Synopsis

Multiple Sclerosis (MS) has been a condition of particular interest to the QSM community, due to the conspicuity that white matter abnormalities show in QSM contrast images. While several studies have examined small cohort case findings of QSM variations within MS subjects at ultra-high field, an analysis of QSM contrast-to-noise ratio across a wide range of lesions has yet to be performed. Here, we present quantitative CNR analysis of 65 MS lesions spread across 10 subjects scanned with a high resolution 3D protocol at 7T.

Introduction

Quantitative susceptibility mapping (QSM) is an MRI post-processing algorithm that enables approximated estimation of isotropic magnetic susceptibility. QSM has used to assess a variety of neurological conditions. Multiple Sclerosis (MS) has been a condition of particular interest to the QSM community, due to the conspicuity that white matter abnormalities show in QSM contrast images ^[1]. One particular feature with QSM analysis of MS is the iron-rich rings that longitudinally form around MS lesions ^[2]. These features are not visible with other diagnostic imaging contrasts.

While several studies have examined small cohort case findings of QSM variations within MS subjects at ultra-high field, an analysis of QSM contrast-to-noise ratio across a wide range of lesions has yet to be performed. The contrast-to-noise of QSM increases with increased field strength, therefore it is anticipated that the routine benefits of QSM should increase at higher field strengths. As 7T begins to move towards clinical utility, with vendor systems now becoming cleared for clinical use, the advantages of QSM for routine diagnostic application in MS patients is of substantial interest.

Here, we present analysis of 65 MS lesions spread across 10 subjects scanned with a high resolution 3D protocol at 7T. The comparison protocol consisted of 3D-FLAIR, 3D-FSE/TSE T2, and 3D-SWI/QSM (SWAN sequence).

Methods

Clinical MS subjects were recruited into a study approved by the local IRB and provided written consent for their participation. Images were acquired with 0.8mm isotropic resolution with the following crucial parameters utilized for each sequence: FLAIR -- TE=96ms, TR=8s, TI=2163ms, T2-FSE -- TE = 66ms, TR=2500ms, echo train length = 200, 3D-SWAN/QSM -- TR= 46.2, number of echoes = 10.

QSM pre-processing was performed using brain masking in SPM, Laplacian phase unwrapping, fitppm from the MEDI Toolbox ^[3] for field estimation, and RESHARP ^[4] background field removal. Inversion was performed using a localized application of MEDI, extracted from the MEDI Toolbox ^[3].

Lesion analysis and measurements were performed in Horos ^[5]. 2D circular ROIs were placed within the axial slice with the maximal dimension for each identified lesion. At least 2 lesions were identified in every subject, and for subjects with substantial white matter damage, a maximum of the most conspicuous 12 lesions were analyzed. While the lesion sampling was not complete for the cohort (i.e not all lesions in all subjects were analyzed), the number of analyzed lesions was sufficient to test the hypothesis of this study. Contrast to noise was measured by capturing a region within normal white matter and a region outside the head. As QSM cannot be computed outside brain regions, a basic estimate of noise was computed from a flat region of white matter in a representative subject. The standard deviation within this flat region was roughly 0.1 ppb and was used as the noise estimate for QSM measurements.

Results and Discussion

Figure 1 provides representative FLAIR, T2, SWAN, and QSM within a MS subject at 7T. In this case, the QSM lesion shows positive contrast. Figure 2 provides the images for another subject, in which the lesion shows negative contrast in QSM. This is a crucial takeaway from the current study. With QSM, the dynamic range of lesion appearances is far greater than with other contrast mechanisms. As indicated by other studies, this is because of the hypothesized sensitivity of QSM to subtle MS pathophysiology.

The quantitative results of the study are summarized in Table 1. SWAN showed the worst contrast of the 4 evaluated acquisitions. FLAIR was the highest of the conventional contrasts. QSM showed over twice the mean CNR of FLAIR, and when signed (using both positive and negative values), showed almost five times the dynamic range of contrasts. Again, this points to the sensitivity of QSM to biological variations within the lesions. Across the 64 analyzed lesions, even with its increased dynamic range (i.e. variation), the CNR of QSM showed a statistically significant mean increase compared to FLAIR (p < 0.001).

The results of this study quantify the substantial potential value of QSM at 7T as a biomarker for MS progression. Further studies will be required to provide longitudinal tracking of QSM values within lesions and the resulting correlations with clinical outcomes.

Acknowledgements

References

[1] Langkammer, Christian, et al. "Quantitative susceptibility mapping in multiple sclerosis." Radiology 267.2 (2013): 551-559.

[2] Eskreis‐Winkler, Sarah, et al. "Multiple sclerosis lesion geometry in quantitative susceptibility mapping (QSM) and phase imaging." Journal of Magnetic Resonance Imaging 42.1 (2015): 224-229.

[3] MEDI Toolbox: http://pre.weill.cornell.edu/mri/pages/qsm.html

[4] Sun H, Wilman AH. Background field removal using spherical mean value filtering and Tikhonov regularization. Magnetic resonance in medicine. 2014; 71 (3): 1151-1157. [5] https://horosproject.org/