Pseudo-Label Assisted nnU-Net enables automatic segmentation of 7T MRI from a single acquisition
IntroductionAutomatic whole brain and lesion segmentation at 7T presents challenges, primarily from bias fields, susceptibility artifacts including distortions, and registration errors. Here, we sought to use deep learning algorithms (D/L) to do both skull stripping and whole brain segmentation on m...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2023-12-01
|
| Series: | Frontiers in Neuroimaging |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fnimg.2023.1252261/full |
| _version_ | 1827159660081184768 |
|---|---|
| author | Corinne Donnay Henry Dieckhaus Charidimos Tsagkas María Inés Gaitán Erin S. Beck Erin S. Beck Andrew Mullins Andrew Mullins Daniel S. Reich Govind Nair |
| author_facet | Corinne Donnay Henry Dieckhaus Charidimos Tsagkas María Inés Gaitán Erin S. Beck Erin S. Beck Andrew Mullins Andrew Mullins Daniel S. Reich Govind Nair |
| author_sort | Corinne Donnay |
| collection | DOAJ |
| description | IntroductionAutomatic whole brain and lesion segmentation at 7T presents challenges, primarily from bias fields, susceptibility artifacts including distortions, and registration errors. Here, we sought to use deep learning algorithms (D/L) to do both skull stripping and whole brain segmentation on multiple imaging contrasts generated in a single Magnetization Prepared 2 Rapid Acquisition Gradient Echoes (MP2RAGE) acquisition on participants clinically diagnosed with multiple sclerosis (MS), bypassing registration errors.MethodsBrain scans Segmentation from 3T and 7T scanners were analyzed with software packages such as FreeSurfer, Classification using Derivative-based Features (C-DEF), nnU-net, and a novel 3T-to-7T transfer learning method, Pseudo-Label Assisted nnU-Net (PLAn). 3T and 7T MRIs acquired within 9 months from 25 study participants with MS (Cohort 1) were used for training and optimizing. Eight MS patients (Cohort 2) scanned only at 7T, but with expert annotated lesion segmentation, was used to further validate the algorithm on a completely unseen dataset. Segmentation results were rated visually by experts in a blinded fashion and quantitatively using Dice Similarity Coefficient (DSC).ResultsOf the methods explored here, nnU-Net and PLAn produced the best tissue segmentation at 7T for all tissue classes. In both quantitative and qualitative analysis, PLAn significantly outperformed nnU-Net (and other methods) in lesion detection in both cohorts. PLAn's lesion DSC improved by 16% compared to nnU-Net.DiscussionLimited availability of labeled data makes transfer learning an attractive option, and pre-training a nnUNet model using readily obtained 3T pseudo-labels was shown to boost lesion detection capabilities at 7T. |
| first_indexed | 2024-03-09T10:42:23Z |
| format | Article |
| id | doaj.art-3b5b15fe013f4545a3f57498e8b49799 |
| institution | Directory Open Access Journal |
| issn | 2813-1193 |
| language | English |
| last_indexed | 2025-03-21T00:00:39Z |
| publishDate | 2023-12-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Neuroimaging |
| spelling | doaj.art-3b5b15fe013f4545a3f57498e8b497992024-08-03T11:37:41ZengFrontiers Media S.A.Frontiers in Neuroimaging2813-11932023-12-01210.3389/fnimg.2023.12522611252261Pseudo-Label Assisted nnU-Net enables automatic segmentation of 7T MRI from a single acquisitionCorinne Donnay0Henry Dieckhaus1Charidimos Tsagkas2María Inés Gaitán3Erin S. Beck4Erin S. Beck5Andrew Mullins6Andrew Mullins7Daniel S. Reich8Govind Nair9Translational Neuroradiology Section, National Institute of Neurological Disorders and Stroke (NINDS), National Institutes of Health, Bethesda, MD, United StatesqMRI Core, NINDS, National Institutes of Health, Bethesda, MD, United StatesTranslational Neuroradiology Section, National Institute of Neurological Disorders and Stroke (NINDS), National Institutes of Health, Bethesda, MD, United StatesTranslational Neuroradiology Section, National Institute of Neurological Disorders and Stroke (NINDS), National Institutes of Health, Bethesda, MD, United StatesTranslational Neuroradiology Section, National Institute of Neurological Disorders and Stroke (NINDS), National Institutes of Health, Bethesda, MD, United StatesDepartment of Neurology, Icahn School of Medicine at Mount Sinai, New York, NY, United StatesTranslational Neuroradiology Section, National Institute of Neurological Disorders and Stroke (NINDS), National Institutes of Health, Bethesda, MD, United StatesqMRI Core, NINDS, National Institutes of Health, Bethesda, MD, United StatesTranslational Neuroradiology Section, National Institute of Neurological Disorders and Stroke (NINDS), National Institutes of Health, Bethesda, MD, United StatesqMRI Core, NINDS, National Institutes of Health, Bethesda, MD, United StatesIntroductionAutomatic whole brain and lesion segmentation at 7T presents challenges, primarily from bias fields, susceptibility artifacts including distortions, and registration errors. Here, we sought to use deep learning algorithms (D/L) to do both skull stripping and whole brain segmentation on multiple imaging contrasts generated in a single Magnetization Prepared 2 Rapid Acquisition Gradient Echoes (MP2RAGE) acquisition on participants clinically diagnosed with multiple sclerosis (MS), bypassing registration errors.MethodsBrain scans Segmentation from 3T and 7T scanners were analyzed with software packages such as FreeSurfer, Classification using Derivative-based Features (C-DEF), nnU-net, and a novel 3T-to-7T transfer learning method, Pseudo-Label Assisted nnU-Net (PLAn). 3T and 7T MRIs acquired within 9 months from 25 study participants with MS (Cohort 1) were used for training and optimizing. Eight MS patients (Cohort 2) scanned only at 7T, but with expert annotated lesion segmentation, was used to further validate the algorithm on a completely unseen dataset. Segmentation results were rated visually by experts in a blinded fashion and quantitatively using Dice Similarity Coefficient (DSC).ResultsOf the methods explored here, nnU-Net and PLAn produced the best tissue segmentation at 7T for all tissue classes. In both quantitative and qualitative analysis, PLAn significantly outperformed nnU-Net (and other methods) in lesion detection in both cohorts. PLAn's lesion DSC improved by 16% compared to nnU-Net.DiscussionLimited availability of labeled data makes transfer learning an attractive option, and pre-training a nnUNet model using readily obtained 3T pseudo-labels was shown to boost lesion detection capabilities at 7T.https://www.frontiersin.org/articles/10.3389/fnimg.2023.1252261/fullbrain and lesion segmentation7T MRIdeep learningtransfer learningmultiple sclerosis |
| spellingShingle | Corinne Donnay Henry Dieckhaus Charidimos Tsagkas María Inés Gaitán Erin S. Beck Erin S. Beck Andrew Mullins Andrew Mullins Daniel S. Reich Govind Nair Pseudo-Label Assisted nnU-Net enables automatic segmentation of 7T MRI from a single acquisition Frontiers in Neuroimaging brain and lesion segmentation 7T MRI deep learning transfer learning multiple sclerosis |
| title | Pseudo-Label Assisted nnU-Net enables automatic segmentation of 7T MRI from a single acquisition |
| title_full | Pseudo-Label Assisted nnU-Net enables automatic segmentation of 7T MRI from a single acquisition |
| title_fullStr | Pseudo-Label Assisted nnU-Net enables automatic segmentation of 7T MRI from a single acquisition |
| title_full_unstemmed | Pseudo-Label Assisted nnU-Net enables automatic segmentation of 7T MRI from a single acquisition |
| title_short | Pseudo-Label Assisted nnU-Net enables automatic segmentation of 7T MRI from a single acquisition |
| title_sort | pseudo label assisted nnu net enables automatic segmentation of 7t mri from a single acquisition |
| topic | brain and lesion segmentation 7T MRI deep learning transfer learning multiple sclerosis |
| url | https://www.frontiersin.org/articles/10.3389/fnimg.2023.1252261/full |
| work_keys_str_mv | AT corinnedonnay pseudolabelassistednnunetenablesautomaticsegmentationof7tmrifromasingleacquisition AT henrydieckhaus pseudolabelassistednnunetenablesautomaticsegmentationof7tmrifromasingleacquisition AT charidimostsagkas pseudolabelassistednnunetenablesautomaticsegmentationof7tmrifromasingleacquisition AT mariainesgaitan pseudolabelassistednnunetenablesautomaticsegmentationof7tmrifromasingleacquisition AT erinsbeck pseudolabelassistednnunetenablesautomaticsegmentationof7tmrifromasingleacquisition AT erinsbeck pseudolabelassistednnunetenablesautomaticsegmentationof7tmrifromasingleacquisition AT andrewmullins pseudolabelassistednnunetenablesautomaticsegmentationof7tmrifromasingleacquisition AT andrewmullins pseudolabelassistednnunetenablesautomaticsegmentationof7tmrifromasingleacquisition AT danielsreich pseudolabelassistednnunetenablesautomaticsegmentationof7tmrifromasingleacquisition AT govindnair pseudolabelassistednnunetenablesautomaticsegmentationof7tmrifromasingleacquisition |