sMRI-based Brain Age Estimation in MCI using Persistent Homology

TL;DR

This study explores brain age estimation and MCI classification using persistent homology Betti-curve features derived from 3D structural MRI data. By applying voxel-level filtration to GM, WM, and CSF and extracting time- and frequency-domain descriptors from Betti curves, the authors train a random forest for age prediction and an SVM for HC vs MCI discrimination, achieving up to 80% accuracy with dimension-1 features. The analysis reveals distinct correlations between clinical features and either predicted brain age or chronological age, enabling differentiation between normal and pathological aging. Despite a small sample size of 100 scans, the work demonstrates a non-invasive framework with potential biomarkers for early detection and monitoring of cognitive decline, and it lays the groundwork for larger, longitudinal studies.

Abstract

In this study, we propose the use of persistent homology -- specifically Betti curves for brain age prediction and for distinguishing between healthy and pathological aging. The proposed framework is applied to 100 structural MRI scans from the publicly available ADNI dataset. Our results indicate that Betti curve features, particularly those from dimension-1 (connected components) and dimension-2 (1D holes), effectively capture structural brain alterations associated with aging. Furthermore, clinical features are grouped into three categories based on their correlation, or lack thereof, with (i) predicted brain age and (ii) chronological age. The findings demonstrate that this approach successfully differentiates normal from pathological aging and provides a novel framework for understanding how structural brain changes relate to cognitive impairment. The proposed method serves as a foundation for developing potential biomarkers for early detection and monitoring of cognitive decline.

Figures (5)

• Figure 1: The first, second, and third columns display segmented MRI images of gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) for a mid-slice of a representative subject. The top row shows images for healthy controls (HC) in Figures (A) GM, (B) WM, and (C) CSF, while the bottom row presents the same for individuals with mild cognitive impairment (MCI) in Figures (D) GM, (E) WM, and (F) CSF.
• Figure 2: Betti curves generated from GM (col-1 and 2) and WM (col-3 and 4) segmentation using persistent homology for one representative HC (blue) and MCI (green) subject. These curves illustrate the evolution of topological features (e.g. dim-1 or loops) across different filtration scales, capturing structural differences in brain tissues between healthy and cognitively impaired subjects.
• Figure 3: Correlation plot comparing the predicted brain age using features derived from Betti curves against the actual chronological age for both HC (left) and MCI (right) subjects.
• Figure 4: Receiver Operating Characteristic (ROC) curves showing the performance of SVM-based classification between healthy controls (HC) and individuals with mild cognitive impairment (MCI) using dimension-1 Betti curve features from different brain tissue regions GM, WM, CSF.
• Figure 5: Visualization of differences in clinical imaging features between healthy aging and pathological aging (MCI). The left-side column illustrates clinical features that correlate significantly with chronological age but not with predicted brain age, representing typical structural changes due to healthy aging. The right-side column displays features significantly correlated with predicted brain age but not with chronological age in MCI subjects, highlighting early structural abnormalities indicative of pathological aging and cognitive decline.