Global tractography

Tractography aims to infer, from a series of diffusion MRI images of the brain, the trajectories of the axonal fibers that generated them; however, as the typical size of the axons is much smaller than the spatial resolution attainable with MRI, this estimation process is ill-posed. The first generation of algorithms approached this reconstruction problem using a greedy line-propagation strategy, where each trajectory is estimated independently of all others and the propagation is guided considering only a portion of the acquired data. The class of algorithms known as “global tractography,” instead, explicitly acknowledges the ill-posed nature of the estimation process and formulates the reconstruction following a global inverse problem perspective, in order to provide higher levels of confidence in the anatomical plausibility of the estimated trajectories as well as other biological properties associated to them. In this chapter, we first define carefully what is meant by “global” within this context, as there has been a sort of gray area around the expression “global tractography” in the literature. Then, we formally define the fundamental paradigm behind all global tractography methods and provide details of the most relevant solutions proposed to implement such paradigm, highlighting their differences from methods presented in previous chapters.