Denoising diffusion-based MRI to CT image translation enables automated spinal segmentation
Abstract Background Automated segmentation of spinal magnetic resonance imaging (MRI) plays a vital role both scientifically and clinically. However, accurately delineating posterior spine structures is challenging. Methods This retrospective study, approved by the ethical committee, involved transl...
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2023-11-01
|
| Series: | European Radiology Experimental |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s41747-023-00385-2 |
| _version_ | 1797578223001796608 |
|---|---|
| author | Robert Graf Joachim Schmitt Sarah Schlaeger Hendrik Kristian Möller Vasiliki Sideri-Lampretsa Anjany Sekuboyina Sandro Manuel Krieg Benedikt Wiestler Bjoern Menze Daniel Rueckert Jan Stefan Kirschke |
| author_facet | Robert Graf Joachim Schmitt Sarah Schlaeger Hendrik Kristian Möller Vasiliki Sideri-Lampretsa Anjany Sekuboyina Sandro Manuel Krieg Benedikt Wiestler Bjoern Menze Daniel Rueckert Jan Stefan Kirschke |
| author_sort | Robert Graf |
| collection | DOAJ |
| description | Abstract Background Automated segmentation of spinal magnetic resonance imaging (MRI) plays a vital role both scientifically and clinically. However, accurately delineating posterior spine structures is challenging. Methods This retrospective study, approved by the ethical committee, involved translating T1-weighted and T2-weighted images into computed tomography (CT) images in a total of 263 pairs of CT/MR series. Landmark-based registration was performed to align image pairs. We compared two-dimensional (2D) paired — Pix2Pix, denoising diffusion implicit models (DDIM) image mode, DDIM noise mode — and unpaired (SynDiff, contrastive unpaired translation) image-to-image translation using “peak signal-to-noise ratio” as quality measure. A publicly available segmentation network segmented the synthesized CT datasets, and Dice similarity coefficients (DSC) were evaluated on in-house test sets and the “MRSpineSeg Challenge” volumes. The 2D findings were extended to three-dimensional (3D) Pix2Pix and DDIM. Results 2D paired methods and SynDiff exhibited similar translation performance and DCS on paired data. DDIM image mode achieved the highest image quality. SynDiff, Pix2Pix, and DDIM image mode demonstrated similar DSC (0.77). For craniocaudal axis rotations, at least two landmarks per vertebra were required for registration. The 3D translation outperformed the 2D approach, resulting in improved DSC (0.80) and anatomically accurate segmentations with higher spatial resolution than that of the original MRI series. Conclusions Two landmarks per vertebra registration enabled paired image-to-image translation from MRI to CT and outperformed all unpaired approaches. The 3D techniques provided anatomically correct segmentations, avoiding underprediction of small structures like the spinous process. Relevance statement This study addresses the unresolved issue of translating spinal MRI to CT, making CT-based tools usable for MRI data. It generates whole spine segmentation, previously unavailable in MRI, a prerequisite for biomechanical modeling and feature extraction for clinical applications. Key points • Unpaired image translation lacks in converting spine MRI to CT effectively. • Paired translation needs registration with two landmarks per vertebra at least. • Paired image-to-image enables segmentation transfer to other domains. • 3D translation enables super resolution from MRI to CT. • 3D translation prevents underprediction of small structures. Graphical Abstract |
| first_indexed | 2024-03-10T22:19:55Z |
| format | Article |
| id | doaj.art-05c5f8ba22ef47cda36ebc1fe94a4f0c |
| institution | Directory Open Access Journal |
| issn | 2509-9280 |
| language | English |
| last_indexed | 2024-03-10T22:19:55Z |
| publishDate | 2023-11-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | European Radiology Experimental |
| spelling | doaj.art-05c5f8ba22ef47cda36ebc1fe94a4f0c2023-11-19T12:19:00ZengSpringerOpenEuropean Radiology Experimental2509-92802023-11-017111410.1186/s41747-023-00385-2Denoising diffusion-based MRI to CT image translation enables automated spinal segmentationRobert Graf0Joachim Schmitt1Sarah Schlaeger2Hendrik Kristian Möller3Vasiliki Sideri-Lampretsa4Anjany Sekuboyina5Sandro Manuel Krieg6Benedikt Wiestler7Bjoern Menze8Daniel Rueckert9Jan Stefan Kirschke10Department of Diagnostic and Interventional Neuroradiology, School of Medicine, Technical University of MunichDepartment of Diagnostic and Interventional Neuroradiology, School of Medicine, Technical University of MunichDepartment of Diagnostic and Interventional Neuroradiology, School of Medicine, Technical University of MunichDepartment of Diagnostic and Interventional Neuroradiology, School of Medicine, Technical University of MunichInstitut Für KI Und Informatik in Der Medizin, Klinikum Rechts Der Isar, Technical University of MunichDepartment of Diagnostic and Interventional Neuroradiology, School of Medicine, Technical University of MunichDepartment of Neurosurgery, Klinikum Rechts Der Isar, School of Medicine, Technical University of MunichDepartment of Diagnostic and Interventional Neuroradiology, School of Medicine, Technical University of MunichDepartment of Quantitative Biomedicine, University of ZurichInstitut Für KI Und Informatik in Der Medizin, Klinikum Rechts Der Isar, Technical University of MunichDepartment of Diagnostic and Interventional Neuroradiology, School of Medicine, Technical University of MunichAbstract Background Automated segmentation of spinal magnetic resonance imaging (MRI) plays a vital role both scientifically and clinically. However, accurately delineating posterior spine structures is challenging. Methods This retrospective study, approved by the ethical committee, involved translating T1-weighted and T2-weighted images into computed tomography (CT) images in a total of 263 pairs of CT/MR series. Landmark-based registration was performed to align image pairs. We compared two-dimensional (2D) paired — Pix2Pix, denoising diffusion implicit models (DDIM) image mode, DDIM noise mode — and unpaired (SynDiff, contrastive unpaired translation) image-to-image translation using “peak signal-to-noise ratio” as quality measure. A publicly available segmentation network segmented the synthesized CT datasets, and Dice similarity coefficients (DSC) were evaluated on in-house test sets and the “MRSpineSeg Challenge” volumes. The 2D findings were extended to three-dimensional (3D) Pix2Pix and DDIM. Results 2D paired methods and SynDiff exhibited similar translation performance and DCS on paired data. DDIM image mode achieved the highest image quality. SynDiff, Pix2Pix, and DDIM image mode demonstrated similar DSC (0.77). For craniocaudal axis rotations, at least two landmarks per vertebra were required for registration. The 3D translation outperformed the 2D approach, resulting in improved DSC (0.80) and anatomically accurate segmentations with higher spatial resolution than that of the original MRI series. Conclusions Two landmarks per vertebra registration enabled paired image-to-image translation from MRI to CT and outperformed all unpaired approaches. The 3D techniques provided anatomically correct segmentations, avoiding underprediction of small structures like the spinous process. Relevance statement This study addresses the unresolved issue of translating spinal MRI to CT, making CT-based tools usable for MRI data. It generates whole spine segmentation, previously unavailable in MRI, a prerequisite for biomechanical modeling and feature extraction for clinical applications. Key points • Unpaired image translation lacks in converting spine MRI to CT effectively. • Paired translation needs registration with two landmarks per vertebra at least. • Paired image-to-image enables segmentation transfer to other domains. • 3D translation enables super resolution from MRI to CT. • 3D translation prevents underprediction of small structures. Graphical Abstracthttps://doi.org/10.1186/s41747-023-00385-2Deep learningImage processing (computer assisted)Magnetic resonance imagingSpineVertebral body |
| spellingShingle | Robert Graf Joachim Schmitt Sarah Schlaeger Hendrik Kristian Möller Vasiliki Sideri-Lampretsa Anjany Sekuboyina Sandro Manuel Krieg Benedikt Wiestler Bjoern Menze Daniel Rueckert Jan Stefan Kirschke Denoising diffusion-based MRI to CT image translation enables automated spinal segmentation European Radiology Experimental Deep learning Image processing (computer assisted) Magnetic resonance imaging Spine Vertebral body |
| title | Denoising diffusion-based MRI to CT image translation enables automated spinal segmentation |
| title_full | Denoising diffusion-based MRI to CT image translation enables automated spinal segmentation |
| title_fullStr | Denoising diffusion-based MRI to CT image translation enables automated spinal segmentation |
| title_full_unstemmed | Denoising diffusion-based MRI to CT image translation enables automated spinal segmentation |
| title_short | Denoising diffusion-based MRI to CT image translation enables automated spinal segmentation |
| title_sort | denoising diffusion based mri to ct image translation enables automated spinal segmentation |
| topic | Deep learning Image processing (computer assisted) Magnetic resonance imaging Spine Vertebral body |
| url | https://doi.org/10.1186/s41747-023-00385-2 |
| work_keys_str_mv | AT robertgraf denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation AT joachimschmitt denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation AT sarahschlaeger denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation AT hendrikkristianmoller denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation AT vasilikisiderilampretsa denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation AT anjanysekuboyina denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation AT sandromanuelkrieg denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation AT benediktwiestler denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation AT bjoernmenze denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation AT danielrueckert denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation AT janstefankirschke denoisingdiffusionbasedmritoctimagetranslationenablesautomatedspinalsegmentation |