Related Experiment Video
Updated: Jun 22, 2025
04:25
Author Spotlight: Bridging Gaps in Anatomy and Establishing a Foundation for Algorithmic Studies
Published on: December 15, 2023
2.3K
High-resolution dataset of manual claustrum segmentation.
Adam Coates1,2, Natalia Zaretskaya1,2
1Department of Psychology, University of Graz, Graz, Austria.
Data in Brief
|July 4, 2024
Summary
This study presents the first high-resolution manual segmentation of the entire claustrum, including the challenging ventral "puddles." These detailed labels improve anatomical MRI analysis for researchers studying the human brain.
Area of Science:
- Neuroimaging
- Neuroanatomy
- Brain Mapping
Background:
- The claustrum's thin, sheet-like structure complicates identification in standard anatomical MRI.
- Existing automated and atlas-based segmentation methods often exclude the ventral claustrum, known as "puddles."
Purpose of the Study:
- To create a comprehensive, high-resolution manual segmentation of the entire human claustrum.
- To provide a reliable dataset for improving claustrum localization in in vivo MRI scans.
Main Methods:
- Manual segmentation of the whole claustrum on ultra-high resolution postmortem MRI data.
- Independent labeling by four trainees, with union and Dice coefficient analysis for label correspondence.
- Size measurements in MNI space using oriented bounding box calculations.
Main Results:
- The first manual segmentation labels encompassing both dorsal and ventral claustrum regions at high resolution.
- Validated label correspondence using Dice coefficients between independent raters.
- Provided size metrics for the segmented claustrum in standard MNI space.
Conclusions:
- This dataset offers a significant advancement for claustrum research by including all its parts.
- The high-resolution manual labels can aid in approximating claustrum location in typical in vivo MRI scans.
