Introduction

Intracerebral hemorrhage (ICH) is a significant cause of mortality throughout the world. Management of ICH in terms of clot lysis and iron scavenging after the initial insult is critical for the recovery and appropriate imaging methods to detect and follow the process are needed¹. The objective of this study was to validate quantitative susceptibility mapping (QSM) methodology in a collagenase-induced ICH model in Wistar rats. Working hypothesis was that QSM could show window to observe the susceptibility changes both in the edematous processes and iron environment.

Methods

Male Wistar rats weighing 230-300 g (n=12 for collagenace ICH, n=4 for SHAM operated) were used for the experiment. Intracerebral hemorrhage was induced by intra-striatal infusion of collagenase IV². In vivo T2- and diffusion-mapping MRI were performed at sub-acute 6 hours, 1, 3 and 14 days time-points to characterize the lesion development. Subgroups of rats were perfused, and fixed brains were subjected to QSM MRI at 1 and 3 days post-ICH. MR experiments were performed utilizing 7T (in vivo scans) and 11.7T (ex vivo scans) Bruker BioSpec MR systems (Bruker Biospin, Ettlingen, Germany) and a Bruker volume transmission/surface receiver cross-coil setup. Absolute T2/volumetry MSME: TR/TE=2.5 s/10-120 ms, 1 mm slices and 117 µm in-plane resolution. 1/3 of the trace of the diffusion tensor was acquired using a DW SE sequence TR/TE 2200/28 ms, b-values 0, 1000 s/mm². Slice positioning and resolution identical to T2 measurement. ROI analysis was performed in MATLAB environment for ipsilateral healthy hemisphere, lesion and contralateral hemisphere. Brain oedema was evaluated as percentage difference for hemispheres. Subset of animals (n=4/group at days 1 and 3) were subjected to transcardial saline+PFA perfusion and the brains in skull (incubated in 4 % PFA for two days) were placed into perfluoropolyether (GALDEN) filled tubes for ex vivo scanning. Ex vivo QSM MRI was performed using 3D multi-echo GRE sequence with a TR = 250 ms, N echoes = 8, echo spacing = 4 ms (first 2.7 ms), flip 30° and isotropic resolution of 100 µm³. Field maps for original data were estimated by fitting the complex data over multiple echo times. Laplacian unwrapping and SHARP filtering (threshold=0.05, kernel diameter=9 voxels) were performed to remove residual wraps and background fields. Susceptibility maps were calculated using threshold k-space division (TKD; threshold=2/3 and correction for susceptibility underestimation). Manual ROI-delineation was applied for total visible lesion in the original 3D GRE data and transferred to QSM maps. QSM data were analyzed by histograms (areas normalized to 1 by division on sum of total histogram bin-events) using [-1 +1] range in forty-one 0.05 ppm width bins.

Results

In vivo T2 mapping showed large hypointense lesions corresponding to actual acute hemorrhage at 6 hours which developed progressively hyperintense lesions at later time points (1, 3 and 14 days), figure 1. Diffusion values showed clear increase (while T2 remaining roughly at the same level) from day 1 to day 3, most likely reflecting reduced cytotoxic contribution in the lesion development. QSM revealed large collagenase-induced ICH lesions with low susceptibility core and high susceptibility outer rim (high iron contribution) surrounded again by low susceptibility region outside the actual lesion, figure 2. This “rim-around-rim” is assumed to reflect ongoing cytotoxic edema process whereas the low susceptibility in the core of the lesion relates to vasogenic edema and cell death. Total lesion QSM results by histogram comparison show clear modulation of spread of susceptibility values from day 1 to day 3; namely, day 1 distribution contain significantly higher proportion of both edema processes and iron content than day 3, figure 3a. Positive side of the susceptibility spread within the lesion was modelled with beta-distribution fit (fixed “b”=7.75 average of all individuals, varying fit parameter “a”, figure 3b). This allowed the comparison of the study days 1 and 3 on iron content with simple metrics; beta distribution a = 1.01±0.05 and a = 0.79±0.02** for Day 1 and Day 3 respectively, figure 3c. Data: mean ± SEM, Student’s t-test: **p < 0.01.

Conclusion

Based on the ex vivo data shown here, QSM method seems particularly suitable for in vivo application in a rat model of ICH due to proper lesion size and the clear presence of iron. This combination of methodology and animal model may provide the window to study novel treatments of ICH.

Acknowledgements

No acknowledgement found.