Automatic selection of lung cancer patients for adaptive radiotherapy using cone-beam CT imaging

M. Van den Bosch; M. Ollers; B. Reymen; W. Van Elmpt

doi:10.1016/j.phro.2017.02.005

Automatic selection of lung cancer patients for adaptive radiotherapy using cone-beam CT imaging

M. Van den Bosch, M. Ollers, B. Reymen, W. Van Elmpt^*

^*Corresponding author for this work

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

Abstract

Background and purpose: Anatomical changes in non-small cell lung cancer (NSCLC) patients may lead to unacceptable treatment results, requiring adaptive radiotherapy (ART). The method proposed in this study describes the proof-of-principle to automatically select patients eligible for ART.Materials and methods: A method was developed flagging patients potentially requiring replanning using changes in density information between the planning CT and cone-beam CT (CBCT) scan. Potential candidates were defined based on anatomical changes in the primary tumour, mediastinum and a region including the lungs. In total, CBCTs of 70 NSCLC patients were retrospectively scored by two independent experts for the need of replanning using the clinical guidelines. Expert evaluation was used as benchmark for the automatic method and accuracy was quantified using a leave-one-out cross-validation strategy.Results: The automatic method correctly selected 79%, 71%, and 89% of the pre-defined candidates based on anatomical change in the tumour, mediastinum and lungs, respectively. The false positive rate was 27%, 9%, and 19% for these regions, respectively. If a single criterion ('flag') per patient is sufficient for replanning, the overall sensitivity was 89% with a false positive rate of 34%. Most common reasons for false positives were changes in breathing pattern or anatomical shifts outside the evaluated region. Excluding these reasons, the false positive rate dropped to 21% and the accuracy was almost similar to expert evaluation: overall sensitivity of approx. 95% with a false positive rate below 19%.Conclusion: Our method based on daily CBCT allows automatic selection of patients that should be investigated for treatment adaptation. (C) 2017 The Authors. Published by Elsevier Ireland Ltd on behalf of European Society of Radiotherapy & Oncology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Original language	English
Pages (from-to)	21-27
Number of pages	7
Journal	Physics & Imaging in Radiation Oncology
Volume	1
DOIs	https://doi.org/10.1016/j.phro.2017.02.005
Publication status	Published - 1 Jan 2017

Keywords

Non-small cell lung cancer
Adaptive radiotherapy
Replanning
Automatic selection method
Cone-beam CT

Access to Document

10.1016/j.phro.2017.02.005Licence: CC BY-NC-ND

Cite this

@article{510faf10e123466ebb0471a826d0fa8b,

title = "Automatic selection of lung cancer patients for adaptive radiotherapy using cone-beam CT imaging",

abstract = "Background and purpose: Anatomical changes in non-small cell lung cancer (NSCLC) patients may lead to unacceptable treatment results, requiring adaptive radiotherapy (ART). The method proposed in this study describes the proof-of-principle to automatically select patients eligible for ART.Materials and methods: A method was developed flagging patients potentially requiring replanning using changes in density information between the planning CT and cone-beam CT (CBCT) scan. Potential candidates were defined based on anatomical changes in the primary tumour, mediastinum and a region including the lungs. In total, CBCTs of 70 NSCLC patients were retrospectively scored by two independent experts for the need of replanning using the clinical guidelines. Expert evaluation was used as benchmark for the automatic method and accuracy was quantified using a leave-one-out cross-validation strategy.Results: The automatic method correctly selected 79%, 71%, and 89% of the pre-defined candidates based on anatomical change in the tumour, mediastinum and lungs, respectively. The false positive rate was 27%, 9%, and 19% for these regions, respectively. If a single criterion ('flag') per patient is sufficient for replanning, the overall sensitivity was 89% with a false positive rate of 34%. Most common reasons for false positives were changes in breathing pattern or anatomical shifts outside the evaluated region. Excluding these reasons, the false positive rate dropped to 21% and the accuracy was almost similar to expert evaluation: overall sensitivity of approx. 95% with a false positive rate below 19%.Conclusion: Our method based on daily CBCT allows automatic selection of patients that should be investigated for treatment adaptation. (C) 2017 The Authors. Published by Elsevier Ireland Ltd on behalf of European Society of Radiotherapy & Oncology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).",

keywords = "Non-small cell lung cancer, Adaptive radiotherapy, Replanning, Automatic selection method, Cone-beam CT",

author = "{Van den Bosch}, M. and M. Ollers and B. Reymen and {Van Elmpt}, W.",

year = "2017",

month = jan,

day = "1",

doi = "10.1016/j.phro.2017.02.005",

language = "English",

volume = "1",

pages = "21--27",

journal = "Physics & Imaging in Radiation Oncology",

issn = "2405-6316",

publisher = "Elsevier Ireland Ltd",

}

TY - JOUR

T1 - Automatic selection of lung cancer patients for adaptive radiotherapy using cone-beam CT imaging

AU - Van den Bosch, M.

AU - Ollers, M.

AU - Reymen, B.

AU - Van Elmpt, W.

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Background and purpose: Anatomical changes in non-small cell lung cancer (NSCLC) patients may lead to unacceptable treatment results, requiring adaptive radiotherapy (ART). The method proposed in this study describes the proof-of-principle to automatically select patients eligible for ART.Materials and methods: A method was developed flagging patients potentially requiring replanning using changes in density information between the planning CT and cone-beam CT (CBCT) scan. Potential candidates were defined based on anatomical changes in the primary tumour, mediastinum and a region including the lungs. In total, CBCTs of 70 NSCLC patients were retrospectively scored by two independent experts for the need of replanning using the clinical guidelines. Expert evaluation was used as benchmark for the automatic method and accuracy was quantified using a leave-one-out cross-validation strategy.Results: The automatic method correctly selected 79%, 71%, and 89% of the pre-defined candidates based on anatomical change in the tumour, mediastinum and lungs, respectively. The false positive rate was 27%, 9%, and 19% for these regions, respectively. If a single criterion ('flag') per patient is sufficient for replanning, the overall sensitivity was 89% with a false positive rate of 34%. Most common reasons for false positives were changes in breathing pattern or anatomical shifts outside the evaluated region. Excluding these reasons, the false positive rate dropped to 21% and the accuracy was almost similar to expert evaluation: overall sensitivity of approx. 95% with a false positive rate below 19%.Conclusion: Our method based on daily CBCT allows automatic selection of patients that should be investigated for treatment adaptation. (C) 2017 The Authors. Published by Elsevier Ireland Ltd on behalf of European Society of Radiotherapy & Oncology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

AB - Background and purpose: Anatomical changes in non-small cell lung cancer (NSCLC) patients may lead to unacceptable treatment results, requiring adaptive radiotherapy (ART). The method proposed in this study describes the proof-of-principle to automatically select patients eligible for ART.Materials and methods: A method was developed flagging patients potentially requiring replanning using changes in density information between the planning CT and cone-beam CT (CBCT) scan. Potential candidates were defined based on anatomical changes in the primary tumour, mediastinum and a region including the lungs. In total, CBCTs of 70 NSCLC patients were retrospectively scored by two independent experts for the need of replanning using the clinical guidelines. Expert evaluation was used as benchmark for the automatic method and accuracy was quantified using a leave-one-out cross-validation strategy.Results: The automatic method correctly selected 79%, 71%, and 89% of the pre-defined candidates based on anatomical change in the tumour, mediastinum and lungs, respectively. The false positive rate was 27%, 9%, and 19% for these regions, respectively. If a single criterion ('flag') per patient is sufficient for replanning, the overall sensitivity was 89% with a false positive rate of 34%. Most common reasons for false positives were changes in breathing pattern or anatomical shifts outside the evaluated region. Excluding these reasons, the false positive rate dropped to 21% and the accuracy was almost similar to expert evaluation: overall sensitivity of approx. 95% with a false positive rate below 19%.Conclusion: Our method based on daily CBCT allows automatic selection of patients that should be investigated for treatment adaptation. (C) 2017 The Authors. Published by Elsevier Ireland Ltd on behalf of European Society of Radiotherapy & Oncology. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

KW - Non-small cell lung cancer

KW - Adaptive radiotherapy

KW - Replanning

KW - Automatic selection method

KW - Cone-beam CT

U2 - 10.1016/j.phro.2017.02.005

DO - 10.1016/j.phro.2017.02.005

M3 - Article

SN - 2405-6316

VL - 1

SP - 21

EP - 27

JO - Physics & Imaging in Radiation Oncology

JF - Physics & Imaging in Radiation Oncology

ER -