Related Experiment Video
Updated: Oct 18, 2025
14:27
Identification of Disease-related Spatial Covariance Patterns using Neuroimaging Data
Published on: June 26, 2013
15.9K
DEEP NETWORK-BASED FEATURE SELECTION FOR IMAGING GENETICS: APPLICATION TO IDENTIFYING BIOMARKERS FOR PARKINSON'S
Mansu Kim1,2,3, Ji Hye Won1,2, Jisu Hong1,2
1Department of Electrical and Computer Engineering, Sungkyunkwan University, Korea.
Proceedings. IEEE International Symposium on Biomedical Imaging
|October 1, 2021
Summary
This study introduces a novel deep learning model for imaging genetics, outperforming traditional methods. The model effectively identifies genetic features linked to brain imaging, potentially uncovering new biomarkers for brain disorders.
Area of Science:
- Neuroimaging
- Genetics
- Computational Biology
- Machine Learning
Background:
- Imaging genetics links brain imaging data with genetic information.
- Current sparse models (e.g., SCCA) primarily capture linear relationships.
- Non-linear, high-level relationships in imaging genetics remain underexplored.
Purpose of the Study:
- To propose a deep learning model for imaging genetics.
- To identify genetic features that effectively explain imaging features.
- To address limitations of linear models in capturing complex relationships.
Main Methods:
- Development of a novel deep learning framework.
- Application of the model to both simulated and real-world datasets.
- Comparison with sparse canonical correlation analysis (SCCA).
Main Results:
- The deep learning model demonstrated superior performance compared to SCCA.
- The method robustly selected important genetic features.
- Empirical validation on diverse datasets confirmed efficacy.
Conclusions:
- Deep learning offers a powerful approach for imaging genetics.
- The proposed model can uncover non-linear imaging-genetics associations.
- This method holds potential for discovering novel biomarkers for brain disorders.
