# Development of a Diagnostic Prediction Model for Post‐Stroke Cognitive Impairment in Acute Large Vessel Occlusion Stroke Using Multimodal MRI and PET/CT: A Study Protocol

## TL;DR

This study aims to create a model using MRI and PET/CT imaging and clinical data to predict cognitive impairment after stroke in patients treated with endovascular therapy.

## Contribution

The novelty lies in integrating multimodal imaging and clinical data to predict post-stroke cognitive impairment in a specific stroke subgroup.

## Key findings

- A machine learning model will be developed to predict PSCI using multimodal imaging and clinical data.
- The model will be evaluated using metrics like AUC, precision, and recall.
- Future multicenter validation is needed to confirm generalizability.

## Abstract

Stroke is a leading cause of morbidity and disability worldwide. Post‐stroke cognitive impairment (PSCI) significantly affects long‐term prognosis in acute anterior circulation large‐vessel occlusion stroke (LVO‐AIS). This study aims to develop a PSCI prediction model integrating multimodal imaging, demographic, and clinical data collected during hospitalization.

This single‐center, prospective cohort study will enroll 379 anterior circulation LVO‐AIS patients undergoing emergency endovascular treatment (EVT) within 24 h of symptom onset. Participants will be categorized into PSCI and non‐PSCI groups and followed up at 90 and 180 days post‐procedure. Primary outcomes include Montreal Cognitive Assessment scores at 3 and 6 months, with the modified Rankin Scale as a secondary outcome. Baseline imaging data will be processed using 3D Slicer for MRI and PET/CT standardization, registration, and feature extraction. Machine learning models will be developed using these imaging features combined with demographic and clinical data and evaluated via metrics such as the area under the receiver operating characteristic curve, precision, and recall. Analyses will be conducted in a blinded manner.

This study will develop a PSCI prediction model based on multimodal imaging and clinical data in EVT‐treated LVO‐AIS patients, providing a tool for early diagnosis and personalized interventions. While limited to a single‐center, future multicenter validation is necessary to establish its generalizability and clinical utility.

This study develops a machine learning‐based model to predict post‐stroke cognitive impairment (PSCI) in patients with anterior circulation large‐vessel occlusion who received endovascular treatment (EVT) within 24 h, using multimodal MRI and PET/CT imaging combined with clinical data for early risk stratification.