Lecture: Longitudinal Image Analysis to Meet Clinical Needs
Longitudinal Image Analysis to Meet Clinical Needs
Guido GERIG, PhD
Institute Professor, IEEE Fellow, AIMBE Fellow
NYU Tandon School of Engineering
Computer Science and Engineering Department, Brooklyn, NY, USA
https://engineering.nyu.edu/faculty/guido-gerig
Abstract: Monitoring of individual disease progression is a fundamental clinical task. Related to imaging data, capturing temporal changes involves acquisition of repeated images which in turn enable analysis of dynamic processes related to development, aging, degeneration, recovery or prediction – information which is not available from single snapshots in time. Image processing of temporal series of 3-D data embedding time-varying anatomical objects and functional measures requires methods and tools that make use of the inherent correlation of repeated image acquisitions. We will discuss crucial aspects of longitudinal imaging such as image harmonization, image curation and synthesis, and longitudinal modeling and segmentation. We will demonstrate that statistical concepts of longitudinal data analysis such as linear and nonlinear mixed-effect modeling can be extended to structures and shapes modeled from longitudinal image data, ranging from modeling of changes of shape or image contrast up to orientation distribution functions (ODFs) as in diffusion MRI. Potentially relevant to clinical studies, we will also show that embedding of subject’s covariates such as sex and diagnostic scores into longitudinal image and shape analysis may lead to improved interpretability of findings. The talk will cover recent learning-based methods and present results from ongoing clinical studies such as analysis of early brain growth in subjects at risk for autism and those with Down’s syndrome, analysis of neurodegeneration in Huntington's disease, and quantitative assessment of progression of glaucoma from OCT imaging.
