Optimization and validation of multi-echo, multi-contrast SAGE acquisition in fMRI

Repeatability and reliability of cerebrovascular reactivity in young adults using multi-echo, multi-contrast MRI

Cerebrovascular reactivity (CVR) shows promise as a biomarker of vascular integrity and may benefit from a repeatable, reliable, and microvasculature-sensitive acquisition. A combined spin- and gradient-echo (SAGE) functional MRI (fMRI) acquisition may improve repeatability and reliability compared to single spin- (SE) and gradient-echo (GRE) fMRI and provide a microvascular-weighted analysis. The most repeatable and reliable MRI acquisition CVR maps were compared across three CVR paradigms: a breath-hold task, a breath modulation task, and a resting state acquisition. SAGE-fMRI data was acquired in fifteen young adults at two timepoints. Mean gray matter (GM) within-subject coefficient of variation (wCV) and intraclass correlation coefficient (ICC) were compared within the quantitative and weighted SAGE-fMRI CVR maps and single GRE- and SE-fMRI CVR. Total and microvascular MRI inputs with lowest wCV and highest ICC were used to compare three CVR paradigms. Total and microvascular weighted SAGE-fMRI CVR had the lowest wCV and highest ICC across paradigms. The breath-hold paradigm produced significantly higher GM CVR estimates. SAGE repeatably and reliably measures CVR and offers a simultaneous, complementary analysis on total and microvascular scales. The breath-hold paradigm showed significantly higher CVR estimates, but less compliance-dependent protocols may be ideal for applications in patient populations.

Asymmetric spin echo multi-echo echo planar imaging (ASEME-EPI) sequence for pre-clinical high-field fMRI

In functional magnetic resonance imaging (fMRI) of the blood oxygen level-dependent (BOLD) contrast, gradient-recalled echo (GRE) acquisitions offer high sensitivity but suffer from susceptibility-induced signal loss and lack specificity to microvasculature. In contrast, spin echo (SE) acquisitions provide improved specificity at the cost of reduced sensitivity. This study introduces Asymmetric Spin Echo Multi-Echo Echo Planar Imaging (ASEME-EPI), a technique designed to combine the benefits of both GRE and SE for high-field preclinical fMRI. ASEME-EPI employs a spin echo readout followed by two asymmetric spin echo (ASE) GRE readouts, providing an initial T2-weighted SE image and subsequent T2*-weighted ASE images. A feasibility study for the technique was implemented on a 9.4 T pre-clinical MRI system and tested using a visual stimulation in northern tree shrews. Comparing ASEME-EPI with conventional GRE echo planar imaging (GRE-EPI) and SE echo planar imaging (SE-EPI) acquisitions, results showed that ASEME-EPI achieved BOLD contrast-to-noise ratio (CNR) comparable to GRE-EPI while offering improved specificity in activation maps. ASEME-EPI activation was more confined to the primary visual cortex (V1), unlike GRE-EPI which showed activation extending beyond anatomical boundaries. Additionally, ASEME-EPI demonstrated the ability to recover signal in areas of severe field inhomogeneity where GRE-EPI suffered from signal loss. The performance of ASEME-EPI is attributed to its multi-echo nature, allowing for SNR-optimized combination of echoes, effectively denoising the data. The inclusion of the initial SE also contributes to signal recovery in areas prone to susceptibility artifacts. This feasibility study demonstrates the potential of ASEME-EPI for high-field pre-clinical fMRI, offering a promising compromise between GRE sensitivity and SE specificity while addressing challenges of T2* decay at high field strengths.

Rapid simultaneous estimation of relaxation rates using multi-echo, multi-contrast MRI

PURPOSE

Multi-echo, multi-contrast methods are increasingly used in dynamic imaging studies to simultaneously quantify R2∗ and R2. To overcome the computational challenges associated with nonlinear least squares (NLSQ) fitting, we propose a generalized linear least squares (LLSQ) solution to rapidly fit R2∗ and R2.

METHODS

Spin- and gradient-echo (SAGE) data were simulated across T2∗ and T2 values at high (200) and low (20) SNR. Full (four-parameter) and reduced (three-parameter) parameter fits were implemented and compared with both LLSQ and NLSQ fitting. Fit data were compared to ground truth using concordance correlation coefficient (CCC) and coefficient of variation (CV). In vivo SAGE perfusion data were acquired in 20 subjects with relapsing-remitting multiple sclerosis. LLSQ R2∗ and R2, as well as cerebral blood volume (CBV), were compared with the standard NLSQ approach.

RESULTS

Across all fitting methods, T2∗ was well-fit at high (CCC = 1, CV = 0) and low (CCC ≥ 0.87, CV ≤ 0.08) SNR. Except for short T2∗ values (5-15 ms), T2 was well-fit at high (CCC = 1, CV = 0) and low (CCC ≥ 0.99, CV ≤ 0.03) SNR. In vivo, LLSQ R2∗ and R2 estimates were similar to NLSQ, and there were no differences in R2∗ across fitting methods at high SNR. However, there were some differences at low SNR and for R2 at high and low SNR. In vivo NLSQ and LLSQ three parameter fits performed similarly, as did NLSQ and LLSQ four-parameter fits. LLSQ CBV nearly matched the standard NLSQ method for R2∗- (0.97 ratio) and R2-CBV (0.98 ratio). Voxel-wise whole-brain fitting was faster for LLSQ (3-4 min) than NLSQ (16-18 h).

CONCLUSIONS

LLSQ reliably fit for R2∗ and R2 in simulated and in vivo data. Use of LLSQ methods reduced the computational demand, enabling rapid estimation of R2∗ and R2.