Deep learning for the fully automated segmentation of the inner ear on MRI
Abstract Segmentation of anatomical structures is valuable in a variety of tasks, including 3D visualization, surgical planning, and quantitative image analysis. Manual segmentation is time-consuming and deals with intra and inter-observer variability. To develop a deep-learning approach for the ful...
| Main Authors: | , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2021-02-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-021-82289-y |
| _version_ | 1818843245107478528 |
|---|---|
| author | Akshayaa Vaidyanathan Marly F. J. A. van der Lubbe Ralph T. H. Leijenaar Marc van Hoof Fadila Zerka Benjamin Miraglio Sergey Primakov Alida A. Postma Tjasse D. Bruintjes Monique A. L. Bilderbeek Hammer Sebastiaan Patrick F. M. Dammeijer Vincent van Rompaey Henry C. Woodruff Wim Vos Seán Walsh Raymond van de Berg Philippe Lambin |
| author_facet | Akshayaa Vaidyanathan Marly F. J. A. van der Lubbe Ralph T. H. Leijenaar Marc van Hoof Fadila Zerka Benjamin Miraglio Sergey Primakov Alida A. Postma Tjasse D. Bruintjes Monique A. L. Bilderbeek Hammer Sebastiaan Patrick F. M. Dammeijer Vincent van Rompaey Henry C. Woodruff Wim Vos Seán Walsh Raymond van de Berg Philippe Lambin |
| author_sort | Akshayaa Vaidyanathan |
| collection | DOAJ |
| description | Abstract Segmentation of anatomical structures is valuable in a variety of tasks, including 3D visualization, surgical planning, and quantitative image analysis. Manual segmentation is time-consuming and deals with intra and inter-observer variability. To develop a deep-learning approach for the fully automated segmentation of the inner ear in MRI, a 3D U-net was trained on 944 MRI scans with manually segmented inner ears as reference standard. The model was validated on an independent, multicentric dataset consisting of 177 MRI scans from three different centers. The model was also evaluated on a clinical validation set containing eight MRI scans with severe changes in the morphology of the labyrinth. The 3D U-net model showed precise Dice Similarity Coefficient scores (mean DSC-0.8790) with a high True Positive Rate (91.5%) and low False Discovery Rate and False Negative Rates (14.8% and 8.49% respectively) across images from three different centers. The model proved to perform well with a DSC of 0.8768 on the clinical validation dataset. The proposed auto-segmentation model is equivalent to human readers and is a reliable, consistent, and efficient method for inner ear segmentation, which can be used in a variety of clinical applications such as surgical planning and quantitative image analysis. |
| first_indexed | 2024-12-19T04:54:48Z |
| format | Article |
| id | doaj.art-35634f516ba54b2385d7644464b85a80 |
| institution | Directory Open Access Journal |
| issn | 2045-2322 |
| language | English |
| last_indexed | 2024-12-19T04:54:48Z |
| publishDate | 2021-02-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj.art-35634f516ba54b2385d7644464b85a802022-12-21T20:35:16ZengNature PortfolioScientific Reports2045-23222021-02-0111111410.1038/s41598-021-82289-yDeep learning for the fully automated segmentation of the inner ear on MRIAkshayaa Vaidyanathan0Marly F. J. A. van der Lubbe1Ralph T. H. Leijenaar2Marc van Hoof3Fadila Zerka4Benjamin Miraglio5Sergey Primakov6Alida A. Postma7Tjasse D. Bruintjes8Monique A. L. Bilderbeek9Hammer Sebastiaan10Patrick F. M. Dammeijer11Vincent van Rompaey12Henry C. Woodruff13Wim Vos14Seán Walsh15Raymond van de Berg16Philippe Lambin17The D-Lab, Department of Precision Medicine, GROW-School for Oncology, Maastricht UniversityDepartment of Otolaryngology and Head and Neck Surgery, Maastricht University Medical CenterOncoradiomics SADepartment of Otolaryngology and Head and Neck Surgery, Maastricht University Medical CenterOncoradiomics SAOncoradiomics SAThe D-Lab, Department of Precision Medicine, GROW-School for Oncology, Maastricht UniversityDepartment of Radiology and Nuclear Medicine, Maastricht University Medical CenterDepartment of Otorhinolaryngology, Gelre HospitalDepartment of Radiology, Viecuri Medical CenterHaga Hospital, RadiologyDepartment of Otorhinolaryngology, Viecuri Medical CenterDepartment of Otorhinolaryngology and Head & Neck Surgery, Antwerp University HospitalThe D-Lab, Department of Precision Medicine, GROW-School for Oncology, Maastricht UniversityOncoradiomics SAOncoradiomics SADepartment of Otolaryngology and Head and Neck Surgery, Maastricht University Medical CenterThe D-Lab, Department of Precision Medicine, GROW-School for Oncology, Maastricht UniversityAbstract Segmentation of anatomical structures is valuable in a variety of tasks, including 3D visualization, surgical planning, and quantitative image analysis. Manual segmentation is time-consuming and deals with intra and inter-observer variability. To develop a deep-learning approach for the fully automated segmentation of the inner ear in MRI, a 3D U-net was trained on 944 MRI scans with manually segmented inner ears as reference standard. The model was validated on an independent, multicentric dataset consisting of 177 MRI scans from three different centers. The model was also evaluated on a clinical validation set containing eight MRI scans with severe changes in the morphology of the labyrinth. The 3D U-net model showed precise Dice Similarity Coefficient scores (mean DSC-0.8790) with a high True Positive Rate (91.5%) and low False Discovery Rate and False Negative Rates (14.8% and 8.49% respectively) across images from three different centers. The model proved to perform well with a DSC of 0.8768 on the clinical validation dataset. The proposed auto-segmentation model is equivalent to human readers and is a reliable, consistent, and efficient method for inner ear segmentation, which can be used in a variety of clinical applications such as surgical planning and quantitative image analysis.https://doi.org/10.1038/s41598-021-82289-y |
| spellingShingle | Akshayaa Vaidyanathan Marly F. J. A. van der Lubbe Ralph T. H. Leijenaar Marc van Hoof Fadila Zerka Benjamin Miraglio Sergey Primakov Alida A. Postma Tjasse D. Bruintjes Monique A. L. Bilderbeek Hammer Sebastiaan Patrick F. M. Dammeijer Vincent van Rompaey Henry C. Woodruff Wim Vos Seán Walsh Raymond van de Berg Philippe Lambin Deep learning for the fully automated segmentation of the inner ear on MRI Scientific Reports |
| title | Deep learning for the fully automated segmentation of the inner ear on MRI |
| title_full | Deep learning for the fully automated segmentation of the inner ear on MRI |
| title_fullStr | Deep learning for the fully automated segmentation of the inner ear on MRI |
| title_full_unstemmed | Deep learning for the fully automated segmentation of the inner ear on MRI |
| title_short | Deep learning for the fully automated segmentation of the inner ear on MRI |
| title_sort | deep learning for the fully automated segmentation of the inner ear on mri |
| url | https://doi.org/10.1038/s41598-021-82289-y |
| work_keys_str_mv | AT akshayaavaidyanathan deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT marlyfjavanderlubbe deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT ralphthleijenaar deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT marcvanhoof deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT fadilazerka deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT benjaminmiraglio deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT sergeyprimakov deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT alidaapostma deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT tjassedbruintjes deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT moniquealbilderbeek deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT hammersebastiaan deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT patrickfmdammeijer deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT vincentvanrompaey deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT henrycwoodruff deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT wimvos deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT seanwalsh deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT raymondvandeberg deeplearningforthefullyautomatedsegmentationoftheinnerearonmri AT philippelambin deeplearningforthefullyautomatedsegmentationoftheinnerearonmri |