shell diffusion imaging (MSDI) requires extensive data acquisition, leading to long scan times unsuitable for clinical use.
• Accurate estimation of the ensemble average diffusion propagator (EAP) is crucial for understanding brain microstructure.

Purpose of the Study:

• To develop a novel method for reconstructing diffusion signals from highly undersampled MSDI data, thereby reducing acquisition time.
• To enable accurate EAP estimation in a clinically feasible timeframe.

Main Methods:

• Proposed a novel extension to spherical ridgelets for sparse representation of diffusion signals across multiple q-shells.
• Incorporated modeling of the radial component and spatial smoothness using total variation (TV) minimization.
• Utilized the Alternating Directions Method of Multipliers (ADMM) algorithm for optimal solution derivation.
• Validated the method using phantom data to determine optimal measurement parameters and compared it against the SHORE method.

Main Results:

• The proposed method enables accurate reconstruction of diffusion signals from undersampled MSDI data.
• Phantom studies indicated that at least 60 measurements (across three b-value shells) are necessary for reliable q-space recovery.
• Demonstrated comparable or superior performance to the SHORE method in terms of angular error, peak detection, signal recovery, and RTOP estimation.
• Successfully applied the technique to human in vivo datasets.

Conclusions:

• The developed method significantly reduces MSDI scan times while maintaining accurate signal reconstruction.
• The findings suggest a minimum of 60 measurements are required for robust q-space recovery using this technique.
• This approach holds promise for improving the clinical utility of diffusion MRI by reducing patient burden.