Establishing the Validity of Compressed Sensing Diffusion Spectrum Imaging.

Area of Science:

• Neuroimaging
• Biophysics

Background:

• Diffusion Spectrum Imaging (DSI) excels at modeling complex white matter architecture but requires long acquisition times.
• Compressed Sensing (CS) reconstruction combined with sparser q-space sampling offers a potential solution to reduce DSI scan times.
• Previous evaluations of CS-DSI were primarily on non-human or post-mortem data, leaving its in vivo human brain utility uncertain.

Approach:

• Evaluated six CS-DSI schemes with up to 80% scan time reduction against a full DSI protocol.
• Utilized a dataset of 26 participants scanned multiple times with full DSI, from which CS-DSI datasets were simulated.
• Compared accuracy and inter-scan reliability of white matter structure measures (bundle segmentation, scalar maps) between CS-DSI and full DSI.

Key Points:

• CS-DSI achieved accuracy and reliability nearly equivalent to full DSI for both bundle segmentation and voxel-wise scalar maps.
• Accuracy and reliability of CS-DSI were notably higher in white matter bundles that were more consistently segmented by the full DSI method.
• Prospective data from 20 participants confirmed the accuracy of CS-DSI.

Conclusions:

• CS-DSI provides a viable method for accurate and reliable in vivo delineation of white matter architecture.
• This technique substantially reduces scan time, enhancing its potential for clinical and research applications.
• CS-DSI promises to accelerate neuroimaging research and improve diagnostic capabilities for neurological conditions.