Tumor margin classification of head and neck cancer using hyperspectral imaging and convolutional neural networks

Martin Halicek, James V. Little, Xu Wang, Mihir Patel, Christopher C. Griffith, Amy Y. Chen, Baowei Fei

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

One of the largest factors affecting disease recurrence after surgical cancer resection is negative surgical margins. Hyperspectral imaging (HSI) is an optical imaging technique with potential to serve as a computer aided diagnostic tool for identifying cancer in gross ex-vivo specimens. We developed a tissue classifier using three distinct convolutional neural network (CNN) architectures on HSI data to investigate the ability to classify the cancer margins from ex-vivo human surgical specimens, collected from 20 patients undergoing surgical cancer resection as a preliminary validation group. A new approach for generating the HSI ground truth using a registered histological cancer margin is applied in order to create a validation dataset. The CNN-based method classifies the tumor-normal margin of squamous cell carcinoma (SCCa) versus normal oral tissue with an area under the curve (AUC) of 0.86 for inter-patient validation, performing with 81% accuracy, 84% sensitivity, and 77% specificity. Thyroid carcinoma cancer-normal margins are classified with an AUC of 0.94 for inter-patient validation, performing with 90% accuracy, 91% sensitivity, and 88% specificity. Our preliminary results on a limited patient dataset demonstrate the predictive ability of HSI-based cancer margin detection, which warrants further investigation with more patient data and additional processing techniques to optimize the proposed deep learning method.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2018
Subtitle of host publicationImage-Guided Procedures, Robotic Interventions, and Modeling
EditorsBaowei Fei, Robert J. Webster
PublisherSPIE
ISBN (Electronic)9781510616417
DOIs
StatePublished - Jan 1 2018
Externally publishedYes
EventMedical Imaging 2018: Image-Guided Procedures, Robotic Interventions, and Modeling - Houston, United States
Duration: Feb 12 2018Feb 15 2018

Publication series

NameProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume10576
ISSN (Print)1605-7422

Other

OtherMedical Imaging 2018: Image-Guided Procedures, Robotic Interventions, and Modeling
CountryUnited States
CityHouston
Period2/12/182/15/18

Fingerprint

Head and Neck Neoplasms
Tumors
margins
tumors
cancer
Neural networks
Neoplasms
Tissue
Aptitude
Thyroid Neoplasms
Area Under Curve
Network architecture
Classifiers
Sensitivity and Specificity
Imaging techniques
Optical Imaging
Hyperspectral imaging
ground truth
Processing
sensitivity

Keywords

  • Cancer margin detection
  • Convolutional neural network
  • Deep learning
  • Head and neck cancer
  • Head and neck surgery
  • Hyperspectral imaging
  • In-traoperative imaging

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Biomaterials
  • Atomic and Molecular Physics, and Optics
  • Radiology Nuclear Medicine and imaging

Cite this

Halicek, M., Little, J. V., Wang, X., Patel, M., Griffith, C. C., Chen, A. Y., & Fei, B. (2018). Tumor margin classification of head and neck cancer using hyperspectral imaging and convolutional neural networks. In B. Fei, & R. J. Webster (Eds.), Medical Imaging 2018: Image-Guided Procedures, Robotic Interventions, and Modeling [1057605] (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 10576). SPIE. https://doi.org/10.1117/12.2293167

Tumor margin classification of head and neck cancer using hyperspectral imaging and convolutional neural networks. / Halicek, Martin; Little, James V.; Wang, Xu; Patel, Mihir; Griffith, Christopher C.; Chen, Amy Y.; Fei, Baowei.

Medical Imaging 2018: Image-Guided Procedures, Robotic Interventions, and Modeling. ed. / Baowei Fei; Robert J. Webster. SPIE, 2018. 1057605 (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 10576).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Halicek, M, Little, JV, Wang, X, Patel, M, Griffith, CC, Chen, AY & Fei, B 2018, Tumor margin classification of head and neck cancer using hyperspectral imaging and convolutional neural networks. in B Fei & RJ Webster (eds), Medical Imaging 2018: Image-Guided Procedures, Robotic Interventions, and Modeling., 1057605, Progress in Biomedical Optics and Imaging - Proceedings of SPIE, vol. 10576, SPIE, Medical Imaging 2018: Image-Guided Procedures, Robotic Interventions, and Modeling, Houston, United States, 2/12/18. https://doi.org/10.1117/12.2293167
Halicek M, Little JV, Wang X, Patel M, Griffith CC, Chen AY et al. Tumor margin classification of head and neck cancer using hyperspectral imaging and convolutional neural networks. In Fei B, Webster RJ, editors, Medical Imaging 2018: Image-Guided Procedures, Robotic Interventions, and Modeling. SPIE. 2018. 1057605. (Progress in Biomedical Optics and Imaging - Proceedings of SPIE). https://doi.org/10.1117/12.2293167
Halicek, Martin ; Little, James V. ; Wang, Xu ; Patel, Mihir ; Griffith, Christopher C. ; Chen, Amy Y. ; Fei, Baowei. / Tumor margin classification of head and neck cancer using hyperspectral imaging and convolutional neural networks. Medical Imaging 2018: Image-Guided Procedures, Robotic Interventions, and Modeling. editor / Baowei Fei ; Robert J. Webster. SPIE, 2018. (Progress in Biomedical Optics and Imaging - Proceedings of SPIE).
@inproceedings{e086b57cfdf5411faee97c5dcd01251e,
title = "Tumor margin classification of head and neck cancer using hyperspectral imaging and convolutional neural networks",
abstract = "One of the largest factors affecting disease recurrence after surgical cancer resection is negative surgical margins. Hyperspectral imaging (HSI) is an optical imaging technique with potential to serve as a computer aided diagnostic tool for identifying cancer in gross ex-vivo specimens. We developed a tissue classifier using three distinct convolutional neural network (CNN) architectures on HSI data to investigate the ability to classify the cancer margins from ex-vivo human surgical specimens, collected from 20 patients undergoing surgical cancer resection as a preliminary validation group. A new approach for generating the HSI ground truth using a registered histological cancer margin is applied in order to create a validation dataset. The CNN-based method classifies the tumor-normal margin of squamous cell carcinoma (SCCa) versus normal oral tissue with an area under the curve (AUC) of 0.86 for inter-patient validation, performing with 81{\%} accuracy, 84{\%} sensitivity, and 77{\%} specificity. Thyroid carcinoma cancer-normal margins are classified with an AUC of 0.94 for inter-patient validation, performing with 90{\%} accuracy, 91{\%} sensitivity, and 88{\%} specificity. Our preliminary results on a limited patient dataset demonstrate the predictive ability of HSI-based cancer margin detection, which warrants further investigation with more patient data and additional processing techniques to optimize the proposed deep learning method.",
keywords = "Cancer margin detection, Convolutional neural network, Deep learning, Head and neck cancer, Head and neck surgery, Hyperspectral imaging, In-traoperative imaging",
author = "Martin Halicek and Little, {James V.} and Xu Wang and Mihir Patel and Griffith, {Christopher C.} and Chen, {Amy Y.} and Baowei Fei",
year = "2018",
month = "1",
day = "1",
doi = "10.1117/12.2293167",
language = "English (US)",
series = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",
publisher = "SPIE",
editor = "Baowei Fei and Webster, {Robert J.}",
booktitle = "Medical Imaging 2018",

}

TY - GEN

T1 - Tumor margin classification of head and neck cancer using hyperspectral imaging and convolutional neural networks

AU - Halicek, Martin

AU - Little, James V.

AU - Wang, Xu

AU - Patel, Mihir

AU - Griffith, Christopher C.

AU - Chen, Amy Y.

AU - Fei, Baowei

PY - 2018/1/1

Y1 - 2018/1/1

N2 - One of the largest factors affecting disease recurrence after surgical cancer resection is negative surgical margins. Hyperspectral imaging (HSI) is an optical imaging technique with potential to serve as a computer aided diagnostic tool for identifying cancer in gross ex-vivo specimens. We developed a tissue classifier using three distinct convolutional neural network (CNN) architectures on HSI data to investigate the ability to classify the cancer margins from ex-vivo human surgical specimens, collected from 20 patients undergoing surgical cancer resection as a preliminary validation group. A new approach for generating the HSI ground truth using a registered histological cancer margin is applied in order to create a validation dataset. The CNN-based method classifies the tumor-normal margin of squamous cell carcinoma (SCCa) versus normal oral tissue with an area under the curve (AUC) of 0.86 for inter-patient validation, performing with 81% accuracy, 84% sensitivity, and 77% specificity. Thyroid carcinoma cancer-normal margins are classified with an AUC of 0.94 for inter-patient validation, performing with 90% accuracy, 91% sensitivity, and 88% specificity. Our preliminary results on a limited patient dataset demonstrate the predictive ability of HSI-based cancer margin detection, which warrants further investigation with more patient data and additional processing techniques to optimize the proposed deep learning method.

AB - One of the largest factors affecting disease recurrence after surgical cancer resection is negative surgical margins. Hyperspectral imaging (HSI) is an optical imaging technique with potential to serve as a computer aided diagnostic tool for identifying cancer in gross ex-vivo specimens. We developed a tissue classifier using three distinct convolutional neural network (CNN) architectures on HSI data to investigate the ability to classify the cancer margins from ex-vivo human surgical specimens, collected from 20 patients undergoing surgical cancer resection as a preliminary validation group. A new approach for generating the HSI ground truth using a registered histological cancer margin is applied in order to create a validation dataset. The CNN-based method classifies the tumor-normal margin of squamous cell carcinoma (SCCa) versus normal oral tissue with an area under the curve (AUC) of 0.86 for inter-patient validation, performing with 81% accuracy, 84% sensitivity, and 77% specificity. Thyroid carcinoma cancer-normal margins are classified with an AUC of 0.94 for inter-patient validation, performing with 90% accuracy, 91% sensitivity, and 88% specificity. Our preliminary results on a limited patient dataset demonstrate the predictive ability of HSI-based cancer margin detection, which warrants further investigation with more patient data and additional processing techniques to optimize the proposed deep learning method.

KW - Cancer margin detection

KW - Convolutional neural network

KW - Deep learning

KW - Head and neck cancer

KW - Head and neck surgery

KW - Hyperspectral imaging

KW - In-traoperative imaging

UR - http://www.scopus.com/inward/record.url?scp=85050665323&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85050665323&partnerID=8YFLogxK

U2 - 10.1117/12.2293167

DO - 10.1117/12.2293167

M3 - Conference contribution

C2 - 30245540

AN - SCOPUS:85050665323

T3 - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

BT - Medical Imaging 2018

A2 - Fei, Baowei

A2 - Webster, Robert J.

PB - SPIE

ER -