Whole Spine Segmentation Using Object Detection and Semantic Segmentation
Objective Virtual and augmented reality have enjoyed increased attention in spine surgery. Preoperative planning, pedicle screw placement, and surgical training are among the most studied use cases. Identifying osseous structures is a key aspect of navigating a 3-dimensional virtual reconstruction....
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Korean Spinal Neurosurgery Society
2024-03-01
|
| Series: | Neurospine |
| Subjects: | |
| Online Access: | http://www.e-neurospine.org/upload/pdf/ns-2347178-589.pdf |
| _version_ | 1797237784154472448 |
|---|---|
| author | Raffaele Da Mutten Olivier Zanier Sven Theiler Seung-Jun Ryu Luca Regli Carlo Serra Victor E. Staartjes |
| author_facet | Raffaele Da Mutten Olivier Zanier Sven Theiler Seung-Jun Ryu Luca Regli Carlo Serra Victor E. Staartjes |
| author_sort | Raffaele Da Mutten |
| collection | DOAJ |
| description | Objective Virtual and augmented reality have enjoyed increased attention in spine surgery. Preoperative planning, pedicle screw placement, and surgical training are among the most studied use cases. Identifying osseous structures is a key aspect of navigating a 3-dimensional virtual reconstruction. To automate the otherwise time-consuming process of labeling vertebrae on each slice individually, we propose a fully automated pipeline that automates segmentation on computed tomography (CT) and which can form the basis for further virtual or augmented reality application and radiomic analysis. Methods Based on a large public dataset of annotated vertebral CT scans, we first trained a YOLOv8m (You-Only-Look-Once algorithm, Version 8 and size medium) to detect each vertebra individually. On the then cropped images, a 2D-U-Net was developed and externally validated on 2 different public datasets. Results Two hundred fourteen CT scans (cervical, thoracic, or lumbar spine) were used for model training, and 40 scans were used for external validation. Vertebra recognition achieved a mAP50 (mean average precision with Jaccard threshold of 0.5) of over 0.84, and the segmentation algorithm attained a mean Dice score of 0.75 ± 0.14 at internal, 0.77 ± 0.12 and 0.82 ± 0.14 at external validation, respectively. Conclusion We propose a 2-stage approach consisting of single vertebra labeling by an object detection algorithm followed by semantic segmentation. In our externally validated pilot study, we demonstrate robust performance for our object detection network in identifying individual vertebrae, as well as for our segmentation model in precisely delineating the bony structures. |
| first_indexed | 2024-04-24T17:25:15Z |
| format | Article |
| id | doaj.art-c25f82f5b55748bab0c73020abef1dca |
| institution | Directory Open Access Journal |
| issn | 2586-6583 2586-6591 |
| language | English |
| last_indexed | 2024-04-24T17:25:15Z |
| publishDate | 2024-03-01 |
| publisher | Korean Spinal Neurosurgery Society |
| record_format | Article |
| series | Neurospine |
| spelling | doaj.art-c25f82f5b55748bab0c73020abef1dca2024-03-28T07:08:10ZengKorean Spinal Neurosurgery SocietyNeurospine2586-65832586-65912024-03-01211576710.14245/ns.2347178.5891539Whole Spine Segmentation Using Object Detection and Semantic SegmentationRaffaele Da Mutten0Olivier Zanier1Sven Theiler2Seung-Jun Ryu3Luca Regli4Carlo Serra5Victor E. Staartjes6 Machine Intelligence in Clinical Neuroscience (MICN) Laboratory, Department of Neurosurgery, Clinical Neuroscience Center, University Hospital Zürich, University of Zürich, Zürich, Switzerland Machine Intelligence in Clinical Neuroscience (MICN) Laboratory, Department of Neurosurgery, Clinical Neuroscience Center, University Hospital Zürich, University of Zürich, Zürich, Switzerland Machine Intelligence in Clinical Neuroscience (MICN) Laboratory, Department of Neurosurgery, Clinical Neuroscience Center, University Hospital Zürich, University of Zürich, Zürich, Switzerland Department of Neurosurgery, Daejeon Eulji University Hospital, Eulji University Medical School, Daejeon, Korea Machine Intelligence in Clinical Neuroscience (MICN) Laboratory, Department of Neurosurgery, Clinical Neuroscience Center, University Hospital Zürich, University of Zürich, Zürich, Switzerland Machine Intelligence in Clinical Neuroscience (MICN) Laboratory, Department of Neurosurgery, Clinical Neuroscience Center, University Hospital Zürich, University of Zürich, Zürich, Switzerland Machine Intelligence in Clinical Neuroscience (MICN) Laboratory, Department of Neurosurgery, Clinical Neuroscience Center, University Hospital Zürich, University of Zürich, Zürich, SwitzerlandObjective Virtual and augmented reality have enjoyed increased attention in spine surgery. Preoperative planning, pedicle screw placement, and surgical training are among the most studied use cases. Identifying osseous structures is a key aspect of navigating a 3-dimensional virtual reconstruction. To automate the otherwise time-consuming process of labeling vertebrae on each slice individually, we propose a fully automated pipeline that automates segmentation on computed tomography (CT) and which can form the basis for further virtual or augmented reality application and radiomic analysis. Methods Based on a large public dataset of annotated vertebral CT scans, we first trained a YOLOv8m (You-Only-Look-Once algorithm, Version 8 and size medium) to detect each vertebra individually. On the then cropped images, a 2D-U-Net was developed and externally validated on 2 different public datasets. Results Two hundred fourteen CT scans (cervical, thoracic, or lumbar spine) were used for model training, and 40 scans were used for external validation. Vertebra recognition achieved a mAP50 (mean average precision with Jaccard threshold of 0.5) of over 0.84, and the segmentation algorithm attained a mean Dice score of 0.75 ± 0.14 at internal, 0.77 ± 0.12 and 0.82 ± 0.14 at external validation, respectively. Conclusion We propose a 2-stage approach consisting of single vertebra labeling by an object detection algorithm followed by semantic segmentation. In our externally validated pilot study, we demonstrate robust performance for our object detection network in identifying individual vertebrae, as well as for our segmentation model in precisely delineating the bony structures.http://www.e-neurospine.org/upload/pdf/ns-2347178-589.pdfmachine learningdeep learningspineartificial intelligencealgorithms |
| spellingShingle | Raffaele Da Mutten Olivier Zanier Sven Theiler Seung-Jun Ryu Luca Regli Carlo Serra Victor E. Staartjes Whole Spine Segmentation Using Object Detection and Semantic Segmentation Neurospine machine learning deep learning spine artificial intelligence algorithms |
| title | Whole Spine Segmentation Using Object Detection and Semantic Segmentation |
| title_full | Whole Spine Segmentation Using Object Detection and Semantic Segmentation |
| title_fullStr | Whole Spine Segmentation Using Object Detection and Semantic Segmentation |
| title_full_unstemmed | Whole Spine Segmentation Using Object Detection and Semantic Segmentation |
| title_short | Whole Spine Segmentation Using Object Detection and Semantic Segmentation |
| title_sort | whole spine segmentation using object detection and semantic segmentation |
| topic | machine learning deep learning spine artificial intelligence algorithms |
| url | http://www.e-neurospine.org/upload/pdf/ns-2347178-589.pdf |
| work_keys_str_mv | AT raffaeledamutten wholespinesegmentationusingobjectdetectionandsemanticsegmentation AT olivierzanier wholespinesegmentationusingobjectdetectionandsemanticsegmentation AT sventheiler wholespinesegmentationusingobjectdetectionandsemanticsegmentation AT seungjunryu wholespinesegmentationusingobjectdetectionandsemanticsegmentation AT lucaregli wholespinesegmentationusingobjectdetectionandsemanticsegmentation AT carloserra wholespinesegmentationusingobjectdetectionandsemanticsegmentation AT victorestaartjes wholespinesegmentationusingobjectdetectionandsemanticsegmentation |