A genetic algorithm-based model for longitudinal changes detection in white matter fiber-bundles of patient with multiple sclerosis

Abstract

Analysis of white matter (WM) tissue is essential to understand the mechanisms of neurodegenerative pathologies like multiple sclerosis (MS). Recently longitudinal studies started to show how the temporal component is important to investigate temporal diffuse effects of neurodegenerative pathologies. Diffusion tensor imaging (DTI) constitutes one of the most sensitive techniques for the detection and characterization of brain related pathological processes and allows also the reconstruction of WM fibers. The analysis of spatial and temporal pathological changes along the fibers are thus possible by merging quantitative maps with structural information provided by DTI. In this work, we present a new genetic algorithm (GA) based method to analyze longitudinal changes occurring along WM fiber-bundles. In the first part of this paper, we describe the data processing pipeline, including data registration and fiber tract post-processing. In the second part, we focus our attention to the description of our GA model. In the last part, we show the tests we performed on simulated and real MS longitudinal data. Our method reached a high level of precision, recall and F-Measure in the detection of longitudinal pathological alterations occurring along different WM fiber-bundles.

Keywords: Diffusion tensor imaging; Genetic algorithm; Longitudinal study; MRI; Multiple sclerosis; White matter fiber-bundles.