Project title
Multimodal imaging of white matter fiber crossing geometry in the brain: towards the next generation of tractography algorithms
Summary
Diffusion MRI (dMRI) is a non-invasive imaging technique used to visualize white matter fibers in the brain. Using tractography algorithms, a map of the brain's structural connectivity is obtained. This provides an invaluable model for studying brain organization. However, most tractography algorithms are based on assumptions about white matter microstructure which, to date, remain difficult to validate experimentally. In particular, what remains unknown is a clear description of how white matter crosses within and between fiber clusters. How the organization of these crossings on a microscopic scale affects tractography algorithms on a macroscopic scale is also poorly understood. An experiment carried out at a recent international symposium concluded that all tracing algorithms managed to detect 90% of true fibers within a synthetic dataset, but each method generated between 4 and 5 times more false detections than true fibers. This was largely attributed to a lack of information about the geometry of fiber crossings. This problem is all the more important considering that, for the resolution of MRI scans used in clinical practice, around 90% of voxels contain multiple fiber orientations. There is therefore an urgent need for a method to validate tractography algorithms.
The overall aim of this project is to combine images of the brain acquired by dMRI and serial histology to study the geometry of white matter fiber crossings and their impact on tractography algorithms. Serial histology is a recent technique in neurophotonics that combines a vibrating blade with a microscope. The blade cuts thin slices of tissue to reveal new layers of brain to be imaged with the microscope. This process is repeated until the entire brain has been imaged. Then, the thousands of images acquired are assembled into a single 3D volume that is precisely aligned with the dMRI data. In a previous research project, we built such an automated microscope, which can acquire whole rodent brains with micrometric resolution in less than a day.
By combining dMRI and optical data, we will develop a new tractography algorithm that will be more reliable in areas where fibers cross. This project will require work in optics, image processing and algorithms. The imaging and analysis pipeline will be developed using rodent brains, and will then be used to study human brain samples. The expected result is an unprecedented correlation between the dMRI signal and the underlying myelin mesostructure in the human brain. Given the growing importance of structural connectivity for understanding brain function and disease, our work will provide essential data to validate tractography algorithms and contribute to a better understanding of the tissue characteristics underlying the dMRI signal.
Co-investigators & Collaborators
- Maxime Descoteaux (Université de Sherbrooke)
- Frédéric Lesage (École Polytechnique de Montréal)
- Laurent Petit (Bordeaux University)
- Josefina Maranzano (Université du Québec à Trois-Rivières)
Financing
- "Réseau de bio-imagerie du Québec(RBIQ) pilot project
- "Team research project" from the Fonds de recherche du Québec - Nature et technologies (FRQNT)