Related Experiment Video
Updated: Jul 13, 2026
09:37
Extracting Metrics for Three-dimensional Root Systems: Volume and Surface Analysis from In-soil X-ray Computed Tomography Data
Published on: April 26, 2016
Combining radiometric and spatial structural information in a new metric for minimal surface segmentation
Olivier Nempont1, Jamal Atif, Elsa Angelini
1Ecole Nationale Supérieure des Télécommunications (GET - Télécom Paris) CNRS UMR 5141 LTCI, Paris, France. name@enst.fr
Information Processing in Medical Imaging : Proceedings of the ... Conference
|July 19, 2007
Summary
This study introduces a novel framework for robust anatomical structure segmentation, improving accuracy even with missing image data. The new metric enhances segmentation of subcortical gray matter structures in MRI scans.
Area of Science:
- Neuroimaging
- Medical Image Analysis
- Computational Anatomy
Background:
- Minimal surface extraction using gradient-based metrics is common for anatomical segmentation.
- This method struggles with weak or missing contour information in medical images.
Purpose of the Study:
- To develop a new framework for defining segmentation metrics robust to missing image information.
- To improve the segmentation of subcortical gray matter structures, particularly the caudate nucleus.
Main Methods:
- A novel framework combines gray-level information and spatial knowledge of cerebral structures into a fuzzy set.
- This fuzzy set guarantees inclusion of object boundaries, from which a robust metric is derived.
- The derived metric is integrated into a minimal surface segmentation framework.
Main Results:
- The proposed metric significantly improves the segmentation of subcortical gray matter structures.
- Quantitative results demonstrate enhanced segmentation of the caudate nucleus in T1 MRI scans.
- The method was validated on 18 normal subjects and 6 pathological cases.
Conclusions:
- The new fuzzy set-based metric framework offers improved robustness for anatomical segmentation.
- This approach addresses limitations of traditional gradient-based methods in cases of incomplete image data.
- The findings are particularly relevant for accurate caudate nucleus segmentation in clinical neuroimaging.
