WP2: Multimodal atlases
Scientific background
Computational neuroanatomy
Dr Anneke Alkemade and her team specialize in integrative computational
neuroanatomical research at the University of Amsterdam. In recent years, their highly interdisciplinary work has included the establishment of the Amsterdam Ultra-high field Adult Lifespan Database (AHEAD) and the creation of probabilistic atlases of the human subcortex using both post-mortem and in vivo 7T MRI. Beyond subcortical mapping, the group has developed image analysis pipelines for 3D histological reconstructions by integrating histochemical data with qMRI data. In a collaboration with Partner 1, these unique, openly available datasets have enabled the precise parcellation of nigrosomes in the SN, allowing the correlation of MRI features with the underlying 3D anatomy at an unprecedented precision, challenging the current conventional MRI-based nigrosome markers found in widely used radiology textbooks. Notably, this project will be carried out in close collaboration with Dr Pierre-Louis Bazin (Full brain picture Analytics), who has developed image processing tools tailored to ultra-high resolution MR images based on multiple contrasts. Together, they recently extended a fully automated method for parcellation of subcortical nuclei based on the 7T ultra-high resolution multiparametric imaging
developed by Weiskopf’s team and translated it to imaging data acquired at 3T clinical field strength by Partners 4’s team.
Working plan and methodology
Subcortical atlasing
We will enrich our subcortical atlases by adding increasingly smaller nuclei and substructures to our efforts, focusing on the nigrosomes of the SN, LC, and hypothalamic nuclei. This goal will be achieved by further integration of data obtained from in vivo and post mortem studies. In our earlier studies, we have laid the groundwork for the translation of microscopy results to in vivo MRI through the use of post mortem MRI. These efforts now allow us to translate delineations made on histological samples to 3D MRI contrasts obtained in the same post mortem specimens. These datasets serve as a bridge, and allow us to register detailed delineations of small structures with limited or no MRI contrast, to MRI reference space. As a result, this anatomical data can be used for unique in vivo MRI analyses of 7T and, more recently, 3T data (such as Partner 4’s data). As a proof of concept, our combined data set allowed us to identify 12 separate nuclei in the thalamus and five nigrosomes within substantia nigra. However, human whole-brain specimens are not readily available, and the processing of a single specimen represents a scientific tour de force. Thus, interindividual variability is not captured. Smaller tissue specimens, containing only parts of the brain, have a higher availability; we will therefore develop a joint microscopy and MRI pipeline that combines data from smaller tissue blocks in 3D and aligns it to whole-brain space, to increase the number of processed samples. We will develop and validate the method on the hypothalamus, a small but very complex region where we expect early disturbances with sleep disorders and PD (see WP3). These microscopy-derived atlases will complement the MASSP subcortical parcellation algorithm, by providing atlas driven predictions on the location of small subcortical nuclei that do not provide sufficient MRI contrast to allow their in vivo visualization and delineations.
To achieve these goals we will improve and extend our existing tissue and data processing pipelines. This work is technically challenging, and we will therefore attract a postdoc with prior experience in this field. Following the JPND collaboration model, detailed post-mortem MRI will be performed at Partners’ 1 and 3 sites, and subsequent tissue processing will be performed at the University of Amsterdam. These efforts will be integrated with further improvements of the MASSP algorithm integrating post mortem and in vivo atlasing efforts, and applied to our growing database of ultra-high field and clinical MRI scans from the consortium partners.
Team – University of Amsterdam
Dr Anneke Alkemade
Team - Full brain picture Analytics
