Modelling longitudinal structural change from serial MRI

TL;DRAbstract

Temporal brain changes such as those in development, plasticity, ageing and neurodegeneration are best studied using longitudinal data. For example, the within-subject trajectory of the brain over the lifespan can differ from that inferred purely from cross-sectional data due to effects of birth-year (nutrition, etc.) in the latter. Serial imaging is also important for experimental studies that test the effects of treatments or training (e.g. Draganski et al., 2004). The most efficient methods for processing (e.g. spatially registering) longitudinal images capitalise on their withinsubject nature; however, doing so invokes a risk of bias from asymmetries or inconsistencies (Fox et al., 2011). Temporally-correlated data also require more sophisticated statistical analysis, and unbalanced data (with unequal numbers of time-points and/or uneven intervals) bring additional challenges (Bernal-Rusiel et al., in press). We have recently developed a longitudinal image registration framework (A

Chat with Paper

AI Agents for this Paper