Related Experiment Video
Updated: Jun 19, 2026
14:08
Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images
Published on: April 13, 2013
Limited view CT reconstruction and segmentation via constrained metric labeling
Vikas Singh1, Lopamudra Mukherjee, Petru M Dinu
1Biostatistics & Medical Informatics and Computer Sciences, UW-Madison.
Computer Vision and Image Understanding : CVIU
|October 6, 2009
Summary
This study introduces a novel discrete optimization framework for limited-view computed tomography (CT) reconstruction and segmentation, crucial for coronary angiography. The method unifies metric labeling and algebraic tomographic reconstruction for improved imaging.
Area of Science:
- Medical Imaging
- Computer Vision
- Optimization
Background:
- Limited-view computed tomography (CT) poses challenges in reconstructing and segmenting volumetric data, particularly in clinical applications like coronary angiography.
- Accurate reconstruction and segmentation are vital for diagnosing cardiovascular conditions.
Purpose of the Study:
- To develop a novel discrete optimization framework for tomographic reconstruction and segmentation using limited projection views.
- To integrate metric labeling classification with classical algebraic tomographic reconstruction (ART) into a unified model.
Main Methods:
- Formulating the limited-view reconstruction and segmentation problem as a constrained metric labeling problem.
- Developing a linear programming framework that combines metric labeling and ART.
- Employing a voxel reassignment strategy to maintain consistency with input reconstruction and ART objectives for volumes with known attenuation coefficients.
Main Results:
- The proposed framework reliably reconstructs and segments CT volumes with multiple contrast objects even with limited projection data.
- Evaluations using cone-beam computed tomography experiments demonstrate the effectiveness of the approach.
Conclusions:
- The unified framework offers a robust solution for limited-view tomographic reconstruction and segmentation.
- This approach has significant potential for improving coronary angiographic imaging and other clinical applications requiring high-resolution volumetric analysis.
