Rapid Simultaneous Mapping of Total and Myelin Water Content, T1 and T2* in Multiple Sclerosis

Quantitative magnetic resonance imaging might provide a more specific insight into disease process, progression and therapeutic response of multiple sclerosis. We present an extension of a previously published approach for the simultaneous mapping of brain T1, T2* and total water content. In addition to those three parameters, the method presented in the current work allows for the measurement of myelin bound water content, a surrogate marker of tissue myelination. Myelin water was measured based on its distinct relaxation with reduced T2*, resulting in a multiexponential decay signal. However, only 10 points could be acquired on the relaxation curve within a maximum echo time of <40ms as the quantitative protocol has been adapted previously for fast acquisitions with whole brain coverage. The sparse sampling required an adaption of the optimisation approach with additional constraints necessary in order to obtain reliable results. Therefore, the corresponding pool fractions were determined using linear optimisation instead of the standard nonnegative least squares (NNLS). The whole approach including the proper choice of constraints was optimised and validated in simulation studies. Furthermore, the independent determination of total water content allows for an absolute quantification of myelin water content, resulting in a more reliable measurement in oedemateous tissue. Whole brain maps of T1, T2*, total and myelin water content were acquired in 12 patients suffering from multiple sclerosis with mild disease grade. With an acquisition time of 10 minutes only, the presented multidimensional quantitative MRI protocol provides an interesting option for the clinical assessment and monitoring of multiple sclerosis, even on a routine basis.

💡 Research Summary

This paper introduces a fast, whole‑brain quantitative MRI protocol that simultaneously maps four parameters—longitudinal relaxation time (T1), effective transverse relaxation time (T2*), total water content, and myelin‑bound water (MW) fraction—in patients with multiple sclerosis (MS). The method builds on a previously published approach for concurrent T1, T2* and total water mapping, extending it by exploiting the markedly shortened T2* of myelin‑associated water to separate it from the bulk water signal using a multiexponential decay model. Because only ten echo points can be acquired within a maximum echo time of less than 40 ms (to keep the scan under ten minutes), conventional non‑negative least‑squares (NNLS) fitting becomes unstable. The authors therefore replace NNLS with a constrained linear optimisation that enforces non‑negativity, limits the sum of pool fractions to ≤1, and incorporates physiologically plausible bounds.

The acquisition employs a 3D gradient‑echo sequence with multiple inversion times and ten gradient echoes. Spatial resolution is 1 mm isotropic, repetition time is about 2 s, and the total scan time is roughly ten minutes, making the protocol feasible for routine clinical use. Data processing proceeds in two stages: first, voxel‑wise non‑linear fitting yields T1, T2* and total water content; second, with total water fixed, a linear system solves for the fractions of myelin‑bound and free water pools under the imposed constraints. Simulation studies demonstrate that, for signal‑to‑noise ratios (SNR) between 30 and 50, the method recovers the true pool fractions with mean absolute errors of 3–5 %, even when the T2* difference between pools is modest. Importantly, because total water is measured independently, the MW fraction can be expressed in absolute terms (percentage of total water), which reduces bias in edematous tissue where bulk water is elevated.

Clinical validation was performed in twelve patients with mild MS (average Expanded Disability Status Scale ≈1.5). Whole‑brain maps of T1, T2*, total water, and MW were successfully generated. Lesional areas showed the expected increase in T1 (≈1100–1300 ms) and decrease in T2* (≈40–45 ms) relative to normal‑appearing white matter (NAWM). MW fraction in lesions was reduced by roughly 15–20 % compared with NAWM, while total water content was elevated in regions with accompanying edema. The absolute quantification of MW prevented the apparent “pseudo‑decrease” that would occur if only relative fractions were reported, thereby providing a more accurate picture of demyelination in edematous lesions.

The study highlights several strengths: (1) simultaneous acquisition of four quantitative biomarkers within a clinically acceptable scan time; (2) robust estimation of MW fraction despite sparse sampling, achieved through constrained linear optimisation; (3) absolute MW quantification that is less susceptible to confounding effects of edema. Limitations include the reliance on high SNR for reliable multiexponential separation, the potential oversimplification inherent in the imposed constraints, and the modest sample size limited to patients with mild disease. Future work could explore higher field strengths (e.g., 7 T), larger and more diverse patient cohorts, and integration with advanced diffusion or susceptibility imaging to further dissect MS pathology.

In summary, the authors present a technically elegant and clinically practical MRI protocol that delivers concurrent maps of T1, T2*, total water, and myelin‑bound water in ten minutes. By providing absolute measures of myelin water content alongside conventional relaxation and hydration metrics, this approach offers a promising tool for more specific monitoring of disease activity, progression, and therapeutic response in multiple sclerosis and potentially other neurodegenerative disorders.