Related Experiment Video
Updated: Jul 10, 2026
12:54
Simultaneous Brightfield, Fluorescence, and Optical Coherence Tomographic Imaging of Contracting Cardiac Trabeculae Ex Vivo
Published on: October 2, 2021
Aortic thrombus segmentation using narrow band active contour model
Bipul Das1, Yogish Mallya, Suryanarayanan Srikanth
1Imaging Technology, GE Global Research, Bangalore, India.
Summary
This study introduces a 2D active contour method for segmenting abdominal aortic aneurysm (AAA) thrombus volume in CT scans. The approach effectively addresses segmentation challenges, achieving high accuracy in thrombus volume measurement.
Area of Science:
- Medical Imaging
- Computational Anatomy
Background:
- Accurate segmentation of thrombus volume in abdominal aortic aneurysms (AAA) is crucial for clinical assessment.
- Challenges include poor contrast at boundaries, soft tissue overlap, and artifacts from medical devices and calcifications.
Purpose of the Study:
- To develop and evaluate a novel 2D active contour algorithm for segmenting thrombus in 3D CT images of AAA.
- To improve the accuracy and robustness of thrombus volume segmentation in the presence of image artifacts.
Main Methods:
- Pre-processing involves bone removal and morphological operations to reduce artifacts.
- A manual contour is initialized and propagated across slices using a 2D active contour (snake) model.
- The snake model is driven by an intensity-based objectness measure and local image properties.
Main Results:
- The algorithm was tested on 7 patient CT datasets.
- Segmentation accuracy ranged from 85.08% to 93.16% when compared to radiologist-defined ground truth.
- The method demonstrated effectiveness in handling image noise and artifacts.
Conclusions:
- The proposed 2D active contour approach provides an accurate and reliable method for abdominal aortic aneurysm thrombus segmentation from CT images.
- Pre-processing steps significantly enhance segmentation performance by mitigating artifacts and improving boundary definition.