Jinhee Jang1, Hyeong-geol Shin2, Yoonho Nam1,3, Jingu Lee2, Jongho Lee2, and Woojun Kim4
1Radiology, Seoul St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea, 2Department of Electrical and Computer Engineering, Seoul National University, Seoul, Republic of Korea, 3Radiology, Eunpyeong St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea, 4Neurology, Seoul St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea
Synopsis
While susceptibility contrast gives details for MS lesions, two major
changes – iron deposition and de-myelination had same contribution on QSM,
increasing bulk magnetic susceptibility. In this work, we applied separation of
positive and negative sources in clinical MS patients, and had a closer look of
in-vivo MS lesions. We demonstrate variable appearances of MS lesions on separation
maps as well as conventional imaging and QSM, and complex distribution and dynamic changes of positive (i.e. iron) and negative (i.e. myelin)
in MS lesions, in cross-sectional and longitudinal observations.
Background
Multiple sclerosis (MS) is a neuro-inflammatory disease with
progressive clinical course. MRI has been a major tool for monitoring disease
extent, activity and treatment response. As MRI has excellent sensitivity to MS
lesions, MR appearance of MS lesions are variable, resulting in relatively low specificity. It is based on the
histopathological variability of MS lesions [1]. Hence, specific MR contrast is
needed that reflects the histopathological characteristics of the MR lesions.
MR images using susceptibility contrast, including phase images
and quantitative susceptibility mapping (QSM), has been one of the new MR
techniques to characterize MS lesion [2]. This popularity is related to that
iron and myelin, those show dynamic changes in MS lesions, have good magnetic
susceptibility contrast (Figure 1A). One caveat is that while the major changes of them,
deposition of paramagnetic iron and loss of diamagnetic myelin could occur simultaneously, both of those changes have the same contribution
to magnetic susceptibility (Figure 1B). Hence, current MR approach, including QSM cannot
differentiate or separate those pathological changes.
Recently, Lee et al [3] proposed a method that enables separation
of positive and negative sources in QSM. The magnetic susceptibility source
separation can provide surrogates for distribution of iron (i.e., positive
susceptibility) and myelin (i.e., negative susceptibility), respectively. This
could be a tailored approach to assess details of MS lesions as an in vivo
histological analytic tool [4]. In this work, we applied the proposed method in
clinical MRI of MS patients, and evaluate findings of separate maps
in variable MS lesions. Methods
From 10 MS patients (all RRMS types) and their 21 MR scans were
analyzed in this study. MRI were done as a part of clinical processes, such as
diagnostic work-up or follow-up.
Diagnosis was done by an
experienced neurologist who was specialized to MS.
MR parameters
MRI scan was done using clinical 3T MRI and 32 channel coil with
following parameters: FOV 216 x 216 x 144 mm3, voxel size 0.5 x 0.5 x 2 mm3, TR
20 ms, TE 5.8 to 36.7 ms with echo spacing of 6.2 ms, flip angle 17°. Parameters of 3D T2-weighted TSE was as follows: FOV 240 X 240 X 175 MM3,
voxel size 1 X 1 X 1 mm3 , TR/TE 2500/300 ms, ETL 135.
Magnetic susceptibility
source separation and QSM processing
The magnetic susceptibility source separation algorithm divides
positive and negative susceptibility source by using magnitude information
(R2’) in addition to phase information used in previous QSM processing [3].
To estimate whole brain R2’ (=R2*-R2) map in clinically feasible scan time,
T2-weighted images (3D TSE) are transformed into R2 images by using a linear
model: R2 = alpha * (log(ITSE/max(ITSE))). The linear coefficient alpha was
determined by linear regression between log(ITSE/max(ITSE)) map and
conventional R2 map estimated from multi-echo spin echo sequence. R2* maps are
acquired by fitting an exponential fit to multi-echo gradient echo signal. Finally,
separated mapping of positive and negative sources were obtained and (χpos And χneg maps, respectively). From frequency estimation using iHARPERELLA [5], QSM calculation was done from iLSQR method [6]. Results
From 10 patients (5 male and 5 female, mean age 32.1 ± 8.3 year old), total of
161 lesions were analyzed (16.1 ± 6.7 lesions/patients).
Mean disease duration was 55.4 ± 32.4 months.
About 3/4 (124/161, 77.0 %) of MS lesions were paramagnetic on
QSM. Among them, 110 (88.7%) lesions showed increased positive source on χpos maps,
and 122 lesions (98.4%) showed loss of negative source on χneg maps
(Figure 2).
Interestingly, details of susceptibility changes differ between
QSM and separation maps (Figure 3a). Fourteen paramagnetic
lesions showed loss on χneg maps
only, suggesting paramagnetic susceptibility is mainly due to demyelination,
not the iron deposition (Figure 3b). In 22 lesions without susceptibility changes on QSM, 2
lesions (10.8%) showed increased paramagnetic contribution on χpos, and
about half of the lesions (12/22) showed decreased contribution of diamagnetic
on χneg.
In total, 27 lesions (16.8%) showed changes during follow-ups. For
lesions were newly noted, and 5 lesions showed contrast-enhancement changes.
Two lesions showed increased size, while 4 lesions showed decreased size.
Interestingly, 11 lesions showed changes on χpos or χneg (Figure 4). Specifically, 8 lesions showed decreased changes on χneg, while 3
lesions showed increased size or degree of defect on χneg. Discussion
In this work, we explored the clinical value of separated
susceptibility mapping in MS lesions characterization and analysis. Proposed
methods were clearly demonstrated pathological changes of MS lesions, those
were well correlated with variable histopathological changes of MS plaques [1,2]. We found that most MS lesions have increased positive sources and decreased negative sources simultaneously. Each of them was associated with iron deposition and demyelination, respectively. Also,
separation maps showed details of longitudinal changes of MS lesions, which
cannot be evaluated using conventional methods. This suggested potential of
separation maps as an in-vivo tool to observe MR lesions characters and their
changes. Acknowledgements
This research was
supported by Basic Science Research Program through the National Research
Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2017R1D1A1B03033829)
and the Catholic Medical Center Research Foundation made in the program year of
2018.References
[1].
Ludwin SK. The pathogenesis of multiple sclerosis: relating human pathology to
experimental studies. J Neuropathol Exp Neurol. 2006;65(4):305-18.
[2]
Absinta M, Sati P, Gaitan MI,
Maggi P, Cortese IC, Filippi M, et al. Seven-tesla phase imaging of acute
multiple sclerosis lesions: a new window into the inflammatory process. Annals
of neurology. 2013;74(5):669-78.
[3]
Lee, J., Nam, YH., Choi, JY., Shin, H., Hwang, T., and Lee, J. (2017),
Separating positive and negative susceptibility sources in QSM, Proc. Intl.
Soc. Mag. Reson. Med., 25, 0751.
[4]
Lee, J., Hwang, T., Nam, Y., Jang, J., Kim, W., Oh, S-H., Fukunaga, M., Lee, J.
(2018), Applications of magnetic susceptibility source separation: multiple
sclerosis lesions and line of Gennari, Proc. Intl. Soc. Mag. Reson. Med., 26, 2207
[5]
Li, W., Avram, A. V., Wu, B.,
Xiao, X., & Liu, C. (2014). Integrated Laplacian‐based phase unwrapping and background
phase removal for quantitative susceptibility mapping. NMR in Biomedicine,
27(2), 219-227.
[6]
Li W, Wang N, Yu F, et al. A method for estimating and removing
streaking artifacts in quantitative susceptibility mapping. Neuroimage
2015;108:111-122