Evaluating Automatic Segmentation for Swallowing-Related Organs for Head and Neck Cancer

Purpose: To evaluate the accuracy of deep-learning-based auto-segmentation of the superior constrictor, middle constrictor, inferior constrictor, and larynx in comparison with a traditional multi-atlas-based method. Methods and Materials: One hundred and five computed tomography image datasets from...

Full description

Bibliographic Details
Main Authors: Yimin Li MD, Shyam Rao MD, Wen Chen MD, Soheila F. Azghadi MD, Ky Nam Bao Nguyen MD, Angel Moran MD, Brittni M Usera MD, Brandon A Dyer MD, Lu Shang PhD, Quan Chen PhD, Yi Rong PhD
Format: Article
Language:English
Published: SAGE Publishing 2022-07-01
Series:Technology in Cancer Research & Treatment
Online Access:https://doi.org/10.1177/15330338221105724
_version_ 1811192287715131392
author Yimin Li MD
Shyam Rao MD
Wen Chen MD
Soheila F. Azghadi MD
Ky Nam Bao Nguyen MD
Angel Moran MD
Brittni M Usera MD
Brandon A Dyer MD
Lu Shang PhD
Quan Chen PhD
Yi Rong PhD
author_facet Yimin Li MD
Shyam Rao MD
Wen Chen MD
Soheila F. Azghadi MD
Ky Nam Bao Nguyen MD
Angel Moran MD
Brittni M Usera MD
Brandon A Dyer MD
Lu Shang PhD
Quan Chen PhD
Yi Rong PhD
author_sort Yimin Li MD
collection DOAJ
description Purpose: To evaluate the accuracy of deep-learning-based auto-segmentation of the superior constrictor, middle constrictor, inferior constrictor, and larynx in comparison with a traditional multi-atlas-based method. Methods and Materials: One hundred and five computed tomography image datasets from 83 head and neck cancer patients were retrospectively collected and the superior constrictor, middle constrictor, inferior constrictor, and larynx were analyzed for deep-learning versus multi-atlas-based segmentation. Eighty-three computed tomography images (40 diagnostic computed tomography and 43 planning computed tomography) were used for training the convolutional neural network, and for atlas-based model training. The remaining 22 computed tomography datasets were used for validation of the atlas-based auto-segmentation versus deep-learning-based auto-segmentation contours, both of which were compared with the corresponding manual contours. Quantitative measures included Dice similarity coefficient, recall, precision, Hausdorff distance, 95th percentile of Hausdorff distance, and mean surface distance. Dosimetric differences between the auto-generated contours and manual contours were evaluated. Subjective evaluation was obtained from 3 clinical observers to blindly score the autosegmented structures based on the percentage of slices that require manual modification. Results: The deep-learning-based auto-segmentation versus atlas-based auto-segmentation results were compared for the superior constrictor, middle constrictor, inferior constrictor, and larynx. The mean Dice similarity coefficient values for the 4 structures were 0.67, 0.60, 0.65, and 0.84 for deep-learning-based auto-segmentation, whereas atlas-based auto-segmentation has Dice similarity coefficient results at 0.45, 0.36, 0.50, and 0.70, respectively. The mean 95th percentile of Hausdorff distance (cm) for the 4 structures were 0.41, 0.57, 0.59, and 0.54 for deep-learning-based auto-segmentation, but 0.78, 0.95, 0.96, and 1.23 for atlas-based auto-segmentation results, respectively. Similar mean dose differences were obtained from the 2 sets of autosegmented contours compared to manual contours. The dose–volume discrepancies and the average modification rates were higher with the atlas-based auto-segmentation contours. Conclusion: Swallowing-related structures are more accurately generated with DL-based versus atlas-based segmentation when compared with manual contours.
first_indexed 2024-04-11T23:50:07Z
format Article
id doaj.art-b9e6feef9e3d42af95d455f769118f79
institution Directory Open Access Journal
issn 1533-0338
language English
last_indexed 2024-04-11T23:50:07Z
publishDate 2022-07-01
publisher SAGE Publishing
record_format Article
series Technology in Cancer Research & Treatment
spelling doaj.art-b9e6feef9e3d42af95d455f769118f792022-12-22T03:56:31ZengSAGE PublishingTechnology in Cancer Research & Treatment1533-03382022-07-012110.1177/15330338221105724Evaluating Automatic Segmentation for Swallowing-Related Organs for Head and Neck CancerYimin Li MD0Shyam Rao MD1Wen Chen MD2Soheila F. Azghadi MD3Ky Nam Bao Nguyen MD4Angel Moran MD5Brittni M Usera MD6Brandon A Dyer MD7Lu Shang PhD8Quan Chen PhD9Yi Rong PhD10 Department of Radiation Oncology, , Sacramento, CA, USA Department of Radiation Oncology, , Sacramento, CA, USA Department of Radiation Oncology, , Changsha, China Department of Radiation Oncology, , Sacramento, CA, USA Department of Radiation Oncology, , Sacramento, CA, USA Department of Radiation Oncology, , Sacramento, CA, USA Department of Radiation Oncology, , Sacramento, CA, USA Department of Radiation Oncology, , Portland, OR, USA Department of Radiation Oncology, , Sacramento, CA, USA Department of Radiation Oncology, City of Hope comprehensive Cancer Center, Duarte, CA, USA Department of Radiation Oncology, , Phoenix, AZ, USAPurpose: To evaluate the accuracy of deep-learning-based auto-segmentation of the superior constrictor, middle constrictor, inferior constrictor, and larynx in comparison with a traditional multi-atlas-based method. Methods and Materials: One hundred and five computed tomography image datasets from 83 head and neck cancer patients were retrospectively collected and the superior constrictor, middle constrictor, inferior constrictor, and larynx were analyzed for deep-learning versus multi-atlas-based segmentation. Eighty-three computed tomography images (40 diagnostic computed tomography and 43 planning computed tomography) were used for training the convolutional neural network, and for atlas-based model training. The remaining 22 computed tomography datasets were used for validation of the atlas-based auto-segmentation versus deep-learning-based auto-segmentation contours, both of which were compared with the corresponding manual contours. Quantitative measures included Dice similarity coefficient, recall, precision, Hausdorff distance, 95th percentile of Hausdorff distance, and mean surface distance. Dosimetric differences between the auto-generated contours and manual contours were evaluated. Subjective evaluation was obtained from 3 clinical observers to blindly score the autosegmented structures based on the percentage of slices that require manual modification. Results: The deep-learning-based auto-segmentation versus atlas-based auto-segmentation results were compared for the superior constrictor, middle constrictor, inferior constrictor, and larynx. The mean Dice similarity coefficient values for the 4 structures were 0.67, 0.60, 0.65, and 0.84 for deep-learning-based auto-segmentation, whereas atlas-based auto-segmentation has Dice similarity coefficient results at 0.45, 0.36, 0.50, and 0.70, respectively. The mean 95th percentile of Hausdorff distance (cm) for the 4 structures were 0.41, 0.57, 0.59, and 0.54 for deep-learning-based auto-segmentation, but 0.78, 0.95, 0.96, and 1.23 for atlas-based auto-segmentation results, respectively. Similar mean dose differences were obtained from the 2 sets of autosegmented contours compared to manual contours. The dose–volume discrepancies and the average modification rates were higher with the atlas-based auto-segmentation contours. Conclusion: Swallowing-related structures are more accurately generated with DL-based versus atlas-based segmentation when compared with manual contours.https://doi.org/10.1177/15330338221105724
spellingShingle Yimin Li MD
Shyam Rao MD
Wen Chen MD
Soheila F. Azghadi MD
Ky Nam Bao Nguyen MD
Angel Moran MD
Brittni M Usera MD
Brandon A Dyer MD
Lu Shang PhD
Quan Chen PhD
Yi Rong PhD
Evaluating Automatic Segmentation for Swallowing-Related Organs for Head and Neck Cancer
Technology in Cancer Research & Treatment
title Evaluating Automatic Segmentation for Swallowing-Related Organs for Head and Neck Cancer
title_full Evaluating Automatic Segmentation for Swallowing-Related Organs for Head and Neck Cancer
title_fullStr Evaluating Automatic Segmentation for Swallowing-Related Organs for Head and Neck Cancer
title_full_unstemmed Evaluating Automatic Segmentation for Swallowing-Related Organs for Head and Neck Cancer
title_short Evaluating Automatic Segmentation for Swallowing-Related Organs for Head and Neck Cancer
title_sort evaluating automatic segmentation for swallowing related organs for head and neck cancer
url https://doi.org/10.1177/15330338221105724
work_keys_str_mv AT yiminlimd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer
AT shyamraomd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer
AT wenchenmd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer
AT soheilafazghadimd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer
AT kynambaonguyenmd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer
AT angelmoranmd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer
AT brittnimuseramd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer
AT brandonadyermd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer
AT lushangphd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer
AT quanchenphd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer
AT yirongphd evaluatingautomaticsegmentationforswallowingrelatedorgansforheadandneckcancer