Improving MRI-based analysis of brain structural changes in patients with hypertension via a privileged information
Bo Peng1, Xinying Yu2, Xinwei Ma3
1Suzhou Institute of Biomedical Engineering and Technology, Chinese Academy of Sciences, Suzhou, China; Suzhou Key Laboratory of Medical and Health Information Technology, Suzhou, China; Jinan Guoke Medical Engineering Technology Development Co., LTD, Jinan, China.
Methods (San Diego, Calif.)
|July 12, 2021
Summary
This study introduces a machine learning method using kernel ridge regression and privileged information to analyze brain changes in hypertension patients. The approach effectively uses single features for diagnosis, aiding clinical judgment from MRI images.
Area of Science:
- Neuroimaging
- Machine Learning
- Cardiovascular Disease
Background:
- Hypertension alters brain structure and function, with varying effects based on blood pressure levels.
- Analyzing these brain changes requires sophisticated machine learning methods.
- Extracting multiple features for analysis is computationally complex.
Purpose of the Study:
- To develop a machine learning framework for analyzing brain structure changes in hypertension patients.
- To enable a single feature to achieve diagnostic accuracy comparable to multiple features.
- To improve the generalization performance of classifiers in analyzing brain MRI data.
Main Methods:
- Proposed a multi-kernel Kernel Ridge Regression (KRR) framework incorporating privileged information (PI).
- Utilized PI to transfer knowledge from multiple feature types to a single primary feature.
- Integrated self-paced learning for optimized sample selection during classifier training.
Main Results:
- The proposed method effectively leverages information from various features for improved classification.
- Achieved enhanced classification performance in analyzing brain structure changes related to hypertension.
- Demonstrated the utility of self-paced learning-based KRR for hypertension-related brain analysis.
Conclusions:
- The developed framework facilitates the analysis of brain structural changes in patients with different blood pressure levels.
- Discriminative features identified may assist clinicians in assessing hypertension severity from brain MRI.
- This approach offers a more efficient method for leveraging complex neuroimaging data in hypertension research.
