Related Experiment Video
Updated: May 3, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.6K
Robust Non-Local Multi-Atlas Segmentation of the Optic Nerve
Andrew J Asman1, Michael P Delisi2, Mawn3
1Electrical Engineering, Vanderbilt University, Nashville, TN, USA 37235.
Summary
We developed a new automated framework for segmenting the optic nerve in medical images. This method improves accuracy and robustness, overcoming challenges in anatomical and imaging variability for better disease understanding.
Area of Science:
- Medical Imaging Analysis
- Neuroscience
- Radiology
Background:
- Accurate segmentation of the optic nerve is crucial for understanding diseases like glaucoma and multiple sclerosis.
- Current automated methods often fail due to anatomical and imaging variability, requiring manual intervention.
Purpose of the Study:
- To propose a robust, fully-automated framework for optic nerve segmentation.
- To improve the accuracy and reliability of optic nerve segmentation in medical imaging.
Main Methods:
- Developed a robust registration procedure for consistent alignment despite varying image resolutions and fields-of-view.
- Applied a non-local label fusion algorithm to correct for minor registration errors.
Main Results:
- The proposed framework achieved accurate optic nerve segmentations on a dataset of 31 computed tomography (CT) brain scans.
- Demonstrated robust registration of optic nerve anatomy across diverse imaging data.
- The non-local label fusion algorithm outperformed state-of-the-art methods.
Conclusions:
- The developed framework offers a significant advancement in automated optic nerve segmentation.
- This automated approach can aid in the clinical and scientific study of optic nerve pathologies.
- Improved segmentation accuracy supports better diagnosis and monitoring of related neurological disorders.
