The microstructure of brain tissues becomes visible using different types of optical microscopy after the tissue sectioning. This preparation procedure introduces stress and strain in the anisotropic and inhomogeneous soft tissue slices, which are several 10 µm thick. Consequently, the three-dimensional dataset, generated out of the two-dimensional images with lateral sub-micrometer resolution, needs algorithms to correct the deformations, which can be significant for mellow tissue such as brain segments. The spatial resolution perpendicular to the slices is much worse with respect to the lateral sub-micrometer resolution. Therefore, we propose as complementary method the synchrotron-radiation-based micro computed tomography (SRµCT), which avoids any kind of preparation artifacts due to sectioning and histological processing and yields true micrometer resolution in the three orthogonal directions. The visualization of soft matter by the use of SRµCT, however, is often based on elaborate staining protocols, since the tissue exhibits (almost) the same x-ray absorption as the surrounding medium. Therefore, it is unexpected that human tissue from the pons and the medulla oblongata in phosphate buffer show several features such as the blood vessels and the inferior olivary nucleus without staining. The value of these tomograms lies especially in the precise non-rigid registration of the different sets of histological slices. Applications of this method to larger pieces of brain tissue, such as the human thalamus are planned in the context of stereotactic functional neurosurgery.