Consistency in contouring of organs at risk by artificial intelligence vs oncologists in head and neck cancer patients

Camilla Panduro Nielsen; Ebbe Laugaard Lorenzen; Kenneth Jensen; Nis Sarup; Carsten Brink; Bob Smulders; Anne Ivalu Sander Holm; Eva Samsøe; Martin Skovmos Nielsen; Patrik Sibolt; Peter Sandegaard Skyt; Ulrik Vindelev Elstrøm; Jørgen Johansen; Ruta Zukauskaite; Jesper Grau Eriksen; Mohammad Farhadi; Maria Andersen; Christian Maare; Jens Overgaard; Cai Grau; Jeppe Friborg; Christian Rønn Hansen

doi:10.1080/0284186X.2023.2256958

Consistency in contouring of organs at risk by artificial intelligence vs oncologists in head and neck cancer patients

Camilla Panduro Nielsen^*, Ebbe Laugaard Lorenzen, Kenneth Jensen, Nis Sarup, Carsten Brink, Bob Smulders, Anne Ivalu Sander Holm, Eva Samsøe, Martin Skovmos Nielsen, Patrik Sibolt, Peter Sandegaard Skyt, Ulrik Vindelev Elstrøm, Jørgen Johansen, Ruta Zukauskaite, Jesper Grau Eriksen, Mohammad Farhadi, Maria Andersen, Christian Maare, Jens Overgaard, Cai GrauJeppe Friborg, Christian Rønn Hansen

^*Corresponding author af dette arbejde

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avis › Tidsskriftartikel › Forskning › peer review

Abstract

BACKGROUND: In the Danish Head and Neck Cancer Group (DAHANCA) 35 trial, patients are selected for proton treatment based on simulated reductions of Normal Tissue Complication Probability (NTCP) for proton compared to photon treatment at the referring departments. After inclusion in the trial, immobilization, scanning, contouring and planning are repeated at the national proton centre. The new contours could result in reduced expected NTCP gain of the proton plan, resulting in a loss of validity in the selection process. The present study evaluates if contour consistency can be improved by having access to AI (Artificial Intelligence) based contours.

MATERIALS AND METHODS: The 63 patients in the DAHANCA 35 pilot trial had a CT from the local DAHANCA centre and one from the proton centre. A nationally validated convolutional neural network, based on nnU-Net, was used to contour OARs on both scans for each patient. Using deformable image registration, local AI and oncologist contours were transferred to the proton centre scans for comparison. Consistency was calculated with the Dice Similarity Coefficient (DSC) and Mean Surface Distance (MSD), comparing contours from AI to AI and oncologist to oncologist, respectively. Two NTCP models were applied to calculate NTCP for xerostomia and dysphagia.

RESULTS: The AI contours showed significantly better consistency than the contours by oncologists. The median and interquartile range of DSC was 0.85 [0.78 - 0.90] and 0.68 [0.51 - 0.80] for AI and oncologist contours, respectively. The median and interquartile range of MSD was 0.9 mm [0.7 - 1.1] mm and 1.9 mm [1.5 - 2.6] mm for AI and oncologist contours, respectively. There was no significant difference in Δ NTCP.

CONCLUSIONS: The study showed that OAR contours made by the AI algorithm were more consistent than those made by oncologists. No significant impact on the Δ NTCP calculations could be discerned.

Originalsprog	Engelsk
Tidsskrift	Acta Oncologica
Vol/bind	62
Nummer	11
Sider (fra-til)	1418-1425
Antal sider	8
ISSN	0284-186X
DOI	https://doi.org/10.1080/0284186X.2023.2256958
Status	Udgivet - nov. 2023

Adgang til dokumentet

10.1080/0284186X.2023.2256958

Andre filer og links

Link to publication in Scopus

Citationsformater

Nielsen, C. P., Lorenzen, E. L., Jensen, K., Sarup, N., Brink, C., Smulders, B., Holm, A. I. S., Samsøe, E., Nielsen, M. S., Sibolt, P., Skyt, P. S., Elstrøm, U. V., Johansen, J., Zukauskaite, R., Eriksen, J. G., Farhadi, M., Andersen, M., Maare, C., Overgaard, J., ... Hansen, C. R. (2023). Consistency in contouring of organs at risk by artificial intelligence vs oncologists in head and neck cancer patients. Acta Oncologica, 62(11), 1418-1425. https://doi.org/10.1080/0284186X.2023.2256958

@article{e81fcd87a9fa4840a89c5ae66f59e063,

title = "Consistency in contouring of organs at risk by artificial intelligence vs oncologists in head and neck cancer patients",

abstract = "BACKGROUND: In the Danish Head and Neck Cancer Group (DAHANCA) 35 trial, patients are selected for proton treatment based on simulated reductions of Normal Tissue Complication Probability (NTCP) for proton compared to photon treatment at the referring departments. After inclusion in the trial, immobilization, scanning, contouring and planning are repeated at the national proton centre. The new contours could result in reduced expected NTCP gain of the proton plan, resulting in a loss of validity in the selection process. The present study evaluates if contour consistency can be improved by having access to AI (Artificial Intelligence) based contours.MATERIALS AND METHODS: The 63 patients in the DAHANCA 35 pilot trial had a CT from the local DAHANCA centre and one from the proton centre. A nationally validated convolutional neural network, based on nnU-Net, was used to contour OARs on both scans for each patient. Using deformable image registration, local AI and oncologist contours were transferred to the proton centre scans for comparison. Consistency was calculated with the Dice Similarity Coefficient (DSC) and Mean Surface Distance (MSD), comparing contours from AI to AI and oncologist to oncologist, respectively. Two NTCP models were applied to calculate NTCP for xerostomia and dysphagia.RESULTS: The AI contours showed significantly better consistency than the contours by oncologists. The median and interquartile range of DSC was 0.85 [0.78 - 0.90] and 0.68 [0.51 - 0.80] for AI and oncologist contours, respectively. The median and interquartile range of MSD was 0.9 mm [0.7 - 1.1] mm and 1.9 mm [1.5 - 2.6] mm for AI and oncologist contours, respectively. There was no significant difference in Δ NTCP. CONCLUSIONS: The study showed that OAR contours made by the AI algorithm were more consistent than those made by oncologists. No significant impact on the Δ NTCP calculations could be discerned. ",

keywords = "AI, contouring, head and neck cancer, organs at risk, proton treatment, Protons, Radiotherapy Planning, Computer-Assisted/methods, Artificial Intelligence, Humans, Organs at Risk, Head and Neck Neoplasms",

author = "Nielsen, {Camilla Panduro} and Lorenzen, {Ebbe Laugaard} and Kenneth Jensen and Nis Sarup and Carsten Brink and Bob Smulders and Holm, {Anne Ivalu Sander} and Eva Sams{\o}e and Nielsen, {Martin Skovmos} and Patrik Sibolt and Skyt, {Peter Sandegaard} and Elstr{\o}m, {Ulrik Vindelev} and J{\o}rgen Johansen and Ruta Zukauskaite and Eriksen, {Jesper Grau} and Mohammad Farhadi and Maria Andersen and Christian Maare and Jens Overgaard and Cai Grau and Jeppe Friborg and Hansen, {Christian R{\o}nn}",

year = "2023",

month = nov,

doi = "10.1080/0284186X.2023.2256958",

language = "English",

volume = "62",

pages = "1418--1425",

journal = "Acta Oncologica",

issn = "0284-186X",

publisher = "Taylor & Francis",

number = "11",

}

Nielsen, CP, Lorenzen, EL, Jensen, K, Sarup, N, Brink, C, Smulders, B, Holm, AIS, Samsøe, E, Nielsen, MS, Sibolt, P, Skyt, PS, Elstrøm, UV, Johansen, J, Zukauskaite, R, Eriksen, JG, Farhadi, M, Andersen, M, Maare, C, Overgaard, J , Grau, C, Friborg, J & Hansen, CR 2023, 'Consistency in contouring of organs at risk by artificial intelligence vs oncologists in head and neck cancer patients', Acta Oncologica, bind 62, nr. 11, s. 1418-1425. https://doi.org/10.1080/0284186X.2023.2256958

Consistency in contouring of organs at risk by artificial intelligence vs oncologists in head and neck cancer patients. / Nielsen, Camilla Panduro; Lorenzen, Ebbe Laugaard; Jensen, Kenneth et al.
I: Acta Oncologica, Bind 62, Nr. 11, 11.2023, s. 1418-1425.

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avis › Tidsskriftartikel › Forskning › peer review

TY - JOUR

T1 - Consistency in contouring of organs at risk by artificial intelligence vs oncologists in head and neck cancer patients

AU - Nielsen, Camilla Panduro

AU - Lorenzen, Ebbe Laugaard

AU - Jensen, Kenneth

AU - Sarup, Nis

AU - Brink, Carsten

AU - Smulders, Bob

AU - Holm, Anne Ivalu Sander

AU - Samsøe, Eva

AU - Nielsen, Martin Skovmos

AU - Sibolt, Patrik

AU - Skyt, Peter Sandegaard

AU - Elstrøm, Ulrik Vindelev

AU - Johansen, Jørgen

AU - Zukauskaite, Ruta

AU - Eriksen, Jesper Grau

AU - Farhadi, Mohammad

AU - Andersen, Maria

AU - Maare, Christian

AU - Overgaard, Jens

AU - Grau, Cai

AU - Friborg, Jeppe

AU - Hansen, Christian Rønn

PY - 2023/11

Y1 - 2023/11

N2 - BACKGROUND: In the Danish Head and Neck Cancer Group (DAHANCA) 35 trial, patients are selected for proton treatment based on simulated reductions of Normal Tissue Complication Probability (NTCP) for proton compared to photon treatment at the referring departments. After inclusion in the trial, immobilization, scanning, contouring and planning are repeated at the national proton centre. The new contours could result in reduced expected NTCP gain of the proton plan, resulting in a loss of validity in the selection process. The present study evaluates if contour consistency can be improved by having access to AI (Artificial Intelligence) based contours.MATERIALS AND METHODS: The 63 patients in the DAHANCA 35 pilot trial had a CT from the local DAHANCA centre and one from the proton centre. A nationally validated convolutional neural network, based on nnU-Net, was used to contour OARs on both scans for each patient. Using deformable image registration, local AI and oncologist contours were transferred to the proton centre scans for comparison. Consistency was calculated with the Dice Similarity Coefficient (DSC) and Mean Surface Distance (MSD), comparing contours from AI to AI and oncologist to oncologist, respectively. Two NTCP models were applied to calculate NTCP for xerostomia and dysphagia.RESULTS: The AI contours showed significantly better consistency than the contours by oncologists. The median and interquartile range of DSC was 0.85 [0.78 - 0.90] and 0.68 [0.51 - 0.80] for AI and oncologist contours, respectively. The median and interquartile range of MSD was 0.9 mm [0.7 - 1.1] mm and 1.9 mm [1.5 - 2.6] mm for AI and oncologist contours, respectively. There was no significant difference in Δ NTCP. CONCLUSIONS: The study showed that OAR contours made by the AI algorithm were more consistent than those made by oncologists. No significant impact on the Δ NTCP calculations could be discerned.

AB - BACKGROUND: In the Danish Head and Neck Cancer Group (DAHANCA) 35 trial, patients are selected for proton treatment based on simulated reductions of Normal Tissue Complication Probability (NTCP) for proton compared to photon treatment at the referring departments. After inclusion in the trial, immobilization, scanning, contouring and planning are repeated at the national proton centre. The new contours could result in reduced expected NTCP gain of the proton plan, resulting in a loss of validity in the selection process. The present study evaluates if contour consistency can be improved by having access to AI (Artificial Intelligence) based contours.MATERIALS AND METHODS: The 63 patients in the DAHANCA 35 pilot trial had a CT from the local DAHANCA centre and one from the proton centre. A nationally validated convolutional neural network, based on nnU-Net, was used to contour OARs on both scans for each patient. Using deformable image registration, local AI and oncologist contours were transferred to the proton centre scans for comparison. Consistency was calculated with the Dice Similarity Coefficient (DSC) and Mean Surface Distance (MSD), comparing contours from AI to AI and oncologist to oncologist, respectively. Two NTCP models were applied to calculate NTCP for xerostomia and dysphagia.RESULTS: The AI contours showed significantly better consistency than the contours by oncologists. The median and interquartile range of DSC was 0.85 [0.78 - 0.90] and 0.68 [0.51 - 0.80] for AI and oncologist contours, respectively. The median and interquartile range of MSD was 0.9 mm [0.7 - 1.1] mm and 1.9 mm [1.5 - 2.6] mm for AI and oncologist contours, respectively. There was no significant difference in Δ NTCP. CONCLUSIONS: The study showed that OAR contours made by the AI algorithm were more consistent than those made by oncologists. No significant impact on the Δ NTCP calculations could be discerned.

KW - AI

KW - contouring

KW - head and neck cancer

KW - organs at risk

KW - proton treatment

KW - Protons

KW - Radiotherapy Planning, Computer-Assisted/methods

KW - Artificial Intelligence

KW - Humans

KW - Organs at Risk

KW - Head and Neck Neoplasms

UR - http://www.scopus.com/inward/record.url?scp=85170701654&partnerID=8YFLogxK

U2 - 10.1080/0284186X.2023.2256958

DO - 10.1080/0284186X.2023.2256958

M3 - Journal article

C2 - 37703300

SN - 0284-186X

VL - 62

SP - 1418

EP - 1425

JO - Acta Oncologica

JF - Acta Oncologica

IS - 11

ER -

Consistency in contouring of organs at risk by artificial intelligence vs oncologists in head and neck cancer patients

Abstract

Adgang til dokumentet

Andre filer og links

Fingeraftryk

Citationsformater