Brain Tortuosity as Biomarker to Classify Mild Cognitive Impairment and Control Subjects

Abstract

Mild cognitive impairment (MCI) is an abnormal deterioration of cognitive functions, whose prevalence is considerable in adults older than 65 years old. Several of these cases will convert to Alzheimer’s disease and therefore, MCI’s simple, proper and opportune diagnosis continues to be a research field with great impact in public health. In this paper we propose tortuosity, which is defined as a shape measure that has been applied to quantify morphological changes in several anatomical structures, as a potential biomarker sensitive enough to depict early brain changes that appear in MCI subjects in comparison with healthy controls (HC). Also, a random forest (RF) classification strategy was implemented to discriminate between MCI and HC populations. A training population selected from the ADNI database and a test group of 21 mexican subjects were analyzed. Statistical analysis showed significant differences (p < 0.05) in tortuosity indices determined for MCI vs HC populations in most of the measured cortical structures. Classification rates increased by 6.7% during training and 4.77% during the test stage, when incorporating tortuosity to other image-based features set. This suggests that tortuosity is a promising morphological parameter to be considered for early stages of Alzheimer disease (AD) and that, combined with an RF classifier, it can adequately separate HC and MCI subjects.