R. Aranda, G. Rocha, Á. Díaz-Pacheco, M. Álvarez-Carmona
SMCCA (2025). 5
Abstract
This work presents a method to estimate the structure of white matter (axon bundles) by integrating microstructural information through convex optimization. The approach locally validates each segment using a physical diffusion model that assigns weights to possible trajectories, reducing spurious connections from the early stages of the process. The method is evaluated against classical algorithms using metrics such as LiFE, connectivity correlation, and the area under the ROC curve. The results show greater structural coherence and a reduction in false positives, with robust performance under noise. The study demonstrates the feasibility of incorporating microstructural information into the estimations, although it also reveals a higher number of false negatives and a high computational demand.
Abstract This work presents a method to estimate the structure of white matter (axon bundles) by integrating microstructural information through convex optimization. The approach locally [...]