Understanding the Early Development of Human Brain: Quantitative Morphometry Analysis of Magnetic Resonance Imaging of the Fetal Brain
In vivo fetal magnetic resonance (MR) imaging provides a unique approach for the study of early human brain development. In-utero cerebral morphometry can potentially be used as a marker of the cerebral maturation and help to distinguish between normal and abnormal development in ambiguous situation. We are mainly interested on the third semester of gestation where the gyral process starts. The main focus of this research project is thus to investigate quantitative morphometric measures (like for instance brain tissue volumes, cortical thickness or gyrification index) for the study of brain development.
The technical challenges are determined by the clinical image data that we want to segment. Fast acquisitions are required to avoid fetal motion, thus, our images are of low spatial resolution (1.09 mm in plane and 5.4 mm between planes). This low spatial resolution increase the partial volume (PV) effect, that is, the presence of several tissues within a same image voxel. Moreover, brain tissues have a non-homogeneous appearance due to MR artefacts but also to the natural myelination and cortical maturation.
In order to make the segmentation problem less challenging, prior information could be included as for instance making use of anatomical atlas priors to segment the different tissue or regions. However, the use of such atlases faces, in our opinion, a risk of circularity: each brain will be analyzed / deformed using the template of its biological age, potentially biasing the effective developmental delay.
In terms of image processing, we address the problem of brain tissue segmentation of the in-utero fetal brain without atlas priors. Specifically, we aim at segmenting the brain tissues (cerebrospinal fluid, central gray matter, cortical gray matter and white matter) by means of powerful image analysis tools: statistical classification in Bayesian framework, local spatial priors by Markov Random Field models, Active Constour segmentation and deformable models. Our results demonstrate the feasibility of quantitative studies of the fetal brain during the third trimester of gestation based on clinically acquired images. As our reconstruction's and segmentation's pipeline do not use any anatomical prior, our method can potentially answer clinically relevant question such as the true developmental age of the fetus without being biased by its biological age.
Collaborators This work is in collaboration with Dr. M. Schaer and Prof. S. Eliez (HUG-UniGe, Switzerland), and with Dr. L. Guibaud (Hopital Femme Mère Enfant, University Hospital Lyon, France).