Evaluation of Shape Variations in Structural MR Images of Fornix in Normal, MCI, and AD Subjects Using Pseudo-Zernike Moments

Alzheimer's disease (AD) is a degenerative disorder of the brain that affects elderly individuals, leading to cognitive decline and memory loss. Mild cognitive impairment (MCI) is a transition stage between normal cognition (NC) and AD. Early detection of MCI is crucial since it allows for timely intervention to delay AD progression. The onset of AD is associated with tissue alterations in the fornix, a white matter region of the brain responsible for cognition, learning, memory consolidation, and attention. In this study, fornix morphometrics in MCI and AD are characterized using structural magnetic resonance (sMR) brain images and pseudo-Zernike moment (PZM) features. For this study, a publicly available database is used. Initially, a standard pipeline is used to pre-process the sMR brain images, followed by segmentation of the fornix structure using the level set without re-initialization (LSWR) algorithm. Subsequently, 64 PZMs are computed from the fornix region. Statistical tests such as the Kolmogorov-Smirnov test, Student's t-test, Wilcoxon-Mann-Whitney test, and one-way ANOVA are employed to identify significant features and machine learning algorithms also performed for binary classification. The outcomes revealed that the LSWR algorithm segmented the fornix structure at an accuracy of 99%. The PZM features exhibited statistical significance (p< 0.05) in distinguishing MCI and AD, emphasizing their effectiveness in capturing fornix shape variations. The proposed approach employed in this study emphasizes the clinical relevance in differentiating MCI from NC and AD subjects.