Ex-vivo MRI of human brain hemispheres provides images at essentially the same time-point as histological examination of the tissue, ensuring that no additional pathology develops between imaging and histology¹. This may allow investigation of the effects of various neuropathologies occurring in aging, on the diffusion anisotropy of different brain regions. However, the effects of death, brain extraction from the skull and chemical fixation on the measures of diffusion anisotropy in brain hemispheres are largely unknown. Thus, the value of ex-vivo brain MR fractional anisotropy (FA) remains uncertain. The purpose of this study was two-fold: 1)longitudinally assess the diffusion anisotropy of various white matter regions measured with ex-vivo MRI, and 2)investigate the relationship between FA values measured in-vivo and ex-vivo.

MRI Data: All paticipants were recruited from a longitudinal clinical-pathologic study of aging, provided written informed consent, and signed an anatomical gift act². Brain hemispheres were immersed in formaldehyde solution immediately after extraction from the skull (4.6±1hrs), and were imaged while in formaldehyde solution³. (Dataset_1) Cerebral hemispheres from five elderly subjects (age-at-death=89.6±3.5yrs) were imaged ex-vivo immediately after death (<24hrs.), one day after death and then weekly for one month. (Dataset_2) Cerebral hemispheres from four older adults (age-at-death=94.8±2yrs) were scanned both in-vivo and ex-vivo. The in-vivo acquisitions were performed less than 2 years before death. All subjects were imaged with a 3T MR system using a T1-weighted MPRAGE sequence and a spin-echo EPI DTI sequence.

Processing: For the DTI data, corrections for distortions due to eddy-currents and field non-uniformities, B-matrix reorientation, and diffusion tensor calculation, were conducted using TORTOISE⁴. In Dataset_1, each subject’s FA map was nonlinearly registered to the FA map of the first time point. In Dataset_2, the in-vivo FA map was matched to the ex-vivo FA map by removing the contralateral hemisphere, the brainstem and the cerebellum digitally. Then, the in-vivo FA map was nonlinearly registered to the ex-vivo FA map from the same subject. Linear and non-linear registration was performed with FLIRT⁵ and FNIRT⁶, respectively. The FA values in well registered regions including internal capsule, cingulum and corpus callosum were selected for further comparison(Fig.1).

Longitudinal Assessment of Diffusion Anisotropy Measured with Ex-vivo MRI: Analysis was done on Dataset_1 for each region per subject. The mean FA value of every time-point for each region was normalized by the region’s mean FA value at the first time point. The normalized mean FA values were then plotted as a function of time postmortem(Fig.2).

Relationship Between Diffusion Anisotropy Performed In-Vivo and Ex-Vivo: For Dataset_2, analysis was conducted in a similar manner. For each region, the linear relationship between ex-vivo and in-vivo FA values was studied. Then the ex-vivo FA values from all regions were plotted as a function of their corresponding in-vivo FA values. Linear regression was used to test significant associations in different brain regions(P<0.05).

For the longitudinal assessment of ex-vivo FA, the FA values dropped quickly in the first few days after death and continued to decrease at a slower rate(Fig.2). These results suggest that the FA values of individual white matter regions decrease with time postmortem throughout the brain. The closer to the time of death, the faster the FA values decrease. This information will be crucial for future studies involving ex-vivo MR diffusion anisotropy of human brain hemispheres imaged at different postmortem intervals.

For all subjects of Dataset_2, linear regression showed a linear relationship (P<0.001) between the FA values of brain regions measured ex-vivo and in-vivo in three white matter structures(Fig.3). The slopes were different across regions, potentially due to the effects of death and the fact that additional brain abnormalities may have developed after the in-vivo scan and prior to death. However, this will not be a problem for future voxelwise analyses since each region will be researched independently. The linear behavior in each region suggests that the presented approach for ex-vivo MR diffusion anisotropy captures information that is linked to the ante-mortem structural characteristics of the brain (first step for translation).

This work demonstrated that white matter diffusion anisotropy measured ex-vivo decreases with time postmortem, and that diffusion anisotropy measured ex-vivo is linearly related to the diffusion anisotropy measured in-vivo on the same subjects. Ex-vivo MRI ensures that no additional pathology develops between imaging and histology. Furthermore, ex-vivo MRI allows higher image quality than in-vivo MRI. Finally, frail older adults, an important population for aging research, can only be imaged ex-vivo. Thus, combination of ex-vivo MR diffusion anisotropy and histopathology may become an effective tool for the assessment of the neuropathologic correlates of structural brain abnormalities.