4793
Brain microstructural changes in stroke patients at subacute/chronic stages by mean of advanced diffusion MRI: a longitudinal study1Centre for Medical Image Computing (CMIC), Department of Computer Science, University College London, London, United Kingdom, 2Paris Brain Institute - ICM, Centre de NeuroImagerie de Recherche - CENIR, Paris, France, 3Sorbonne Université, UMR S 1127, Inserm U 1127, CNRS UMR 7225, ICM, F-75013, Paris, France, 4Center for Magnetic Resonance Research and Department of Radiology, University of Minnesota, Minneapolis, MN, United States, 5Department of Neurology, Pitié-Salpétrière Hospital, Paris, France, 6Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, 7Cardiff University Brain Research Imaging Centre (CUBRIC), School of Psychology, Cardiff University, Cardiff, United Kingdom, 8School of Computer Science and Informatics, Cardiff University, Cardiff, United Kingdom
Synopsis
This study characterises microstructural changes in stroke lesions using NODDI and WMTI, two advanced diffusion MRI techniques, and compares their ability to detect WM alterations on the ipsi-lesional and contralesional sides of 12 patients scanned 2 weeks, 1 month and 3 months after onset. Region-of-interest analysis showed that ODI and FWF from NODDI were significantly altered at the lesion. Significant changes were also found in the WMTI parameters of several ipsi-lesional white matter tracts. Our prospective analysis suggests that the application of enhanced dMRI methods such as NODDI and WMTI can help to further comprehend the biophysical mechanism behind ischemia.
Introduction
Traditional diffusion-weighted MRI (dMRI) is a promising non-invasive technique for stroke imaging, although the specificity of traditional diffusion metrics (e.g. mean diffusivity) might be impaired by partial volume contamination from free fluid owing to, for example, vasogenic edema in the subacute period1-3. Neurite Orientation Dispersion and Density Imaging (NODDI) can explicitly account for extracellular fluid by modeling the brain tissue as three compartments (intra-neurite, extra-neurite and free water) to examine the microstructure of neurites4. Previous investigations have revealed that another technique, called White Matter Tract Integrity (WMTI), is also sensitive to specific microstructural alterations associated with white matter (WM) pathological disorders5, 6. Therefore, WMTI may give possible insights to the underlying biophysical causes of WM ischemia, complementary to NODDI.The purpose of this prospective study was to investigate the added value of both NODDI and WMTI to characterize microstructural changes over time of stroke lesions. In addition, we also examine the ability of these techniques to detect alterations in the WM on the ipsi-lesional side with contralesional side at various stages.Method and Materials
12 patients with stroke (men/women=5/7; mean age 64 yrs, range 43-77) were included in this prospective study. All participants were scanned on a Siemens Prisma 3T scanner at three different time points (baseline [after ~2 weeks], first follow-up [after ~1 month] and second follow-up [after ~3 months]). Multi-shell dMRI data were acquired using a Pulsed-Gradient-Spin-Echo sequence with: b-shells: b=300 (8 directions), b=1000 (50 directions), b=2500 (50 directions). The other scan parameters were: TR=3900 ms, TE=62 ms, resolution pixel size = 1.8 x 1.8 x 1.8 mm3, 78 slices. T2-weighted images were also acquired with: TR=9000 ms, TE=95 ms, pixel size = 1 x 1 x 2.5 mm3, and 70 slices. Diffusion tensor metrics (FA, MD) were estimated by FMRIB Software Library (https://fsl.fmrib.ox.ac.uk/fsl)7. NODDI maps (ODI, NDI and FWF) were computed using the AMICO software package (https://github.com/daducci/AMICO) [4]. Diffusion Kurtosis Imaging (MK, Kaxial, Kradial) and WMTI maps (Axonal Water Fraction (AWF), Intra-axonal Diffusivity (Dia), Axial and Radial Extra-axonal Diffusivities (Dea, Der) and Tortuosity (T)) were estimated using the DIPY toolkit (https://dipy.org/)8, 9. Fig 1. shows the flowchart of image processing procedures. The Statistical analyses carried out using MATLAB. For comparisons of lesions and contralateral tissue, paired t-tests were performed at the same time points and between different time points. For comparisons of white matter tracts on ipsi-lesional side with contralesional side, paired t-tests were implemented at the same time points and between different time points. Bonferroni correction was used to correct for multiple comparisons. The threshold for statistical significance was at corrected p < 0.05.Results
Examples of parametric maps obtained for a representative patient, over time, are reported in Fig.2. Our statistical analysis reported in Fig.3 showed that, at each time point, FA was considerably less than that of contralateral tissue, but there was no pattern of change with time. After one month, MD rose considerably and was significantly greater than that of contralateral tissue. After three months, all three DKI indices had a significant decreasing trend and were significantly lower than the contralateral tissue. Two weeks later, the ODI was considerably greater than that of contralateral tissue and trended significantly down with time. After one month, the FWF was much greater than that of contralateral tissue and tended to rise significantly over time.Focusing on WM highly coherent tracts, Fig.4 shows that after three months, the Superior Corona Radiata's AWF decreased significantly on the ipsi-lesional side. Both Dea and Der indices increased considerably after three months in comparison to the second week in the ipsi-lesional Sagittal Stratum. Additionally, mean tortuosity values in the ipsi-lesional Fornix/ Stria Terminalis decreased significantly after three months compared to the second week. After three months, the Superior Longitudinal Fasciculus on the psi-lesional side showed a substantial reduction in the mean values of AWF and Kradial.
Discussion and Conclusion
The observed significant increase of ODI within stroke lesions may indicate axonal disorganization10, however this will gradually return to normal tissue levels over time. After one month, the large rise in FWF may reflect an elevation of free water space after tissue death and clearance. The MD and three DKI metrics exhibited a similar pattern to the FWF. The Superior Corona Radiata's AWF on the ipsi-lesional side may reflect axonal loss11. The alterations in Dea and Der in the ipsi-lesional Sagittal Stratum indicate an increase in extracellular space, presumably as a result of extracellular inflammation12. Der in the ipsi-lesional Cingulum Hippocampus may be attributed to a decrease in extracellular volume produced by axonal beading during ischemia11. Tortuosity abnormalities on the contralateral side of Fornix/Stria Terminalis may be indicative of persistent white matter injury6. Reduced AWF and Kradial on the ipsi-lesional side of the Superior Longitudinal Fasciculus may signify axonal loss.Our results suggest that NODDI is sensitive to microstructural changes in stroke lesions and may be used as a prognostic biomarker. The WMTI indices can detect microstructural alterations in WM tissue after ischemic stroke, with a greater sensitivity than other diffusion indices. However, further investigations, involving larger cohorts, are needed to confirm these findings.
Acknowledgements
MP is supported by UKRI Future Leaders Fellowship (MR/T020296/1). FB and SL acknowledge support from Investissements d’avenir [grant number ANR-10-IAIHU-06 and ANR-11-INBS-0006]. GG is supported by National Institutes of Health [grant numbers R01MH113700, P41 EB015894, P30 NS076408] and the W.M. Keck Foundation.
References
1. González, R.G., et al., Acute ischemic stroke. 2011: Springer.
2. Rosenberg, G.A.J.P.i.c.d., Ischemic brain edema. 1999. 42(3): p. 209-216.
3. Dostovic, Z., et al., Brain edema after ischaemic stroke. 2016. 70(5): p. 339.
4. Zhang, H., et al., NODDI: practical in vivo neurite orientation dispersion and density imaging of the human brain. 2012. 61(4): p. 1000-1016.
5. Chung, S., et al., White matter tract integrity: an indicator of axonal pathology after mild traumatic brain injury. 2018. 35(8): p. 1015-1020.
6. de Kouchkovsky, I., et al., Quantification of normal-appearing white matter tract integrity in multiple sclerosis: a diffusion kurtosis imaging study. 2016. 263(6): p. 1146-1155.
7. Basser, P.J., J. Mattiello, and D.J.B.j. LeBihan, MR diffusion tensor spectroscopy and imaging. 1994. 66(1): p. 259-267.
8. Jensen, J.H., et al., Diffusional kurtosis imaging: the quantification of non‐gaussian water diffusion by means of magnetic resonance imaging. 2005. 53(6): p. 1432-1440.
9. Fieremans, E., J.H. Jensen, and J.A.J.N. Helpern, White matter characterization with diffusional kurtosis imaging. 2011. 58(1): p. 177-188.
10. Mitchell, T., et al., Neurite orientation dispersion and density imaging (NODDI) and free‐water imaging in Parkinsonism. 2019. 40(17): p. 5094-5107.
11. Hui, E.S., et al., Stroke assessment with diffusional kurtosis imaging. 2012. 43(11): p. 2968-2973.
12. Dabrowska, S., et al., Neuroinflammation as a target for treatment of stroke using mesenchymal stem cells and extracellular vesicles. 2019. 16(1): p. 1-17.
13. Keihaninejad, S., et al., An unbiased longitudinal analysis framework for tracking white matter changes using diffusion tensor imaging with application to Alzheimer's disease. 2013. 72: p. 153-163.
14. Mori, S., et al., Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template. 2008. 40(2): p. 570-582.
15. Wang, Z., et al., A study of neurite orientation dispersion and density imaging in ischemic stroke. 2019. 57: p. 28-33.
Figures
Figure 1. This figure presents the flowchart of image processing procedure. (a) This study used a previously published analysis framework for image spatial normalization using DTI-TK software package (http://dti-tk.sourceforge.net/)13. (b) The JHU ICBM-DTI-81 white matter atlas was used to locate 48 white matter ROIs14. (c) The lesion masks were selected manually by experienced neurologists. The contralateral tissue mask were generated by flipping the lesion masks along the sagittal line15. (d) Means of diffusion metrics over the ROIs were computed for statistical analysis.
Figure 2. This figure presents the various diffusion metrics at three points.
Figure 3. This figure presents comparison of lesions with contralateral tissue. (Represented bar chart are colored by different side, orange: lesions, blue: contralateral tissue).
Figure 4. This figure presents comparison of white matter tracts on ipsi-lesional side with contralesional side. (Represented bar chart are colored by different side, orange: ipsi-lesional side, blue: contralesional side).