Related Experiment Video
Updated: Mar 11, 2026
06:08
Quantitative Visualization and Detection of Skin Cancer Using Dynamic Thermal Imaging
Published on: May 5, 2011
17.3K
Parameter estimation of breast tumour using dynamic neural network from thermal pattern.
Elham Saniei1, Saeed Setayeshi1, Mohammad Esmaeil Akbari2
1Energy Engineering and Physics Faculty, Amirkabir University of Technology, Tehran, Iran.
Journal of Advanced Research
|November 19, 2016
Summary
This study introduces a novel method using thermal imaging and dynamic neural networks to estimate tumor characteristics like depth and size. The approach shows promise for non-invasively retrieving crucial tumor parameters from breast thermal images.
Area of Science:
- Biomedical Engineering
- Medical Imaging
- Computational Biology
Background:
- Accurate tumor characterization is crucial for effective cancer treatment planning.
- Non-invasive methods for estimating tumor depth, size, and metabolic heat generation are highly desirable.
Purpose of the Study:
- To develop and validate a novel computational approach for estimating tumor parameters using breast thermal imaging.
- To assess the feasibility of using dynamic neural networks (DNNs) combined with finite element modeling for this purpose.
Main Methods:
- A two-step approach involving a forward and an inverse problem was employed.
- The forward step utilized a finite element model solving Pennes' bio-heat equation to simulate temperature distributions.
- The inverse step employed a trained DNN to estimate subsurface temperature profiles from surface temperature data.
Main Results:
- The finite element model and the DNN produced comparable temperature distribution results.
- The trained DNN successfully estimated depth temperature distribution from thermal image surface profiles.
- Tumor parameters were accurately retrieved from the estimated depth temperature distribution.
Conclusions:
- The proposed method demonstrates a promising non-invasive approach for retrieving critical tumor parameters.
- Dynamic neural networks combined with thermal imaging offer a viable tool for enhanced tumor assessment.
