# Association between white matter structural damage and cognitive impairment in patients with cerebral small vessel disease based on TBSS technology

## TL;DR

This study shows that specific white matter damage in the brain can predict cognitive impairment in patients with cerebral small vessel disease.

## Contribution

A multivariate model using DTI metrics effectively identifies cognitive impairment in CSVD patients.

## Key findings

- DTI metrics from specific white matter tracts correlate with cognitive impairment in CSVD patients.
- A multivariate model based on these metrics achieved a testing AUC of 0.809 in predicting cognitive impairment.
- Calibration and decision curve analysis confirmed the model's clinical utility.

## Abstract

Cognitive impairment in patients with cerebral small vessel disease (CSVD) is closely associated with white matter injury. This study aims to evaluate whether diffusion tensor imaging (DTI) metrics can predict the risk of cognitive impairment in CSVD patients.

We retrospectively analyzed data from 54 CSVD patients, classified into a cognitive impairment group (CI, n = 25) and a non-cognitive impairment group (NCI, n = 29). Using tract-based spatial statistics (TBSS), we computed fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) across 48 major white matter tracts. Significant DTI metrics identified by univariate logistic regression were used to construct a multivariate logistic regression model. Model performance was evaluated via 5-fold cross-validation based on the area under the ROC curve (AUC), calibration curves, and decision curve analysis.

Several DTI metrics showed significant correlations with cognitive impairment, including FA (fornix, left corticospinal tract, bilateral medial lemniscus/inferior cerebellar peduncle, left cerebral peduncle, right cingulum hippocampus), MD (right superior cerebellar peduncle, left cerebral peduncle), and RD (bilateral medial lemniscus, right inferior/superior cerebellar peduncle, left cerebral peduncle, right external capsule, cingulum hippocampus). The multivariate model constructed based on these metrics demonstrated the best predictive performance, with a mean training AUC of 0.940 and testing AUC of 0.809. The calibration curves showed good agreement between predicted and observed outcomes, and decision curve analysis confirmed the clinical utility of the model.

The multivariate logistic regression model incorporating DTI metrics can effectively identify cognitive impairment in CSVD patients. This study establishes a link between damage in specific white matter tracts and cognitive dysfunction, providing a practical tool for assessing the risk of cognitive impairment in clinical settings.