We have located links that may give you full text access.
JOURNAL ARTICLE
RESEARCH SUPPORT, NON-U.S. GOV'T
Quantitative Comparison of Spherical Deconvolution Approaches to Resolve Complex Fiber Configurations in Diffusion MRI: ISRA-Based vs L2L0 Sparse Methods.
IEEE Transactions on Bio-medical Engineering 2017 December
OBJECTIVE: To quantitatively compare different spherical deconvolution (SD) approaches (ISRA-based and sparse L2L0 algorithms) to resolve crossing fiber in diffusion MRI. The purpose of this comparison is to address the area of application in which each approach could better perform.
METHODS: Image Space Reconstruction Algorithm (ISRA)-based [Richardson-Lucy (RL), damped-RL] and sparse L2L0 algorithms were implemented and evaluated on both simulated data and in vivo datasets. Simulations were performed at different crossing angles (30°-90°), b-values (1000-3000 s/mm2 ), SNR (10-30), number of fibers (1-3). Isotropic compartments and different fiber volume fractions were included to obtain more realistic configurations. In vivo datasets were acquired to confirm simulated results.
RESULTS: A decrease of SNR or b-value reduces the performances of both approaches. L2L0 methods have better performances at low crossing angles (30°-45°) whereas ISRA methods slightly prevail at high crossing angles (>70°). In the medium crossing angle range, the performance depends on the b-value. In the case of single and 3 fibers configurations as well as in complex scenarios (isotropic components, different partial volumes), ISRA algorithms were able to resolve fiber crossing more accurately and they outperform sparse L2L0 methods. In vivo results confirmed simulated trends.
CONCLUSION: Both classes of algorithms can effectively resolve fiber crossing. L2L0 methods are more effective at low crossing angles whereas ISRA approaches have better performances at high angles and are more robust in more realistic configurations.
SIGNIFICANCE: This work provides useful indications to select the best performing SD algorithm depending on the specific application.
METHODS: Image Space Reconstruction Algorithm (ISRA)-based [Richardson-Lucy (RL), damped-RL] and sparse L2L0 algorithms were implemented and evaluated on both simulated data and in vivo datasets. Simulations were performed at different crossing angles (30°-90°), b-values (1000-3000 s/mm2 ), SNR (10-30), number of fibers (1-3). Isotropic compartments and different fiber volume fractions were included to obtain more realistic configurations. In vivo datasets were acquired to confirm simulated results.
RESULTS: A decrease of SNR or b-value reduces the performances of both approaches. L2L0 methods have better performances at low crossing angles (30°-45°) whereas ISRA methods slightly prevail at high crossing angles (>70°). In the medium crossing angle range, the performance depends on the b-value. In the case of single and 3 fibers configurations as well as in complex scenarios (isotropic components, different partial volumes), ISRA algorithms were able to resolve fiber crossing more accurately and they outperform sparse L2L0 methods. In vivo results confirmed simulated trends.
CONCLUSION: Both classes of algorithms can effectively resolve fiber crossing. L2L0 methods are more effective at low crossing angles whereas ISRA approaches have better performances at high angles and are more robust in more realistic configurations.
SIGNIFICANCE: This work provides useful indications to select the best performing SD algorithm depending on the specific application.
Full text links
Related Resources
Get seemless 1-tap access through your institution/university
For the best experience, use the Read mobile app
All material on this website is protected by copyright, Copyright © 1994-2024 by WebMD LLC.
This website also contains material copyrighted by 3rd parties.
By using this service, you agree to our terms of use and privacy policy.
Your Privacy Choices
You can now claim free CME credits for this literature searchClaim now
Get seemless 1-tap access through your institution/university
For the best experience, use the Read mobile app