Related Experiment Videos
Neural networks for electrical impedance tomography image characterisation.
A S Miller1, B H Blott, T K Hames
1Department of Physics, University of Southampton, UK.
Summary
This study introduces a novel neural network approach for real-time analysis of electrical impedance tomography (EIT) images. The back-projection network identifies key regions in thoracic EIT scans, aiding in the diagnosis of cardiac and pulmonary functions.
Area of Science:
- Medical Imaging
- Biomedical Engineering
- Artificial Intelligence
Background:
- Electrical Impedance Tomography (EIT) systems, like the Southampton system, reconstruct internal conductivity distributions.
- Real-time imaging of cardiac and pulmonary function is crucial for medical diagnosis.
- Current diagnostic methods often lack real-time image analysis capabilities.
Purpose of the Study:
- To apply neural networks for real-time analysis of EIT images.
- To identify regions of interest within thoracic EIT images for functional assessment.
- To develop automated real-time activity plots for specific organs using EIT data.
Main Methods:
- Utilized a Southampton EIT system with Sheffield data acquisition and a PC-based transputer card.
- Employed a back-projection neural network for image analysis.
- Focused on identifying characteristic regions and their temporal activity, such as ventricular ejection.
Main Results:
- The back-projection network successfully identified characteristic regions within EIT images.
- Automated real-time activity plots were facilitated by defining the extent of specific organs.
- Demonstrated the potential for real-time neural network application in medical imaging.
Conclusions:
- Neural networks offer a novel and effective method for real-time EIT image analysis.
- This approach can automate the identification of functional regions in cardiac and pulmonary imaging.
- The study paves the way for advanced, AI-driven diagnostic tools in medical imaging.