Related Experiment Video
Updated: Jul 21, 2025
Fiber Connections of the Supplementary Motor Area Revisited: Methodology of Fiber Dissection, DTI, and Three Dimensional Documentation
Published on: May 23, 2017
FIESTA: Autoencoders for accurate fiber segmentation in tractography.
Félix Dumais1, Jon Haitz Legarreta2, Carl Lemaire3
1Sherbrooke Connectivity Imaging Lab (SCIL), Department of Computer Science, Université de Sherbrooke, Canada; Videos & Images Theory and Analytics Lab (VITAL), Department of Computer Science, Université de Sherbrooke, Canada.
FIESTA (FIbEr Segmentation in Tractography using Autoencoders) offers automated white matter bundle segmentation for brain connectivity studies. This deep learning approach enhances tractography reliability and coverage, outperforming existing methods.
Area of Science:
- Neuroimaging
- Computational Neuroscience
- Medical Image Analysis
Background:
- White matter bundle segmentation is crucial for understanding brain structural connectivity.
- Current tractography methods face challenges in accurately segmenting and recovering all white matter bundles.
Purpose of the Study:
- To present FIESTA (FIbEr Segmentation in Tractography using Autoencoders), a novel automated pipeline for white matter bundle segmentation.
- To improve the reliability, robustness, and coverage of bundle segmentation in diffusion MRI tractography.
Main Methods:
- Utilized deep autoencoders with contrastive learning to model a latent space of streamlines.
- Employed generative sampling and latent space seeding for recovering hard-to-track bundles.
- Segmented tractograms based on autoencoder latent distance to an atlas of bundles in MNI space.
Main Results:
- FIESTA demonstrated improved intra-subject bundle reliability by generating anatomically correct, novel streamlines.
- The method showed superior performance compared to state-of-the-art automated virtual dissection techniques.
- Achieved enhanced spatial coverage and recovery of challenging white matter bundles.
Conclusions:
- FIESTA provides a reliable, automated, and adaptable framework for white matter bundle segmentation.
- The pipeline enhances the practicality and usability of tractography analysis for neurological disorders, neurosurgery, and aging research.
- Offers flexible transition between different anatomical bundle definitions with minimal calibration.
