Currently there are no publicly available segmentation tools and atlases that would automatically label human brain MRI images and characterize neuroanatomy in the age range of 0 to 5 years. The development of fully automated, highly accurate and robust post-processing tools applicable to this population has been significantly lagging compared to those introduced for adults given a number of factors - the difficulty of obtaining data from non-compliant neonates and toddlers, ethical issues regarding research imaging of subjects who cannot consent, and critically, the rapid change in contrast and geometry displayed during development in infants. Additionally, the majority of the existing imaging data are obtained from clinical subjects (for example, for premature borns), that cannot be used to characterize normal variation in the infant population and thus to produce a baseline for further quantitative comparison between healthy and disease. We believe that it is crucial that atlases and image processing tools are first adapted and introduced for a healthy control population. During the proposed project, we intend to develop a FreeSurfer-compatible pipeline that, when used with standard clinical MRI acquisitions, will compute cortical and subcortical segmentation maps and create topologically correct and geometrically accurate white matter and pial surfaces. Additionally, such surfaces will be used to drive novel pair- and group-wise registration methods to encode spatial correspondences and age-related information in the form of 2D and 3D atlases. This work will be performed at the MGH/Harvard/MIT Martinos Center for Biomedical Imaging in collaboration with the Boston Children's Hospital, where we will take advantage of the cutting-edge imaging facilities as well as leading imaging and clinical expertise.
Currently there are no publicly available segmentation tools and atlases that would automatically label human brain MRI images and characterize neuroanatomy in the age range of 0 to 5 years. We intend to develop a FreeSurfer-compatible pipeline that, when used with standard clinical MRI images, will compute cortical and subcortical segmentation maps and create topologically correct and geometrically accurate white matter and pial surfaces. Additionally, such surfaces will be used to drive novel pair- and group-wise registration methods to encode spatial correspondences and age-related information in the form of 2D and 3D atlases. These tools will finally all assist with the characterization of healthy control pediatric neuroanatomical morphometry.
