A practical evaluation of measures derived from compressed sensing diffusion spectrum imaging.

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Radiology

Background:

• Diffusion Spectrum Imaging (DSI) excels at modeling white matter architecture but requires long scan times.
• Compressed Sensing (CS) with sparser q-space sampling offers reduced DSI acquisition time.
• Previous CS-DSI evaluations were limited to non-human or post-mortem data, leaving in vivo accuracy uncertain.

Purpose of the Study:

• To evaluate the accuracy and inter-scan reliability of Compressed Sensing DSI (CS-DSI) for in vivo human brain white matter analysis.
• To compare various CS-DSI schemes with up to 80% scan time reduction against a full DSI acquisition.
• To determine the feasibility of CS-DSI for clinical and research applications.

Main Methods:

• Developed and tested six CS-DSI schemes by subsampling data from a full DSI acquisition in 26 participants across eight sessions.
• Compared accuracy and reliability of white matter bundle segmentation and scalar maps derived from CS-DSI versus full DSI.
• Validated CS-DSI accuracy in a separate prospective dataset of 20 participants.

Main Results:

• CS-DSI demonstrated nearly equivalent accuracy and reliability to full DSI for both bundle segmentation and voxel-wise scalar maps.
• Accuracy and reliability of CS-DSI were higher in white matter bundles that were more consistently segmented by the full DSI.
• Prospective data replication confirmed the accuracy of CS-DSI.

Conclusions:

• CS-DSI is a viable technique for accurate and reliable in vivo delineation of white matter architecture.
• Significant scan time reduction is achievable with CS-DSI without compromising data quality.
• CS-DSI holds substantial promise for advancing clinical neuroimaging and neuroscience research.