Abstract:Purpose: To propose a novel Numerical fitting method of the Extrapolated semisolid Magnetization transfer Reference (NEMR) signal for quantifying the CEST effect. Theory and methods: Modified two-pool Bloch-McConnell equations were used to numerically fit the magnetization transfer (MT) and direct water saturation (DS) signals at far off-resonance frequencies, which was subsequently extrapolated into the frequency range of amide proton transfer (APT) and nuclear Overhauser enhancement (NOE) pools. Then the subtraction of the fitted two-pool z-spectrum and the experimentally acquired z-spectrum yielded APT# and NOE# signals mostly free of MT and DS contamination. Several strategies were used to accelerate the NEMR fitting. Furthermore, the proposed NEMR method was compared with the conventional extrapolated semisolid magnetization transfer reference (EMR) and magnetization transfer ratio asymmetry (MTRasym ) methods in simulations and stroke patients. Results: The combination of RF downsampling, MT lineshape look-up table, and conversion of MATLAB code to C code accelerated the NEMR fitting by over 2700-fold. Monte-Carlo simulations showed that NEMR had higher accuracy than EMR and eliminated the requirement of the steady-state condition. In ischemic stroke patients, the NEMR maps at 1 μT removed hypointense artifacts seen on EMR and MTRasym images, and better depicted stroke lesions than EMR. For NEMR, NOE# yielded significantly (p < 0.05) stronger signal contrast between stroke and normal tissues than APT# at 1 μT. Conclusion: The proposed NEMR method is suitable for arbitrary saturation settings and can remove MT and DS contamination from the CEST signal for improved detection of ischemic stroke.

Numerical fitting of Extrapolated semisolid Magnetization transfer Reference (NEMR) signals: Improved detection of ischemic stroke