DL-based segmentation of endoscopic scenes for mitral valve repair
Minimally invasive surgery is increasingly utilized for mitral valve repair and replacement. The intervention is performed with an endoscopic field of view on the arrested heart. Extracting the necessary information from the live endoscopic video stream is challenging due to the moving camera positi...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
De Gruyter
2020-09-01
|
| Series: | Current Directions in Biomedical Engineering |
| Subjects: | |
| Online Access: | https://doi.org/10.1515/cdbme-2020-0017 |
| _version_ | 1830192545811922944 |
|---|---|
| author | Ivantsits Matthias Tautz Lennart Sündermann Simon Wamala Isaac Kempfert Jörg Kuehne Titus Falk Volkmar Hennemuth Anja |
| author_facet | Ivantsits Matthias Tautz Lennart Sündermann Simon Wamala Isaac Kempfert Jörg Kuehne Titus Falk Volkmar Hennemuth Anja |
| author_sort | Ivantsits Matthias |
| collection | DOAJ |
| description | Minimally invasive surgery is increasingly utilized for mitral valve repair and replacement. The intervention is performed with an endoscopic field of view on the arrested heart. Extracting the necessary information from the live endoscopic video stream is challenging due to the moving camera position, the high variability of defects, and occlusion of structures by instruments. During such minimally invasive interventions there is no time to segment regions of interest manually. We propose a real-time-capable deep-learning-based approach to detect and segment the relevant anatomical structures and instruments. For the universal deployment of the proposed solution, we evaluate them on pixel accuracy as well as distance measurements of the detected contours. The U-Net, Google’s DeepLab v3, and the Obelisk-Net models are cross-validated, with DeepLab showing superior results in pixel accuracy and distance measurements. |
| first_indexed | 2024-12-17T23:58:24Z |
| format | Article |
| id | doaj.art-015a861c7a8a438b8f48af4ac4d1b893 |
| institution | Directory Open Access Journal |
| issn | 2364-5504 |
| language | English |
| last_indexed | 2024-12-17T23:58:24Z |
| publishDate | 2020-09-01 |
| publisher | De Gruyter |
| record_format | Article |
| series | Current Directions in Biomedical Engineering |
| spelling | doaj.art-015a861c7a8a438b8f48af4ac4d1b8932022-12-21T21:28:01ZengDe GruyterCurrent Directions in Biomedical Engineering2364-55042020-09-01616082110.1515/cdbme-2020-0017cdbme-2020-0017DL-based segmentation of endoscopic scenes for mitral valve repairIvantsits Matthias0Tautz Lennart1Sündermann Simon2Wamala Isaac3Kempfert Jörg4Kuehne Titus5Falk Volkmar6Hennemuth Anja7Charité – Universitätsmedizin Berlin, Berlin, GermanyCharité – Universitätsmedizin Berlin, Berlin, GermanyCharité – Universitätsmedizin Berlin, Berlin, GermanyDZHK (German Centre for Cardiovascular Research), Berlin, GermanyCharité – Universitätsmedizin Berlin, Berlin, GermanyCharité – Universitätsmedizin Berlin, Berlin, GermanyCharité – Universitätsmedizin Berlin, Berlin, GermanyCharité – Universitätsmedizin Berlin, Berlin, GermanyMinimally invasive surgery is increasingly utilized for mitral valve repair and replacement. The intervention is performed with an endoscopic field of view on the arrested heart. Extracting the necessary information from the live endoscopic video stream is challenging due to the moving camera position, the high variability of defects, and occlusion of structures by instruments. During such minimally invasive interventions there is no time to segment regions of interest manually. We propose a real-time-capable deep-learning-based approach to detect and segment the relevant anatomical structures and instruments. For the universal deployment of the proposed solution, we evaluate them on pixel accuracy as well as distance measurements of the detected contours. The U-Net, Google’s DeepLab v3, and the Obelisk-Net models are cross-validated, with DeepLab showing superior results in pixel accuracy and distance measurements.https://doi.org/10.1515/cdbme-2020-0017deep learningdetectionendoscopicmachine learningmitral valvemitral valve leafletsegmentationsurgery |
| spellingShingle | Ivantsits Matthias Tautz Lennart Sündermann Simon Wamala Isaac Kempfert Jörg Kuehne Titus Falk Volkmar Hennemuth Anja DL-based segmentation of endoscopic scenes for mitral valve repair Current Directions in Biomedical Engineering deep learning detection endoscopic machine learning mitral valve mitral valve leaflet segmentation surgery |
| title | DL-based segmentation of endoscopic scenes for mitral valve repair |
| title_full | DL-based segmentation of endoscopic scenes for mitral valve repair |
| title_fullStr | DL-based segmentation of endoscopic scenes for mitral valve repair |
| title_full_unstemmed | DL-based segmentation of endoscopic scenes for mitral valve repair |
| title_short | DL-based segmentation of endoscopic scenes for mitral valve repair |
| title_sort | dl based segmentation of endoscopic scenes for mitral valve repair |
| topic | deep learning detection endoscopic machine learning mitral valve mitral valve leaflet segmentation surgery |
| url | https://doi.org/10.1515/cdbme-2020-0017 |
| work_keys_str_mv | AT ivantsitsmatthias dlbasedsegmentationofendoscopicscenesformitralvalverepair AT tautzlennart dlbasedsegmentationofendoscopicscenesformitralvalverepair AT sundermannsimon dlbasedsegmentationofendoscopicscenesformitralvalverepair AT wamalaisaac dlbasedsegmentationofendoscopicscenesformitralvalverepair AT kempfertjorg dlbasedsegmentationofendoscopicscenesformitralvalverepair AT kuehnetitus dlbasedsegmentationofendoscopicscenesformitralvalverepair AT falkvolkmar dlbasedsegmentationofendoscopicscenesformitralvalverepair AT hennemuthanja dlbasedsegmentationofendoscopicscenesformitralvalverepair |