Automated peripheral nerve segmentation for MR-neurography.
Artificial intelligence
Magnetic resonance imaging
Neural networks (computer)
Peripheral nervous system
Sciatic nerve
Journal
European radiology experimental
ISSN: 2509-9280
Titre abrégé: Eur Radiol Exp
Pays: England
ID NLM: 101721752
Informations de publication
Date de publication:
26 Aug 2024
26 Aug 2024
Historique:
received:
18
12
2023
accepted:
01
08
2024
medline:
26
8
2024
pubmed:
26
8
2024
entrez:
26
8
2024
Statut:
epublish
Résumé
Magnetic resonance neurography (MRN) is increasingly used as a diagnostic tool for peripheral neuropathies. Quantitative measures enhance MRN interpretation but require nerve segmentation which is time-consuming and error-prone and has not become clinical routine. In this study, we applied neural networks for the automated segmentation of peripheral nerves. A neural segmentation network was trained to segment the sciatic nerve and its proximal branches on the MRN scans of the right and left upper leg of 35 healthy individuals, resulting in 70 training examples, via 5-fold cross-validation (CV). The model performance was evaluated on an independent test set of one-sided MRN scans of 60 healthy individuals. Mean Dice similarity coefficient (DSC) in CV was 0.892 (95% confidence interval [CI]: 0.888-0.897) with a mean Jaccard index (JI) of 0.806 (95% CI: 0.799-0.814) and mean Hausdorff distance (HD) of 2.146 (95% CI: 2.184-2.208). For the independent test set, DSC and JI were lower while HD was higher, with a mean DSC of 0.789 (95% CI: 0.760-0.815), mean JI of 0.672 (95% CI: 0.642-0.699), and mean HD of 2.118 (95% CI: 2.047-2.190). The deep learning-based segmentation model showed a good performance for the task of nerve segmentation. Future work will focus on extending training data and including individuals with peripheral neuropathies in training to enable advanced peripheral nerve disease characterization. The results will serve as a baseline to build upon while developing an automated quantitative MRN feature analysis framework for application in routine reading of MRN examinations. Quantitative measures enhance MRN interpretation, requiring complex and challenging nerve segmentation. We present a deep learning-based segmentation model with good performance. Our results may serve as a baseline for clinical automated quantitative MRN segmentation.
Sections du résumé
BACKGROUND
BACKGROUND
Magnetic resonance neurography (MRN) is increasingly used as a diagnostic tool for peripheral neuropathies. Quantitative measures enhance MRN interpretation but require nerve segmentation which is time-consuming and error-prone and has not become clinical routine. In this study, we applied neural networks for the automated segmentation of peripheral nerves.
METHODS
METHODS
A neural segmentation network was trained to segment the sciatic nerve and its proximal branches on the MRN scans of the right and left upper leg of 35 healthy individuals, resulting in 70 training examples, via 5-fold cross-validation (CV). The model performance was evaluated on an independent test set of one-sided MRN scans of 60 healthy individuals.
RESULTS
RESULTS
Mean Dice similarity coefficient (DSC) in CV was 0.892 (95% confidence interval [CI]: 0.888-0.897) with a mean Jaccard index (JI) of 0.806 (95% CI: 0.799-0.814) and mean Hausdorff distance (HD) of 2.146 (95% CI: 2.184-2.208). For the independent test set, DSC and JI were lower while HD was higher, with a mean DSC of 0.789 (95% CI: 0.760-0.815), mean JI of 0.672 (95% CI: 0.642-0.699), and mean HD of 2.118 (95% CI: 2.047-2.190).
CONCLUSION
CONCLUSIONS
The deep learning-based segmentation model showed a good performance for the task of nerve segmentation. Future work will focus on extending training data and including individuals with peripheral neuropathies in training to enable advanced peripheral nerve disease characterization.
RELEVANCE STATEMENT
CONCLUSIONS
The results will serve as a baseline to build upon while developing an automated quantitative MRN feature analysis framework for application in routine reading of MRN examinations.
KEY POINTS
CONCLUSIONS
Quantitative measures enhance MRN interpretation, requiring complex and challenging nerve segmentation. We present a deep learning-based segmentation model with good performance. Our results may serve as a baseline for clinical automated quantitative MRN segmentation.
Identifiants
pubmed: 39186183
doi: 10.1186/s41747-024-00503-8
pii: 10.1186/s41747-024-00503-8
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
97Informations de copyright
© 2024. The Author(s).
Références
Balsiger F, Steindel C, Arn M et al (2018) Segmentation of peripheral nerves from magnetic resonance neurography: a fully-automatic, deep learning-based approach. Front Neurol 9:777. https://doi.org/10.3389/FNEUR.2018.00777
doi: 10.3389/FNEUR.2018.00777
pubmed: 30283397
pmcid: 6156270
Bendszus M, Wessig C, Solymosi L et al (2004) MRI of peripheral nerve degeneration and regeneration: Correlation with electrophysiology and histology. Exp Neurol 188:171–177. https://doi.org/10.1016/j.expneurol.2004.03.025
doi: 10.1016/j.expneurol.2004.03.025
pubmed: 15191813
Chen Y, Mark Haacke E, Li J (2019) Peripheral nerve magnetic resonance imaging. F1000Res. https://doi.org/10.12688/F1000RESEARCH.19695.1
Filler AG, Kliot M, Winn HR et al (1993) Magnetic resonance neurography. Lancet 341:659–661. https://doi.org/10.1016/0140-6736(93)90422-D
doi: 10.1016/0140-6736(93)90422-D
pubmed: 8095572
Jende JME, Groener JB, Oikonomou D et al (2018) Diabetic neuropathy differs between type 1 and type 2 diabetes: insights from magnetic resonance neurography. Ann Neurol 83:588–598. https://doi.org/10.1002/ANA.25182
doi: 10.1002/ANA.25182
pubmed: 29443416
Kollmer J, Bäumer P, Milford D et al (2012) T2-signal of ulnar nerve branches at the wrist in Guyon’s canal syndrome. PLoS One 7:e47295. https://doi.org/10.1371/JOURNAL.PONE.0047295
doi: 10.1371/JOURNAL.PONE.0047295
pubmed: 23071777
pmcid: 3468548
Kronlage M, Schwehr V, Schwarz D et al (2019) Magnetic resonance neurography: normal values and demographic determinants of nerve caliber and T2 relaxometry in 60 healthy individuals. Clin Neuroradiol 29:19–26. https://doi.org/10.1007/S00062-017-0633-5
doi: 10.1007/S00062-017-0633-5
pubmed: 29030674
Pham M, Oikonomou D, Hornung B et al (2015) Magnetic resonance neurography detects diabetic neuropathy early and with proximal predominance. Ann Neurol 78:939. https://doi.org/10.1002/ANA.24524
doi: 10.1002/ANA.24524
pubmed: 26381658
pmcid: 5132066
Schwarz D, Weiler M, Pham M et al (2015) Diagnostic signs of motor neuropathy in MR neurography: nerve lesions and muscle denervation. Eur Radiol 25:1497–1503. https://doi.org/10.1007/S00330-014-3498-X
doi: 10.1007/S00330-014-3498-X
pubmed: 25433411
Pham M, Wessig C, Brinkhoff J et al (2011) MR neurography of sciatic nerve injection injury. J Neurol 258:1120–1125. https://doi.org/10.1007/S00415-010-5895-7
doi: 10.1007/S00415-010-5895-7
pubmed: 21225276
Pham M, Oikonomou D, Bäumer P et al (2011) Proximal neuropathic lesions in distal symmetric diabetic polyneuropathy: findings of high-resolution magnetic resonance neurography. Diabetes Care 34:721. https://doi.org/10.2337/DC10-1491
doi: 10.2337/DC10-1491
pubmed: 21266652
pmcid: 3041214
Kronlage M, Schwehr V, Schwarz D et al (2018) Peripheral nerve diffusion tensor imaging (DTI): normal values and demographic determinants in a cohort of 60 healthy individuals. Eur Radiol 28:1801–1808. https://doi.org/10.1007/S00330-017-5134-Z
doi: 10.1007/S00330-017-5134-Z
pubmed: 29230526
Lichtenstein T, Sprenger A, Weiss K et al (2018) MRI biomarkers of proximal nerve injury in CIDP. Ann Clin Transl Neurol 5:19–28. https://doi.org/10.1002/ACN3.502
doi: 10.1002/ACN3.502
pubmed: 29376089
Pitarokoili K, Kronlage M, Bäumer P et al (2018) High-resolution nerve ultrasound and magnetic resonance neurography as complementary neuroimaging tools for chronic inflammatory demyelinating polyneuropathy. Ther Adv Neurol Disord 11:1756286418759974. https://doi.org/10.1177/1756286418759974
doi: 10.1177/1756286418759974
pubmed: 29552093
pmcid: 5846906
Simon NG, Lagopoulos J, Gallagher T et al (2016) Peripheral nerve diffusion tensor imaging is reliable and reproducible. J Magn Reson Imaging 43:962–969. https://doi.org/10.1002/jmri.25056
doi: 10.1002/jmri.25056
pubmed: 26397723
Felisaz PF, Balducci F, Gitto S et al (2016) Nerve fascicles and epineurium volume segmentation of peripheral nerve using magnetic resonance micro-neurography. Acad Radiol 23(8):1000–1007. https://doi.org/10.1016/j.acra.2016.03.013
doi: 10.1016/j.acra.2016.03.013
pubmed: 27209266
Felisaz PF, Maugeri G, Busi V et al (2017) MR micro-neurography and a segmentation protocol applied to diabetic neuropathy. Radiology Res Pract 2017:2761818. https://doi.org/10.1155/2017/2761818
doi: 10.1155/2017/2761818
Rossi F, Bignotti B, Bianchi L et al (2020) Radiomics of peripheral nerves MRI in mild carpal and cubital tunnel syndrome. Radiol Med 125:197–203. https://doi.org/10.1007/S11547-019-01110-Z
doi: 10.1007/S11547-019-01110-Z
pubmed: 31773457
Ronneberger O, Fischer P, Brox T(2015) U-Net: convolutional networks for biomedical image segmentation. MICCAI. Lecture Notes Comp Science 9351:234–241. https://doi.org/10.1007/978-3-319-24574-4_28
doi: 10.1007/978-3-319-24574-4_28
Isensee F, Jaeger PF, Kohl SAA et al (2020) nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat Methods 18:203–211. https://doi.org/10.1038/s41592-020-01008-z
doi: 10.1038/s41592-020-01008-z
pubmed: 33288961
Fan DP, Zhou T, Ji GP et al (2020) Inf-Net: automatic COVID-19 lung infection segmentation from CT images. IEEE Trans Med Imaging 39:2626–2637. https://doi.org/10.1109/TMI.2020.2996645
doi: 10.1109/TMI.2020.2996645
pubmed: 32730213
Futrega M, Milesi A, Marcinkiewicz M et al (2022) Optimized U-Net for brain tumor segmentation. In: Crimi A, Bakas S (eds) Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries. BrainLes 2021. Lecture Notes in Computer Science, vol 12963. https://doi.org/10.1007/978-3-031-09002-8_2
Liu X, Song L, Liu S et al (2021) A review of deep-learning-based medical image segmentation methods. Sustainability 13:1224. https://doi.org/10.3390/SU13031224
doi: 10.3390/SU13031224
Siddique N, Paheding S, Elkin CP et al (2021) U-net and its variants for medical image segmentation: a review of theory and applications. IEEE Access 9:82031–82057. https://doi.org/10.1109/ACCESS.2021.3086020
doi: 10.1109/ACCESS.2021.3086020
van der Laak J, Litjens G, Ciompi F (2021) Deep learning in histopathology: the path to the clinic. Nat Med 27:775–784. https://doi.org/10.1038/s41591-021-01343-4
doi: 10.1038/s41591-021-01343-4
pubmed: 33990804
Walsh J, Othmani A, Jain M et al (2022) Using U-Net network for efficient brain tumor segmentation in MRI images. Healthc Anal 2:100098. https://doi.org/10.1016/J.HEALTH.2022.100098
doi: 10.1016/J.HEALTH.2022.100098
Zhang K, Liu X, Shen J et al (2020) Clinically applicable AI system for accurate diagnosis, quantitative measurements, and prognosis of COVID-19 pneumonia using computed tomography. Cell 181:1423–1433.e11. https://doi.org/10.1016/J.CELL.2020.04.045
doi: 10.1016/J.CELL.2020.04.045
pubmed: 32416069
pmcid: 7196900
Dice, Lee R (1945) Measures of the amount of ecologic association between species. Ecology 26:297–302. https://doi.org/10.2307/1932409
doi: 10.2307/1932409
Zou KH, Warfield SK, Bharatha A et al (2004) Statistical validation of image segmentation quality based on a spatial overlap index. Acad Radiol 11:178. https://doi.org/10.1016/S1076-6332(03)00671-8
doi: 10.1016/S1076-6332(03)00671-8
pubmed: 14974593
pmcid: 1415224
Hancock JM (2004) Jaccard distance (Jaccard index, Jaccard similarity coefficient). Dictionary of bioinformatics and computational biology [in book]. https://doi.org/10.1002/9780471650126.DOB0956
Huttenlocher DP, Klanderman GA, Rucklidge WJ (1993) Comparing images using the Hausdorff distance. IEEE Trans Pattern Anal Mach Intell 15:850–863. https://doi.org/10.1109/34.232073
doi: 10.1109/34.232073
Nesterov Y (1983) A method for solving the convex programming problem with convergence rate O(1/k
Taha AA, Hanbury A (2015) Metrics for evaluating 3D medical image segmentation: analysis, selection, and tool. BMC Med Imaging 15:29. https://doi.org/10.1186/S12880-015-0068-X
doi: 10.1186/S12880-015-0068-X
pubmed: 26263899
pmcid: 4533825
Pietsch AM, Viehöver A, Diem R et al (2023) Quantification and proximal-to-distal distribution pattern of tibial nerve lesions in relapsing-remitting multiple sclerosis: assessment by MR neurography. Clin Neuroradiol 33:383–392. https://doi.org/10.1007/S00062-022-01219-1
doi: 10.1007/S00062-022-01219-1
pubmed: 36264352
Mooshage CM, Schimpfle L, Kender Z et al (2023) Association of small fiber function with microvascular perfusion of peripheral nerves in patients with type 2 diabetes: study using quantitative sensory testing and magnetic resonance neurography. Clin Neuroradiol 34:55–66. https://doi.org/10.1007/S00062-023-01328-5
doi: 10.1007/S00062-023-01328-5
pubmed: 37548682
pmcid: 10881621
Bernard O, Lalande A, Zotti C et al (2018) Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: Is the problem solved? IEEE Trans Med Imaging 37:2514–2525. https://doi.org/10.1109/tmi.2018.2837502
doi: 10.1109/tmi.2018.2837502
pubmed: 29994302
Isensee F, Maier-Hein KH (2019) An attempt at beating the 3D U-Net. Preprint at https://doi.org/10.24926/548719.001
Menze BH, Jakab A, Bauer S et al (2015) The multimodal brain tumor image segmentation benchmark (BRATS). IEEE Trans Med Imaging 34:1993–2024. https://doi.org/10.1109/tmi.2014.2377694
doi: 10.1109/tmi.2014.2377694
pubmed: 25494501
Jende JME, Groener JB, Kender Z et al (2020) Structural nerve remodeling at 3-T MR neurography differs between painful and painless diabetic polyneuropathy in type 1 or 2 diabetes. Radiology 294:405–414. https://doi.org/10.1148/radiol.2019191347
doi: 10.1148/radiol.2019191347
pubmed: 31891321
Aydin OU, Taha AA, Hilbert A et al (2021) On the usage of average Hausdorff distance for segmentation performance assessment: hidden error when used for ranking. Eur Radiol Exp 5:4. https://doi.org/10.1186/s41747-020-00200-2
doi: 10.1186/s41747-020-00200-2
pubmed: 33474675
pmcid: 7817746