Related Experiment Video
Updated: Aug 5, 2026
Fiber Connections of the Supplementary Motor Area Revisited: Methodology of Fiber Dissection, DTI, and Three Dimensional Documentation
Published on: May 23, 2017
GEMO: A Deep Learning Method for Brain Fiber Classification and Tract Segmentation Using Geometrical and
Amin Barati Shoorche1, Bahador Makkiabadi1, Parastoo Farnia1
1Department of Medical Physics and Biomedical Engineering, School of Medicine, Tehran University of Medical Science (TUMS), Tehran, Iran (A.B., B.M., P.F.); Research Center for Intelligent Technologies in Medicine (RCITM), Advanced Medical Technologies and Equipment Institute (AMTEI), Tehran University of Medical Sciences (TUMS), Tehran, Iran (A.B., B.M., P.F.).
This study introduces GEMO, a deep learning method that accurately classifies brain fibers using geometrical and morphological features. GEMO improves streamline tractography for better brain structure analysis and neurosurgical planning.
Area of Science:
- Neuroimaging
- Computational Neuroscience
- Artificial Intelligence in Medicine
Background:
- Diffusion-weighted magnetic resonance imaging (dMRI) is crucial for studying brain structure.
- Streamline tractography, enabled by dMRI, is vital for evaluating structural connectivity and surgical planning.
- Accurate localization of brain fibers is essential for enhancing these applications.
Purpose of the Study:
- To propose a supervised deep learning approach named GEMO for fiber classification.
- To leverage geometrical and morphological features alongside convolutional neural network (CNN) extracted features for improved classification performance.
Main Methods:
- Streamlines were converted into 2D color-encoded images using the "xyz2RGB" mapping.
- These images were processed by a CNN for feature extraction and classification.
- The GEMO approach integrated CNN features with geometrical and morphological streamline features.
Main Results:
- Incorporating geometric and morphological features significantly enhanced the deep learning network's accuracy in fiber classification.
- These features alone increased network accuracy to 98.34%.
- GEMO outperformed existing approaches across six segmentation accuracy metrics.
Conclusions:
- GEMO performs streamline classification using individual streamline features (CNN, geometric, morphologic) from whole brain tractograms.
- The method requires no structural images or preprocessing and surpasses state-of-the-art segmentation techniques.
- GEMO facilitates brain structure assessment through automated white matter tract segmentation.
