# An Integrated QSM‐Radiomics Nomogram With Clinical and Imaging Markers for Stratifying Cognitive Impairment in Hypertension

## TL;DR

This study creates a new tool combining brain imaging and clinical data to better identify cognitive issues in people with high blood pressure.

## Contribution

The study introduces a novel integrated QSM-radiomics nomogram for cognitive impairment stratification in hypertension.

## Key findings

- The multi-region radiomics model outperformed single-parameter models in diagnosing cognitive impairment.
- The combined model achieved high accuracy in both training and validation cohorts (AUCs of 0.860 and 0.872).
- Decision curve analysis confirmed the clinical utility of the nomogram for risk stratification.

## Abstract

Hypertensive cognitive impairment is associated with increased iron deposition in deep gray matter. This study aimed to evaluate the potential clinical application of quantitative susceptibility mapping (QSM)‐based radiomics for stratifying cognitive impairment in hypertensive patients.

We prospectively enrolled 178 hypertensive patients who underwent QSM examination and categorized them into cognitive impairment group and normal cognition group. The workflow included: (1) Precise 3D segmentation and susceptibility quantification of the basal ganglia; (2) Initial radiomics feature extraction from the target regions, followed by optimal feature selection via variance thresholding, maximum relevance minimum redundancy (mRMR) algorithm, and LASSO regression; (3) Construction of a multiparametric model integrating radiomics scores and clinical risk factors. The model's performance was assessed by receiver operating characteristic (ROC) curve and its clinical utility for stratification was further evaluated using decision curve analysis (DCA).

The multi‐region radiomics model demonstrated superior diagnostic performance in the training cohort (AUC = 0.812), significantly outperforming the QSM and WMH models (AUCs = 0.620 and 0.688, respectively). In the validation cohort, clinically meaningful improvements were observed (ΔAUC = 0.117 and 0.136, respectively). The combined model, which integrated Radscore, susceptibility values, WMH scores, age, and education level, achieved the highest discriminability in both the training (AUC = 0.860) and validation cohorts (AUC = 0.872). DCA further indicated that the nomogram derived from the combined model provided the greatest net clinical benefit for stratifying hypertensive cognitive impairment.

A nomogram integrating QSM‐based radiomics with clinical and imaging markers accurately stratified hypertensive cognitive impairment, offering an objective tool for early risk assessment.

This study developed an integrated nomogram combining QSM‐based radiomics, brain iron deposition, white matter hyperintensity, and clinical markers. This model demonstrated superior performance in stratifying hypertensive cognitive impairment compared to traditional single‐parameter models. It offers a valuable non‐invasive tool for early risk assessment and personalized intervention.