Related Experiment Videos
Localization and segmentation of aortic endografts using marker detection.
Marleen de Bruijne1, Wiro J Niessen, J B Antoine Maintz
1Image Sciences Institute, University Medical Center Utrecht, Room E.01.335, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands.
IEEE Transactions on Medical Imaging
|May 31, 2003
Summary
This study presents an automated method for segmenting bifurcated aortic endografts in CT angiography (CTA) images using radiopaque markers. The approach accurately localizes and segments endografts, achieving high overlap with expert segmentations.
Area of Science:
- Medical Imaging
- Computer-Aided Diagnosis
- Biomedical Engineering
Background:
- Accurate segmentation of aortic endografts in computed tomographic angiography (CTA) is crucial for post-procedural assessment.
- Existing segmentation methods can be time-consuming and subject to inter-observer variability.
Purpose of the Study:
- To develop and evaluate an automated method for the localization and segmentation of bifurcated aortic endografts in CTA images.
- To improve the efficiency and reproducibility of endograft segmentation compared to manual methods.
Main Methods:
- A novel method utilizing radiopaque markers on the endograft for automatic localization and segmentation in CTA images.
- Marker detection employs second-order scaled derivative analysis combined with prior knowledge of graft and marker configuration.
- Three segmentation approaches were tested, with the best involving iterative refinement orthogonal to the estimated graft axis.
Main Results:
- The automated method successfully detected 262 out of 266 markers across ten CTA images.
- The best segmentation approach achieved an average relative volume overlap of 92% with expert segmentations.
- The automated method's volume difference from experts was comparable to inter-expert variability (3.5%).
Conclusions:
- The proposed automated method provides accurate and reproducible segmentation of bifurcated aortic endografts in CTA.
- This technique has the potential to streamline endograft assessment and reduce variability in clinical practice.
- The method's performance is comparable to expert manual segmentation, offering a valuable tool for medical imaging analysis.