Ex-vivo and in-vivo ultra-High-Field R2* and QSM microimaging in Alzheimer’s disease

Purpose/Introduction: Conspicuous advances in MRI imaging in the last decade catalyzed the quest for novel approaches to investigate Alzheimer’s disease. Till now a diagnosis is only unequivocally defined by post-mortem histology. High resolution imaging is potentially feasible at ultra-high magnetic field strength, allowing imaging of pathologic processes at a unique level of detail. Recently it has been demonstrated that cortical phase changes in T2* weighted MRI are characteristic for Alzheimer’s disease [1]. b-amyloid deposits likely contribute significantly to the observed phase effect [2,3]. The orientation dependence of such phase effects can be overcome by the use of quantitative susceptibility mapping (QSM). Our purpose is to explore the source of the observed MR phase signal changes by comparing quantitative susceptibility and R2* maps obtained in vivo at 9.4T as well as in post mortem samples at both 9.4T and 14T. The maps are also investigated by histology and here preliminary data are presented. Subjects and Methods: The same frontal cortex area of two post mortem samples from an Alzheimer’s diseased patient and a healthy subject, respectively, were examined at 14T using a GRE-T2*-weighted image (50 lm isotropic voxels, matrix = 1000 9 749 9 512, FOV = 50 9 37.45 9 25.6 mm3, TR = 34.4 ms, TE = 17.5 ms, 4 averages, total scan time = 14.67 h) for QSM and a multi echo sequence for R2* mapping (100 lm isotropic voxels, matrix = 500 9 300 9 256; FOV = 50 9 30 9 25.6 mm3; TR = 27 ms, TE = 4.5, 11, 17.5 ms, acquisition time = 2.3 h). Phase shift information was used to generate data sets for QSM analysis. At 9.4T coronal post mortem brain slices of the same donors, and in vivo measurements of four patients with AD and frontal lobe dementia were also investigated using multi echo (N = 5) 3D-GRE imaging (0.375 9 0.0.375 9 0.8 mm3 voxelsize, FOV = 192 9 174 9 70.4 mm3, matrixsize = 512 9 464 9 88, TR = 35 ms; TE = 6 to 30 ms in steps of 6 ms, total acquisition time = 8,7 s) and high resolution acquisition-weighted 3D-GRE imaging (0.130 9 0.130 9 0.6 mm3 voxelsize, TR = 30 ms, TE = 18 ms, acquisition time = 14 min). Results: Both in vivo and ex vivo R2* and QSM maps showed distinct cortical layering patterns (Fig 1, 2). Compared to healthy subjects, we observed an apparent broadening of the central cortical layer with increased R2* and QSM values consistent with paramagnetic effects in AD. No single plaques could be observed post mortem at the current isotropic voxel size of 50 micrometers. Discussion/Conclusion: Clinical valid methods for studying and, eventually, diagnosing AD in vivo by MRI are emerging. Quantitative R2* and QSM methods at ultra-high-field hold promise for this endeavor and detect changes that involve the layering pattern of the cortical rim. Future studies that target these structures by multi-modal means are necessary to further characterize the signal sources are necessary