Towards Exercise Radiomics: Deep Neural Network-Based Automatic Analysis of Thermal Images Captured During Exercise.
- Publication type:
- Journal article
- Metadata:
-
- Autoren
- Barlo Hillen
- Daniel Andres Lopez
- Elmar Schoemer
- Markus Naegele
- Perikles Simon
- Autoren-URL
- https://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=fis-test-1&SrcAuth=WosAPI&KeyUT=WOS:000852247000020&DestLinkType=FullRecord&DestApp=WOS_CPL
- DOI
- 10.1109/JBHI.2022.3186530
- eISSN
- 2168-2208
- Externe Identifier
- Clarivate Analytics Document Solution ID: 4K9FX
- PubMed Identifier: 35759601
- ISSN
- 2168-2194
- Ausgabe der Veröffentlichung
- 9
- Zeitschrift
- IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
- Schlüsselwörter
- Artificial neural networks
- biotechnology
- incremental exercise testing
- semantic segmentation
- thermal imaging
- Paginierung
- 4530 - 4540
- Datum der Veröffentlichung
- 2022
- Status
- Published
- Titel
- Towards Exercise Radiomics: Deep Neural Network-Based Automatic Analysis of Thermal Images Captured During Exercise
- Sub types
- Article
- Ausgabe der Zeitschrift
- 26
Data source: Web of Science (Lite)
- Other metadata sources:
-
- Autoren
- Barlo Hillen
- Daniel Andres Lopez
- Elmar Schomer
- Markus Nagele
- Perikles Simon
- DOI
- 10.1109/jbhi.2022.3186530
- eISSN
- 2168-2208
- ISSN
- 2168-2194
- Ausgabe der Veröffentlichung
- 9
- Zeitschrift
- IEEE Journal of Biomedical and Health Informatics
- Paginierung
- 4530 - 4540
- Datum der Veröffentlichung
- 2022
- Status
- Published
- Herausgeber
- Institute of Electrical and Electronics Engineers (IEEE)
- Herausgeber URL
- http://dx.doi.org/10.1109/jbhi.2022.3186530
- Datum der Datenerfassung
- 2024
- Titel
- Towards Exercise Radiomics: Deep Neural Network-Based Automatic Analysis of Thermal Images Captured During Exercise
- Ausgabe der Zeitschrift
- 26
Data source: Crossref
- Abstract
- Infrared thermography is increasingly applied in sports science due to promising observations regarding changes in skin's surface radiation temperature ( T<sub>sr</sub>) before, during, and after exercise. The common manual thermogram analysis limits an objective and reproducible measurement of T<sub>sr</sub>. Previous analysis approaches depend on expert knowledge and have not been applied during movement. We aimed to develop a deep neural network (DNN) capable of automatically and objectively segmenting body parts, recognizing blood vessel-associated T<sub>sr</sub> distributions, and continuously measuring T<sub>sr</sub> during exercise. We conducted 38 cardiopulmonary exercise tests on a treadmill. We developed two DNNs: body part network and vessel network, to perform semantic segmentation of 1 107 855 thermal images. Both DNNs were trained with 263 training and 75 validation images. Additionally, we compare the results of a common manual thermogram analysis with these of the DNNs. Performance analysis identified a mean IoU of 0.8 for body part network and 0.6 for vessel network. There is a high agreement between manual and automatic analysis (r = 0.999; p 0.001; T-test: p = 0.116), with a mean difference of 0.01 <sup>°</sup>C (0.08). Non-parametric Bland Altman's analysis showed that the 95% agreement ranges between - 0.086 <sup>°</sup>C and 0.228 <sup>°</sup>C. The developed DNNs enable automatic, objective, and continuous measurement of T<sub>sr</sub> and recognition of blood vessel-associated T<sub>sr</sub> distributions in resting and moving legs. Hence, the DNNs surpass previous algorithms by eliminating manual region of interest selection and form the currently needed foundation to extensively investigate T<sub>sr</sub> distributions related to non-invasive diagnostics of (patho-)physiological traits in means of exercise radiomics.
- Autoren
- Barlo Hillen
- Daniel Andres Lopez
- Elmar Schomer
- Markus Nagele
- Perikles Simon
- DOI
- 10.1109/jbhi.2022.3186530
- eISSN
- 2168-2208
- Externe Identifier
- PubMed Identifier: 35759601
- Funding acknowledgements
- German Parliament: ZF4211603GR9
- Bundesministerium für Wirtschaft und Energie:
- Open access
- false
- ISSN
- 2168-2194
- Ausgabe der Veröffentlichung
- 9
- Zeitschrift
- IEEE journal of biomedical and health informatics
- Schlüsselwörter
- Humans
- Thermography
- Exercise
- Algorithms
- Neural Networks, Computer
- Sprache
- eng
- Medium
- Print-Electronic
- Online publication date
- 2022
- Paginierung
- 4530 - 4540
- Datum der Veröffentlichung
- 2022
- Status
- Published
- Datum der Datenerfassung
- 2022
- Titel
- Towards Exercise Radiomics: Deep Neural Network-Based Automatic Analysis of Thermal Images Captured During Exercise.
- Sub types
- Research Support, Non-U.S. Gov't
- Journal Article
- Ausgabe der Zeitschrift
- 26
Data source: Europe PubMed Central
- Abstract
- Infrared thermography is increasingly applied in sports science due to promising observations regarding changes in skin's surface radiation temperature ( Tsr) before, during, and after exercise. The common manual thermogram analysis limits an objective and reproducible measurement of Tsr. Previous analysis approaches depend on expert knowledge and have not been applied during movement. We aimed to develop a deep neural network (DNN) capable of automatically and objectively segmenting body parts, recognizing blood vessel-associated Tsr distributions, and continuously measuring Tsr during exercise. We conducted 38 cardiopulmonary exercise tests on a treadmill. We developed two DNNs: body part network and vessel network, to perform semantic segmentation of 1 107 855 thermal images. Both DNNs were trained with 263 training and 75 validation images. Additionally, we compare the results of a common manual thermogram analysis with these of the DNNs. Performance analysis identified a mean IoU of 0.8 for body part network and 0.6 for vessel network. There is a high agreement between manual and automatic analysis (r = 0.999; p 0.001; T-test: p = 0.116), with a mean difference of 0.01 °C (0.08). Non-parametric Bland Altman's analysis showed that the 95% agreement ranges between - 0.086 °C and 0.228 °C. The developed DNNs enable automatic, objective, and continuous measurement of Tsr and recognition of blood vessel-associated Tsr distributions in resting and moving legs. Hence, the DNNs surpass previous algorithms by eliminating manual region of interest selection and form the currently needed foundation to extensively investigate Tsr distributions related to non-invasive diagnostics of (patho-)physiological traits in means of exercise radiomics.
- Autoren
- Barlo Hillen
- Daniel Andres Lopez
- Elmar Schomer
- Markus Nagele
- Perikles Simon
- Autoren-URL
- https://www.ncbi.nlm.nih.gov/pubmed/35759601
- DOI
- 10.1109/JBHI.2022.3186530
- eISSN
- 2168-2208
- Ausgabe der Veröffentlichung
- 9
- Zeitschrift
- IEEE J Biomed Health Inform
- Schlüsselwörter
- Algorithms
- Exercise
- Humans
- Neural Networks, Computer
- Thermography
- Sprache
- eng
- Country
- United States
- Paginierung
- 4530 - 4540
- Datum der Veröffentlichung
- 2022
- Status
- Published
- Datum, an dem der Datensatz öffentlich gemacht wurde
- 2022
- Titel
- Towards Exercise Radiomics: Deep Neural Network-Based Automatic Analysis of Thermal Images Captured During Exercise.
- Sub types
- Journal Article
- Research Support, Non-U.S. Gov't
- Ausgabe der Zeitschrift
- 26
Data source: PubMed
- Autoren
- Barlo Hillen
- Daniel Andres Lopez
- Elmar Schömer
- Markus Nägele
- Perikles Simon
- DOI
- 10.1109/JBHI.2022.3186530
- eISSN
- 2168-2208
- ISSN
- 2168-2194
- Zeitschrift
- IEEE Journal of Biomedical and Health Informatics
- Paginierung
- 1 - 11
- Datum der Veröffentlichung
- 2022
- Status
- Published
- Herausgeber
- Institute of Electrical and Electronics Engineers (IEEE)
- Datum der Datenerfassung
- 2022
- Titel
- Towards Exercise Radiomics: Deep Neural Network-Based Automatic Analysis of Thermal Images Captured During Exercise.
- Sub types
- Article
Data source: Manual
- Beziehungen:
- Property of