Related Experiment Video
Updated: May 20, 2025
Simultaneous PET/MRI Imaging During Mouse Cerebral Hypoxia-ischemia
Published on: September 20, 2015
Radiomics-based MRI model to predict hypoperfusion in lacunar infarction.
Chia-Peng Chang1, Yen-Chu Huang2, Yuan-Hsiung Tsai3
1Department of Emergency Medicine, Chang Gung Memorial Hospital, Chiayi, Taiwan; Department of Nursing, Chang Gung University of Science and Technology, Chiayi Campus, Chiayi, Taiwan; In-service Master Program in Artificial Intelligence in Medicine, College of Medicine, Taipei Medical University.
Radiomics analysis of MRI scans can predict hypoperfusion in lacunar stroke patients, aiding early intervention. Combining imaging features with clinical data like NIHSS scores improves prediction accuracy for better patient outcomes.
Area of Science:
- Neurology
- Radiology
- Artificial Intelligence
Background:
- Lacunar infarction (a type of ischemic stroke) affects 20-30% of patients with early neurological deterioration.
- Hemodynamic perfusion deficits are implicated in stroke progression and poor prognosis.
- Early prediction of perfusion deficits is crucial for timely treatment and monitoring.
Purpose of the Study:
- To develop a predictive model for hypoperfusion in lacunar stroke.
- To utilize radiomic features from MRI and machine learning for prediction.
- To identify patients at risk of neurological deterioration.
Main Methods:
- Retrospective analysis of 92 lacunar stroke patients (2011-2020).
- Extraction of radiomic features from Diffusion Weighted Imaging (DWI), Apparent Diffusion Coefficient (ADC), and Fluid Attenuated Inversion Recovery (FLAIR) MRI sequences.
- Development of a machine learning model using an 80% training and 20% testing split.
Main Results:
- A model using DWI + FLAIR sequences achieved 84.1% accuracy and 0.92 AUC.
- Significant clinical factors for hypoperfusion included NIHSS scores and infarct size.
- A combined model with nine features (7 radiomic + NIHSS + infarct size) reached 88.9% accuracy and 0.91 AUC in the test set.
Conclusions:
- Radiomics on DWI and FLAIR MRI can predict hypoperfusion in lacunar stroke.
- Integrating stroke volume and NIHSS scores improves predictive model performance.
- Larger-scale studies are needed for validation.
