Prospective development and validation of a volumetric laser endomicroscopy computer algorithm for detection of Barrett’s neoplasia

Maarten R. Struyvenberg; Albert J. de Groof; Roger Fonollà; Fons van der Sommen; Peter H.N. de With; Erik J. Schoon; Bas L.A.M. Weusten; Cadman L. Leggett; Allon Kahn; Arvind J. Trindade; Eric K. Ganguly; Vani J.A. Konda; Charles J. Lightdale; Douglas K. Pleskow; Amrita Sethi; Michael S. Smith; Michael B. Wallace; Herbert C. Wolfsen; Gary J. Tearney; Sybren L. Meijer; Michael Vieth; Roos E. Pouw; Wouter L. Curvers; Jacques J. Bergman

doi:10.1016/j.gie.2020.07.052

Prospective development and validation of a volumetric laser endomicroscopy computer algorithm for detection of Barrett’s neoplasia

Maarten R. Struyvenberg, Albert J. de Groof, Roger Fonollà, Fons van der Sommen, Peter H.N. de With, Erik J. Schoon, Bas L.A.M. Weusten, Cadman L. Leggett, Allon Kahn, Arvind J. Trindade, Eric K. Ganguly, Vani J.A. Konda, Charles J. Lightdale, Douglas K. Pleskow, Amrita Sethi, Michael S. Smith, Michael B. Wallace, Herbert C. Wolfsen, Gary J. Tearney, Sybren L. MeijerMichael Vieth, Roos E. Pouw, Wouter L. Curvers, Jacques J. Bergman (Corresponding author)

Research output: Contribution to journal › Article › Academic › peer-review

15 Citations (Scopus)

Abstract

Background and Aims
Volumetric laser endomicroscopy (VLE) is an advanced imaging modality used to detect Barrett’s esophagus (BE) dysplasia. However, real-time interpretation of VLE scans is complex and time-consuming. Computer-aided detection (CAD) may aid in the process of VLE image interpretation. Our aim was to train and validate a CAD algorithm for VLE-based detection of BE neoplasia.

Methods
The multicenter, VLE PREDICT study, prospectively enrolled 47 BE patients. In total, 229 nondysplastic BE, and 89 neoplastic (HGD/EAC) targets were laser marked under VLE guidance and subsequently biopsied for histological diagnosis. Deep convolutional neural networks were used to construct a CAD algorithm for differentiation between nondysplastic and neoplastic BE tissue. The CAD algorithm was trained on a set consisting of the first 22 patients (134 NDBE and 38 neoplastic targets) and validated on a separate test set of patients 23 to 47 (95 NDBE and 51 neoplastic targets). Finally, algorithm performance was benchmarked against the performance of 10 VLE experts.

Results
Using the Training set to construct the algorithm resulted in an accuracy of 92%, sensitivity of 95% and specificity of 92%. When performance was assessed on the Test set, accuracy, sensitivity, and specificity were 85%, 91%, and 82%, respectively. The algorithm outperformed all 10 VLE experts, who demonstrated an overall accuracy of 77%, sensitivity of 70%, and specificity of 81%.

Conclusions
We developed, validated, and benchmarked a VLE CAD algorithm for detection of BE neoplasia using prospectively collected and biopsy-correlated VLE targets. The algorithm detected neoplasia with high accuracy and outperformed 10 VLE experts.

Original language	English
Pages (from-to)	871-879
Number of pages	9
Journal	Gastrointestinal Endoscopy
Volume	93
Issue number	4
Early online date	29 Jul 2020
DOIs	https://doi.org/10.1016/j.gie.2020.07.052
Publication status	Published - Apr 2021

Funding

DISCLOSURE: Dr Weusten has received research support and speakers fees from Pentax Medical. Dr Leggett has received indirect research support from NinePoint Medical. Dr Kahn received an unrestricted travel grant and research equipment from NinePoint Medical. Dr Trindade has received research support NinePoint Medical and consultant fees from Olympus America and Pentax Medical. Dr Ganguly has received consultant fees from Boston Scientific. Dr Konda has received a research grant from Pentax. Dr Pleskow has received consultant fees from NinePoint Medical, Boston Scientific , Olympus, Fuji, Medtronic, and CSA. Dr Sethi has received consultant fees from Boston Scientific, Olympus, and Fujifilm. Dr Smith has received consultant fees from NinePoint Medical. Dr Wallace has received research grants from NinePoint Medical, Fujifilm , Boston Scientific, Olympus, Medtronic, NinePoint Medical, Cosmo/Aries Pharmaceuticals; consultant fees from Virgo Inc, Cosmo/Aries Pharmaceuticals, Anx Robotica (2019), Covidien , and GI Supply; he holds stock options in Virgo Inc and has consulted on behalf of Mayo Clinic for GI Supply (2018), Endokey, Endostart, Boston Scientific, and Microtek; he has received general payments from Synergy Pharmaceuticals, Boston Scientific, and Cook Medical . Dr Tearney has received consultant fees and royalties from NinePoint Medical and research support from Boston Scientific, iLumen, and CN USA Biotech Holdings. Dr Vieth has received honoraria for lecturing from Falk, Shire, and Olympus. Dr Bergman has received research support from NinePoint Medical and speaker fees from Fujifilm. All other authors disclosed no financial relationships. The collaboration project is financed by the Ministry of Economic Affairs of the Netherlands by means of the PPP Allowance made available by the Top Sector Life Sciences & Health to Academic Medical Center Amsterdam to stimulate public-private partnerships.

Funders	Funder number
European Union's Horizon 2020 - Research and Innovation Framework Programme	721766
Ministerie van Economische Zaken en Klimaat

Access to Document

10.1016/j.gie.2020.07.052

Cite this

Struyvenberg, M. R., de Groof, A. J., Fonollà, R., van der Sommen, F., de With, P. H. N., Schoon, E. J., Weusten, B. L. A. M., Leggett, C. L., Kahn, A., Trindade, A. J., Ganguly, E. K., Konda, V. J. A., Lightdale, C. J., Pleskow, D. K., Sethi, A., Smith, M. S., Wallace, M. B., Wolfsen, H. C., Tearney, G. J., ... Bergman, J. J. (2021). Prospective development and validation of a volumetric laser endomicroscopy computer algorithm for detection of Barrett’s neoplasia. Gastrointestinal Endoscopy, 93(4), 871-879. https://doi.org/10.1016/j.gie.2020.07.052

@article{25a51d5bf7e14f23a3780f6f41107c0a,

title = "Prospective development and validation of a volumetric laser endomicroscopy computer algorithm for detection of Barrett{\textquoteright}s neoplasia",

abstract = "Background and AimsVolumetric laser endomicroscopy (VLE) is an advanced imaging modality used to detect Barrett{\textquoteright}s esophagus (BE) dysplasia. However, real-time interpretation of VLE scans is complex and time-consuming. Computer-aided detection (CAD) may aid in the process of VLE image interpretation. Our aim was to train and validate a CAD algorithm for VLE-based detection of BE neoplasia.MethodsThe multicenter, VLE PREDICT study, prospectively enrolled 47 BE patients. In total, 229 nondysplastic BE, and 89 neoplastic (HGD/EAC) targets were laser marked under VLE guidance and subsequently biopsied for histological diagnosis. Deep convolutional neural networks were used to construct a CAD algorithm for differentiation between nondysplastic and neoplastic BE tissue. The CAD algorithm was trained on a set consisting of the first 22 patients (134 NDBE and 38 neoplastic targets) and validated on a separate test set of patients 23 to 47 (95 NDBE and 51 neoplastic targets). Finally, algorithm performance was benchmarked against the performance of 10 VLE experts.ResultsUsing the Training set to construct the algorithm resulted in an accuracy of 92%, sensitivity of 95% and specificity of 92%. When performance was assessed on the Test set, accuracy, sensitivity, and specificity were 85%, 91%, and 82%, respectively. The algorithm outperformed all 10 VLE experts, who demonstrated an overall accuracy of 77%, sensitivity of 70%, and specificity of 81%.ConclusionsWe developed, validated, and benchmarked a VLE CAD algorithm for detection of BE neoplasia using prospectively collected and biopsy-correlated VLE targets. The algorithm detected neoplasia with high accuracy and outperformed 10 VLE experts.",

author = "Struyvenberg, {Maarten R.} and {de Groof}, {Albert J.} and Roger Fonoll{\`a} and {van der Sommen}, Fons and {de With}, {Peter H.N.} and Schoon, {Erik J.} and Weusten, {Bas L.A.M.} and Leggett, {Cadman L.} and Allon Kahn and Trindade, {Arvind J.} and Ganguly, {Eric K.} and Konda, {Vani J.A.} and Lightdale, {Charles J.} and Pleskow, {Douglas K.} and Amrita Sethi and Smith, {Michael S.} and Wallace, {Michael B.} and Wolfsen, {Herbert C.} and Tearney, {Gary J.} and Meijer, {Sybren L.} and Michael Vieth and Pouw, {Roos E.} and Curvers, {Wouter L.} and Bergman, {Jacques J.}",

year = "2021",

month = apr,

doi = "10.1016/j.gie.2020.07.052",

language = "English",

volume = "93",

pages = "871--879",

journal = "Gastrointestinal Endoscopy",

issn = "0016-5107",

publisher = "Elsevier",

number = "4",

}

Struyvenberg, MR, de Groof, AJ, Fonollà, R, van der Sommen, F , de With, PHN, Schoon, EJ, Weusten, BLAM, Leggett, CL, Kahn, A, Trindade, AJ, Ganguly, EK, Konda, VJA, Lightdale, CJ, Pleskow, DK, Sethi, A, Smith, MS, Wallace, MB, Wolfsen, HC, Tearney, GJ, Meijer, SL, Vieth, M, Pouw, RE, Curvers, WL & Bergman, JJ 2021, 'Prospective development and validation of a volumetric laser endomicroscopy computer algorithm for detection of Barrett’s neoplasia', Gastrointestinal Endoscopy, vol. 93, no. 4, pp. 871-879. https://doi.org/10.1016/j.gie.2020.07.052

TY - JOUR

T1 - Prospective development and validation of a volumetric laser endomicroscopy computer algorithm for detection of Barrett’s neoplasia

AU - Struyvenberg, Maarten R.

AU - de Groof, Albert J.

AU - Fonollà, Roger

AU - van der Sommen, Fons

AU - de With, Peter H.N.

AU - Schoon, Erik J.

AU - Weusten, Bas L.A.M.

AU - Leggett, Cadman L.

AU - Kahn, Allon

AU - Trindade, Arvind J.

AU - Ganguly, Eric K.

AU - Konda, Vani J.A.

AU - Lightdale, Charles J.

AU - Pleskow, Douglas K.

AU - Sethi, Amrita

AU - Smith, Michael S.

AU - Wallace, Michael B.

AU - Wolfsen, Herbert C.

AU - Tearney, Gary J.

AU - Meijer, Sybren L.

AU - Vieth, Michael

AU - Pouw, Roos E.

AU - Curvers, Wouter L.

AU - Bergman, Jacques J.

PY - 2021/4

Y1 - 2021/4

N2 - Background and AimsVolumetric laser endomicroscopy (VLE) is an advanced imaging modality used to detect Barrett’s esophagus (BE) dysplasia. However, real-time interpretation of VLE scans is complex and time-consuming. Computer-aided detection (CAD) may aid in the process of VLE image interpretation. Our aim was to train and validate a CAD algorithm for VLE-based detection of BE neoplasia.MethodsThe multicenter, VLE PREDICT study, prospectively enrolled 47 BE patients. In total, 229 nondysplastic BE, and 89 neoplastic (HGD/EAC) targets were laser marked under VLE guidance and subsequently biopsied for histological diagnosis. Deep convolutional neural networks were used to construct a CAD algorithm for differentiation between nondysplastic and neoplastic BE tissue. The CAD algorithm was trained on a set consisting of the first 22 patients (134 NDBE and 38 neoplastic targets) and validated on a separate test set of patients 23 to 47 (95 NDBE and 51 neoplastic targets). Finally, algorithm performance was benchmarked against the performance of 10 VLE experts.ResultsUsing the Training set to construct the algorithm resulted in an accuracy of 92%, sensitivity of 95% and specificity of 92%. When performance was assessed on the Test set, accuracy, sensitivity, and specificity were 85%, 91%, and 82%, respectively. The algorithm outperformed all 10 VLE experts, who demonstrated an overall accuracy of 77%, sensitivity of 70%, and specificity of 81%.ConclusionsWe developed, validated, and benchmarked a VLE CAD algorithm for detection of BE neoplasia using prospectively collected and biopsy-correlated VLE targets. The algorithm detected neoplasia with high accuracy and outperformed 10 VLE experts.

AB - Background and AimsVolumetric laser endomicroscopy (VLE) is an advanced imaging modality used to detect Barrett’s esophagus (BE) dysplasia. However, real-time interpretation of VLE scans is complex and time-consuming. Computer-aided detection (CAD) may aid in the process of VLE image interpretation. Our aim was to train and validate a CAD algorithm for VLE-based detection of BE neoplasia.MethodsThe multicenter, VLE PREDICT study, prospectively enrolled 47 BE patients. In total, 229 nondysplastic BE, and 89 neoplastic (HGD/EAC) targets were laser marked under VLE guidance and subsequently biopsied for histological diagnosis. Deep convolutional neural networks were used to construct a CAD algorithm for differentiation between nondysplastic and neoplastic BE tissue. The CAD algorithm was trained on a set consisting of the first 22 patients (134 NDBE and 38 neoplastic targets) and validated on a separate test set of patients 23 to 47 (95 NDBE and 51 neoplastic targets). Finally, algorithm performance was benchmarked against the performance of 10 VLE experts.ResultsUsing the Training set to construct the algorithm resulted in an accuracy of 92%, sensitivity of 95% and specificity of 92%. When performance was assessed on the Test set, accuracy, sensitivity, and specificity were 85%, 91%, and 82%, respectively. The algorithm outperformed all 10 VLE experts, who demonstrated an overall accuracy of 77%, sensitivity of 70%, and specificity of 81%.ConclusionsWe developed, validated, and benchmarked a VLE CAD algorithm for detection of BE neoplasia using prospectively collected and biopsy-correlated VLE targets. The algorithm detected neoplasia with high accuracy and outperformed 10 VLE experts.

UR - http://www.scopus.com/inward/record.url?scp=85095854894&partnerID=8YFLogxK

U2 - 10.1016/j.gie.2020.07.052

DO - 10.1016/j.gie.2020.07.052

M3 - Article

C2 - 32735947

SN - 0016-5107

VL - 93

SP - 871

EP - 879

JO - Gastrointestinal Endoscopy

JF - Gastrointestinal Endoscopy

IS - 4

ER -

Prospective development and validation of a volumetric laser endomicroscopy computer algorithm for detection of Barrett’s neoplasia

Abstract

Funding

Access to Document

Other files and links

Fingerprint

Cite this