Keywords: Analysis/Processing, Quantitative Imaging, Nonlinear Least Squares, Mono-exponential, Deep Learning, Multi-component T1$$$\rho$$$ maps
Motivation: Non-linear least squares (NLS)-based T1$$$\rho$$$ fitting converges slowly and sensitive to initial guesses. Deep learning-based methods offer efficient alternatives, but suffer from overfitting and generalization problems.
Goal(s): To overcome the limitations of NLS and DL, we introduce HDNLS, which uses DL-based T1ρ fitting as an initial guess for NLS.
Approach: The optimal number of NLS iterations is determined by analyzing the trade-off between precision and fitting time. Finally, four HDNLS variants are suggested: Ultrafast-HDNLS, Superfast-HDNLS, HDNLS, and Relaxed-HDNLS.
Results: HDNLS exhibits similar behavior to NLS and Regularized NLS (RNLS) with a minimum of 13-times increase in fitting speed.
Impact: Our results indicate that HDNLS is faster and exhibits similar behavior to NLS for whole knee joint T1$$$\rho$$$ mapping than NLS. Thus, HDNLS is an alternative to replace NLS and RNLS for T1$$$\rho$$$ mapping when computational time is an issue.
How to access this content:
For one year after publication, abstracts and videos are only open to registrants of this annual meeting. Registrants should use their existing login information. Non-registrant access can be purchased via the ISMRM E-Library.
After one year, current ISMRM & ISMRT members get free access to both the abstracts and videos. Non-members and non-registrants must purchase access via the ISMRM E-Library.
After two years, the meeting proceedings (abstracts) are opened to the public and require no login information. Videos remain behind password for access by members, registrants and E-Library customers.
Keywords