Related Experiment Videos
Level set based cerebral vasculature segmentation and diameter quantification in CT angiography.
R Manniesing1, B K Velthuis, M S van Leeuwen
1Department of Radiology, Image Sciences Institute, University Medical Center Utrecht, Heidelberglaan 100, Room E01.335, 3584 CX Utrecht, The Netherlands. rashindra@isi.uu.nl
Medical Image Analysis
|November 3, 2005
Summary
This study presents a novel level set method for segmenting cerebral vascular trees in computed tomography angiography (CTA) scans. The automated approach demonstrates high accuracy and reproducibility, outperforming human observers in specific validation tests.
Area of Science:
- Medical Imaging
- Image Segmentation
- Computational Anatomy
Background:
- Cerebral vascular tree segmentation is crucial for diagnosing cerebrovascular diseases.
- Computed tomography angiography (CTA) is a common imaging modality for visualizing cerebral vasculature.
- Accurate and reproducible segmentation methods are needed to improve quantitative analysis.
Purpose of the Study:
- To develop and validate a level set-based method for automated cerebral vascular tree segmentation from CTA data.
- To assess the accuracy and reproducibility of the proposed method compared to expert observers and ground truth.
- To evaluate the method's potential for automated and accurate diameter quantification.
Main Methods:
- A level set-based segmentation approach is employed for cerebral vascular tree extraction.
- The method incorporates bone masking using registered scans and intensity histogram analysis to guide level set evolution.
- Parameter optimization is performed using a training dataset, and validation is conducted on phantom and patient data.
Main Results:
- The level set method achieved similar accuracy to expert observers on phantom data but was unbiased.
- Method reproducibility was slightly superior to inter- and intra-observer variability.
- In patient studies, the method showed agreement with observers, with comparable reproducibility.
Conclusions:
- The presented level set method offers a promising approach for automated cerebral vascular tree segmentation in CTA.
- The method demonstrates high accuracy and reproducibility, making it suitable for quantitative analysis, including diameter quantification.
- This automated technique has the potential to enhance the diagnostic capabilities of CTA in cerebrovascular assessments.