Improved Barrett's cancer detection in volumetric laser endomicroscopy scans using multiple-frame voting

A. Rikos, F. van der Sommen, S. Zinger, P.H.N. de With, W.L. Curvers, E.J. (Erik) Schoon, A.-F. Swager, J.J. (Jacques) Bergman

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

2 Downloads (Pure)

Abstract

This paper explores the feasibility of using multiframe analysis to increase the classification performance of machine learning methods for cancer detection in Volumetric Laser Endomicroscopy (VLE). VLE is a novel and promising modality for the detection of neoplasia in patients with Baretts Esophagus (BE). It produces hundreds of high-resolution, cross-sectional images of the esophagus and offers considerable advantages compared to current methods. While some recent studies have proposed cancer detection algorithms for single VLE frames, the study described in this paper is the first to make use of VLE volumes for the differentiation between dysplastic and non-dysplastic tissue. We explore the use of various voting schemes for a broad range of features and classification methods. Our results demonstrate that multi-frame analysis leads to superior performance, irrespective of the chosen feature-classifier combination. By using multi-frame analysis with straightforward voting methods, the Area Under the receiver operating Curve (AUC) is increased by an average of over 12% compared to using single VLE frames. When only considering methods that achieve expert performance or higher (AUC≥0.81), an even larger performance improvement of up to 16.9% is observed. Furthermore, with many feature/classifier combinations showing AUC values ranging from 0.90 to 0.98, our experiments indicate that computeraided methods can considerably outperform medical experts, who demonstrate an AUC of 0.81 using a recently proposed clinical prediction model.
Original languageDutch
Title of host publication30th IEEE International Conference on Computer-Based Medical Systems (CBMS), 20-22 June 2017, Thessaloniki, Greece
Place of PublicationPiscataway
PublisherInstitute of Electrical and Electronics Engineers
Pages708-713
Number of pages6
ISBN (Electronic)978-1-5386-1710-6
ISBN (Print)978-1-5386-1711-3
DOIs
Publication statusPublished - 14 Nov 2017

Cite this

Rikos, A., van der Sommen, F., Zinger, S., de With, P. H. N., Curvers, W. L., Schoon, E. J. E., ... Bergman, J. J. J. (2017). Improved Barrett's cancer detection in volumetric laser endomicroscopy scans using multiple-frame voting. In 30th IEEE International Conference on Computer-Based Medical Systems (CBMS), 20-22 June 2017, Thessaloniki, Greece (pp. 708-713). Piscataway: Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/CBMS.2017.31
Rikos, A. ; van der Sommen, F. ; Zinger, S. ; de With, P.H.N. ; Curvers, W.L. ; Schoon, E.J. (Erik) ; Swager, A.-F. ; Bergman, J.J. (Jacques). / Improved Barrett's cancer detection in volumetric laser endomicroscopy scans using multiple-frame voting. 30th IEEE International Conference on Computer-Based Medical Systems (CBMS), 20-22 June 2017, Thessaloniki, Greece. Piscataway : Institute of Electrical and Electronics Engineers, 2017. pp. 708-713
@inproceedings{8988fe854493465f9b24c3cfdc6551b2,
title = "Improved Barrett's cancer detection in volumetric laser endomicroscopy scans using multiple-frame voting",
abstract = "This paper explores the feasibility of using multiframe analysis to increase the classification performance of machine learning methods for cancer detection in Volumetric Laser Endomicroscopy (VLE). VLE is a novel and promising modality for the detection of neoplasia in patients with Baretts Esophagus (BE). It produces hundreds of high-resolution, cross-sectional images of the esophagus and offers considerable advantages compared to current methods. While some recent studies have proposed cancer detection algorithms for single VLE frames, the study described in this paper is the first to make use of VLE volumes for the differentiation between dysplastic and non-dysplastic tissue. We explore the use of various voting schemes for a broad range of features and classification methods. Our results demonstrate that multi-frame analysis leads to superior performance, irrespective of the chosen feature-classifier combination. By using multi-frame analysis with straightforward voting methods, the Area Under the receiver operating Curve (AUC) is increased by an average of over 12{\%} compared to using single VLE frames. When only considering methods that achieve expert performance or higher (AUC≥0.81), an even larger performance improvement of up to 16.9{\%} is observed. Furthermore, with many feature/classifier combinations showing AUC values ranging from 0.90 to 0.98, our experiments indicate that computeraided methods can considerably outperform medical experts, who demonstrate an AUC of 0.81 using a recently proposed clinical prediction model.",
author = "A. Rikos and {van der Sommen}, F. and S. Zinger and {de With}, P.H.N. and W.L. Curvers and Schoon, {E.J. (Erik)} and A.-F. Swager and Bergman, {J.J. (Jacques)}",
year = "2017",
month = "11",
day = "14",
doi = "10.1109/CBMS.2017.31",
language = "Nederlands",
isbn = "978-1-5386-1711-3",
pages = "708--713",
booktitle = "30th IEEE International Conference on Computer-Based Medical Systems (CBMS), 20-22 June 2017, Thessaloniki, Greece",
publisher = "Institute of Electrical and Electronics Engineers",
address = "Verenigde Staten van Amerika",

}

Rikos, A, van der Sommen, F, Zinger, S, de With, PHN, Curvers, WL, Schoon, EJE, Swager, A-F & Bergman, JJJ 2017, Improved Barrett's cancer detection in volumetric laser endomicroscopy scans using multiple-frame voting. in 30th IEEE International Conference on Computer-Based Medical Systems (CBMS), 20-22 June 2017, Thessaloniki, Greece. Institute of Electrical and Electronics Engineers, Piscataway, pp. 708-713. https://doi.org/10.1109/CBMS.2017.31

Improved Barrett's cancer detection in volumetric laser endomicroscopy scans using multiple-frame voting. / Rikos, A.; van der Sommen, F.; Zinger, S.; de With, P.H.N.; Curvers, W.L.; Schoon, E.J. (Erik); Swager, A.-F.; Bergman, J.J. (Jacques).

30th IEEE International Conference on Computer-Based Medical Systems (CBMS), 20-22 June 2017, Thessaloniki, Greece. Piscataway : Institute of Electrical and Electronics Engineers, 2017. p. 708-713.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Improved Barrett's cancer detection in volumetric laser endomicroscopy scans using multiple-frame voting

AU - Rikos, A.

AU - van der Sommen, F.

AU - Zinger, S.

AU - de With, P.H.N.

AU - Curvers, W.L.

AU - Schoon, E.J. (Erik)

AU - Swager, A.-F.

AU - Bergman, J.J. (Jacques)

PY - 2017/11/14

Y1 - 2017/11/14

N2 - This paper explores the feasibility of using multiframe analysis to increase the classification performance of machine learning methods for cancer detection in Volumetric Laser Endomicroscopy (VLE). VLE is a novel and promising modality for the detection of neoplasia in patients with Baretts Esophagus (BE). It produces hundreds of high-resolution, cross-sectional images of the esophagus and offers considerable advantages compared to current methods. While some recent studies have proposed cancer detection algorithms for single VLE frames, the study described in this paper is the first to make use of VLE volumes for the differentiation between dysplastic and non-dysplastic tissue. We explore the use of various voting schemes for a broad range of features and classification methods. Our results demonstrate that multi-frame analysis leads to superior performance, irrespective of the chosen feature-classifier combination. By using multi-frame analysis with straightforward voting methods, the Area Under the receiver operating Curve (AUC) is increased by an average of over 12% compared to using single VLE frames. When only considering methods that achieve expert performance or higher (AUC≥0.81), an even larger performance improvement of up to 16.9% is observed. Furthermore, with many feature/classifier combinations showing AUC values ranging from 0.90 to 0.98, our experiments indicate that computeraided methods can considerably outperform medical experts, who demonstrate an AUC of 0.81 using a recently proposed clinical prediction model.

AB - This paper explores the feasibility of using multiframe analysis to increase the classification performance of machine learning methods for cancer detection in Volumetric Laser Endomicroscopy (VLE). VLE is a novel and promising modality for the detection of neoplasia in patients with Baretts Esophagus (BE). It produces hundreds of high-resolution, cross-sectional images of the esophagus and offers considerable advantages compared to current methods. While some recent studies have proposed cancer detection algorithms for single VLE frames, the study described in this paper is the first to make use of VLE volumes for the differentiation between dysplastic and non-dysplastic tissue. We explore the use of various voting schemes for a broad range of features and classification methods. Our results demonstrate that multi-frame analysis leads to superior performance, irrespective of the chosen feature-classifier combination. By using multi-frame analysis with straightforward voting methods, the Area Under the receiver operating Curve (AUC) is increased by an average of over 12% compared to using single VLE frames. When only considering methods that achieve expert performance or higher (AUC≥0.81), an even larger performance improvement of up to 16.9% is observed. Furthermore, with many feature/classifier combinations showing AUC values ranging from 0.90 to 0.98, our experiments indicate that computeraided methods can considerably outperform medical experts, who demonstrate an AUC of 0.81 using a recently proposed clinical prediction model.

U2 - 10.1109/CBMS.2017.31

DO - 10.1109/CBMS.2017.31

M3 - Conferentiebijdrage

SN - 978-1-5386-1711-3

SP - 708

EP - 713

BT - 30th IEEE International Conference on Computer-Based Medical Systems (CBMS), 20-22 June 2017, Thessaloniki, Greece

PB - Institute of Electrical and Electronics Engineers

CY - Piscataway

ER -

Rikos A, van der Sommen F, Zinger S, de With PHN, Curvers WL, Schoon EJE et al. Improved Barrett's cancer detection in volumetric laser endomicroscopy scans using multiple-frame voting. In 30th IEEE International Conference on Computer-Based Medical Systems (CBMS), 20-22 June 2017, Thessaloniki, Greece. Piscataway: Institute of Electrical and Electronics Engineers. 2017. p. 708-713 https://doi.org/10.1109/CBMS.2017.31