Related Experiment Videos
An efficient tissue classifier for building patient-specific finite element models from X-ray CT images
N Shrinidhi1, D R Haynor, Y Wang
1Department of Electrical Engineering, University of Washington, Seattle 98195, USA.
IEEE Transactions on Bio-Medical Engineering
|March 1, 1996
Summary
We created an efficient semiautomatic tissue classifier for X-ray CT images to build patient-specific finite element models for bioelectric studies. This 3-D approach significantly improves accuracy and reduces manual effort in creating bioelectric models.
Area of Science:
- Biomedical Engineering
- Medical Imaging
- Computational Biology
Background:
- Accurate patient-specific finite element (FE) models are crucial for bioelectric studies.
- Manual segmentation of medical images for FE model creation is time-consuming and labor-intensive.
- Existing methods often lack efficiency and accuracy in tissue classification.
Purpose of the Study:
- To develop an efficient semiautomatic tissue classifier for X-ray computed tomography (CT) images.
- To enable the construction of patient- or animal-specific FE models for bioelectric applications.
- To reduce manual effort and improve accuracy in FE model generation from medical imaging data.
Main Methods:
- A semiautomatic tissue classifier utilizing grayscale histogram and 3-D neighborhood information was developed.
- The classifier was applied to 537 CT images from four pigs for tissue classification.
- Comparison of classification accuracy against manual segmentation by a radiologist was performed.
- The impact of 3-D versus 2-D information on error reduction was evaluated.
Main Results:
- The semiautomatic classifier achieved an average accuracy of 96.5% compared to manual classification.
- The use of 3-D information reduced the classification error rate by 78% compared to 2-D methods.
- FE models generated with minimal or full manual editing showed substantially identical voltage profiles.
- Specific tissue editing (e.g., myocardium) is necessary for accurate voltage gradient and current density calculations.
Conclusions:
- The developed semiautomatic classifier is an efficient tool for creating patient-specific FE models from CT images.
- The 3-D approach significantly enhances classification accuracy and reduces errors.
- The level of manual effort can be adjusted based on specific application requirements, offering flexibility.
- This method facilitates the generation of accurate bioelectric models with reduced manual intervention.