Three-Dimensional Thermal Tomography with Physics-Informed Neural Networks

<b>Background</b>: Accurate reconstruction of internal temperature fields from surface temperature data is critical for applications such as non-invasive thermal imaging, particularly in scenarios involving small temperature gradients, like those in the human body. <b>Methods</b...

Full description

Saved in:
Bibliographic Details
Main Authors: Theodoros Leontiou, Anna Frixou, Marios Charalambides, Efstathios Stiliaris, Costas N. Papanicolas, Sofia Nikolaidou, Antonis Papadakis
Format: Article
Language:English
Published: MDPI AG 2024-11-01
Series:Tomography
Subjects:
Online Access:https://www.mdpi.com/2379-139X/10/12/140
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846102554244546560
author Theodoros Leontiou
Anna Frixou
Marios Charalambides
Efstathios Stiliaris
Costas N. Papanicolas
Sofia Nikolaidou
Antonis Papadakis
author_facet Theodoros Leontiou
Anna Frixou
Marios Charalambides
Efstathios Stiliaris
Costas N. Papanicolas
Sofia Nikolaidou
Antonis Papadakis
author_sort Theodoros Leontiou
collection DOAJ
description <b>Background</b>: Accurate reconstruction of internal temperature fields from surface temperature data is critical for applications such as non-invasive thermal imaging, particularly in scenarios involving small temperature gradients, like those in the human body. <b>Methods</b>: In this study, we employed 3D convolutional neural networks (CNNs) to predict internal temperature fields. The network’s performance was evaluated under both ideal and non-ideal conditions, incorporating noise and background temperature variations. A physics-informed loss function embedding the heat equation was used in conjunction with statistical uncertainty during training to simulate realistic scenarios. <b>Results</b>: The CNN achieved high accuracy for small phantoms (e.g., 10 cm in diameter). However, under non-ideal conditions, the network’s predictive capacity diminished in larger domains, particularly in regions distant from the surface. The introduction of physical constraints in the training processes improved the model’s robustness in noisy environments, enabling accurate reconstruction of hot-spots in deeper regions where traditional CNNs struggled. <b>Conclusions</b>: Combining deep learning with physical constraints provides a robust framework for non-invasive thermal imaging and other applications requiring high-precision temperature field reconstruction, particularly under non-ideal conditions.
format Article
id doaj-art-337df041f7f542e8aa54218daf28e75e
institution Kabale University
issn 2379-1381
2379-139X
language English
publishDate 2024-11-01
publisher MDPI AG
record_format Article
series Tomography
spelling doaj-art-337df041f7f542e8aa54218daf28e75e2024-12-27T14:56:22ZengMDPI AGTomography2379-13812379-139X2024-11-0110121930194610.3390/tomography10120140Three-Dimensional Thermal Tomography with Physics-Informed Neural NetworksTheodoros Leontiou0Anna Frixou1Marios Charalambides2Efstathios Stiliaris3Costas N. Papanicolas4Sofia Nikolaidou5Antonis Papadakis6Department of Mechanical Engineering, Frederick University, Nicosia 1036, CyprusComputation-Based Science and Technology Research Center (CaSToRC), The Cyprus Institute, 20 Kavafi Street, Nicosia 2121, CyprusDepartment of Business Administration, Frederick University, Nicosia 1036, CyprusComputation-Based Science and Technology Research Center (CaSToRC), The Cyprus Institute, 20 Kavafi Street, Nicosia 2121, CyprusComputation-Based Science and Technology Research Center (CaSToRC), The Cyprus Institute, 20 Kavafi Street, Nicosia 2121, CyprusKYAMOS Ltd., 37 Polyneikis Street, Strovolos, Nicosia 2047, CyprusKYAMOS Ltd., 37 Polyneikis Street, Strovolos, Nicosia 2047, Cyprus<b>Background</b>: Accurate reconstruction of internal temperature fields from surface temperature data is critical for applications such as non-invasive thermal imaging, particularly in scenarios involving small temperature gradients, like those in the human body. <b>Methods</b>: In this study, we employed 3D convolutional neural networks (CNNs) to predict internal temperature fields. The network’s performance was evaluated under both ideal and non-ideal conditions, incorporating noise and background temperature variations. A physics-informed loss function embedding the heat equation was used in conjunction with statistical uncertainty during training to simulate realistic scenarios. <b>Results</b>: The CNN achieved high accuracy for small phantoms (e.g., 10 cm in diameter). However, under non-ideal conditions, the network’s predictive capacity diminished in larger domains, particularly in regions distant from the surface. The introduction of physical constraints in the training processes improved the model’s robustness in noisy environments, enabling accurate reconstruction of hot-spots in deeper regions where traditional CNNs struggled. <b>Conclusions</b>: Combining deep learning with physical constraints provides a robust framework for non-invasive thermal imaging and other applications requiring high-precision temperature field reconstruction, particularly under non-ideal conditions.https://www.mdpi.com/2379-139X/10/12/140thermal tomographyconvolutional neural networksphysics-informed neural networks3D temperature fieldheat conductioninverse problems
spellingShingle Theodoros Leontiou
Anna Frixou
Marios Charalambides
Efstathios Stiliaris
Costas N. Papanicolas
Sofia Nikolaidou
Antonis Papadakis
Three-Dimensional Thermal Tomography with Physics-Informed Neural Networks
Tomography
thermal tomography
convolutional neural networks
physics-informed neural networks
3D temperature field
heat conduction
inverse problems
title Three-Dimensional Thermal Tomography with Physics-Informed Neural Networks
title_full Three-Dimensional Thermal Tomography with Physics-Informed Neural Networks
title_fullStr Three-Dimensional Thermal Tomography with Physics-Informed Neural Networks
title_full_unstemmed Three-Dimensional Thermal Tomography with Physics-Informed Neural Networks
title_short Three-Dimensional Thermal Tomography with Physics-Informed Neural Networks
title_sort three dimensional thermal tomography with physics informed neural networks
topic thermal tomography
convolutional neural networks
physics-informed neural networks
3D temperature field
heat conduction
inverse problems
url https://www.mdpi.com/2379-139X/10/12/140
work_keys_str_mv AT theodorosleontiou threedimensionalthermaltomographywithphysicsinformedneuralnetworks
AT annafrixou threedimensionalthermaltomographywithphysicsinformedneuralnetworks
AT marioscharalambides threedimensionalthermaltomographywithphysicsinformedneuralnetworks
AT efstathiosstiliaris threedimensionalthermaltomographywithphysicsinformedneuralnetworks
AT costasnpapanicolas threedimensionalthermaltomographywithphysicsinformedneuralnetworks
AT sofianikolaidou threedimensionalthermaltomographywithphysicsinformedneuralnetworks
AT antonispapadakis threedimensionalthermaltomographywithphysicsinformedneuralnetworks