In vivo detection of head and neck tumors by hyperspectral imaging combined with deep learning methods
Standard
In vivo detection of head and neck tumors by hyperspectral imaging combined with deep learning methods. / Eggert, Dennis; Bengs, Marcel; Westermann, Stephan; Gessert, Nils; Gerstner, Andreas O H; Mueller, Nina A; Bewarder, Julian; Schlaefer, Alexander; Betz, Christian; Laffers, Wiebke.
In: J BIOPHOTONICS, Vol. 15, No. 3, e202100167, 03.2022.Research output: SCORING: Contribution to journal › SCORING: Journal article › Research › peer-review
Harvard
APA
Vancouver
Bibtex
}
RIS
TY - JOUR
T1 - In vivo detection of head and neck tumors by hyperspectral imaging combined with deep learning methods
AU - Eggert, Dennis
AU - Bengs, Marcel
AU - Westermann, Stephan
AU - Gessert, Nils
AU - Gerstner, Andreas O H
AU - Mueller, Nina A
AU - Bewarder, Julian
AU - Schlaefer, Alexander
AU - Betz, Christian
AU - Laffers, Wiebke
N1 - © 2021 The Authors. Journal of Biophotonics published by Wiley-VCH GmbH.
PY - 2022/3
Y1 - 2022/3
N2 - Currently, there are no fast and accurate screening methods available for head and neck cancer, the eighth most common tumor entity. For this study, we used hyperspectral imaging, an imaging technique for quantitative and objective surface analysis, combined with deep learning methods for automated tissue classification. As part of a prospective clinical observational study, hyperspectral datasets of laryngeal, hypopharyngeal and oropharyngeal mucosa were recorded in 98 patients before surgery in vivo. We established an automated data interpretation pathway that can classify the tissue into healthy and tumorous using convolutional neural networks with 2D spatial or 3D spatio-spectral convolutions combined with a state-of-the-art Densenet architecture. Using 24 patients for testing, our 3D spatio-spectral Densenet classification method achieves an average accuracy of 81%, a sensitivity of 83% and a specificity of 79%.
AB - Currently, there are no fast and accurate screening methods available for head and neck cancer, the eighth most common tumor entity. For this study, we used hyperspectral imaging, an imaging technique for quantitative and objective surface analysis, combined with deep learning methods for automated tissue classification. As part of a prospective clinical observational study, hyperspectral datasets of laryngeal, hypopharyngeal and oropharyngeal mucosa were recorded in 98 patients before surgery in vivo. We established an automated data interpretation pathway that can classify the tissue into healthy and tumorous using convolutional neural networks with 2D spatial or 3D spatio-spectral convolutions combined with a state-of-the-art Densenet architecture. Using 24 patients for testing, our 3D spatio-spectral Densenet classification method achieves an average accuracy of 81%, a sensitivity of 83% and a specificity of 79%.
U2 - 10.1002/jbio.202100167
DO - 10.1002/jbio.202100167
M3 - SCORING: Journal article
C2 - 34889065
VL - 15
JO - J BIOPHOTONICS
JF - J BIOPHOTONICS
SN - 1864-063X
IS - 3
M1 - e202100167
ER -