Automated thalamic nuclei segmentation using multi-planar cascaded convolutional neural networks.
Clinical analysis
Convolutional neural network
Thalamic nuclei segmentation
White-matter-nulled MPRAGE
Journal
Magnetic resonance imaging
ISSN: 1873-5894
Titre abrégé: Magn Reson Imaging
Pays: Netherlands
ID NLM: 8214883
Informations de publication
Date de publication:
11 2020
11 2020
Historique:
received:
19
05
2020
revised:
25
07
2020
accepted:
17
08
2020
pubmed:
24
8
2020
medline:
28
1
2021
entrez:
24
8
2020
Statut:
ppublish
Résumé
To develop a fast and accurate convolutional neural network based method for segmentation of thalamic nuclei. A cascaded multi-planar scheme with a modified residual U-Net architecture was used to segment thalamic nuclei on conventional and white-matter-nulled (WMn) magnetization prepared rapid gradient echo (MPRAGE) data. A single network was optimized to work with images from healthy controls and patients with multiple sclerosis (MS) and essential tremor (ET), acquired at both 3 T and 7 T field strengths. WMn-MPRAGE images were manually delineated by a trained neuroradiologist using the Morel histological atlas as a guide to generate reference ground truth labels. Dice similarity coefficient and volume similarity index (VSI) were used to evaluate performance. Clinical utility was demonstrated by applying this method to study the effect of MS on thalamic nuclei atrophy. Segmentation of each thalamus into twelve nuclei was achieved in under a minute. For 7 T WMn-MPRAGE, the proposed method outperforms current state-of-the-art on patients with ET with statistically significant improvements in Dice for five nuclei (increase in the range of 0.05-0.18) and VSI for four nuclei (increase in the range of 0.05-0.19), while performing comparably for healthy and MS subjects. Dice and VSI achieved using 7 T WMn-MPRAGE data are comparable to those using 3 T WMn-MPRAGE data. For conventional MPRAGE, the proposed method shows a statistically significant Dice improvement in the range of 0.14-0.63 over FreeSurfer for all nuclei and disease types. Effect of noise on network performance shows robustness to images with SNR as low as half the baseline SNR. Atrophy of four thalamic nuclei and whole thalamus was observed for MS patients compared to healthy control subjects, after controlling for the effect of parallel imaging, intracranial volume, gender, and age (p < 0.004). The proposed segmentation method is fast, accurate, performs well across disease types and field strengths, and shows great potential for improving our understanding of thalamic nuclei involvement in neurological diseases.
Identifiants
pubmed: 32828985
pii: S0730-725X(20)30311-8
doi: 10.1016/j.mri.2020.08.005
pmc: PMC7530079
mid: NIHMS1624346
pii:
doi:
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
45-54Subventions
Organisme : NIAAA NIH HHS
ID : R21 AA023582
Pays : United States
Informations de copyright
Copyright © 2020 Elsevier Inc. All rights reserved.
Références
Neuroimage. 2003 Jun;19(2 Pt 1):391-401
pubmed: 12814588
Biol Psychiatry. 2018 Apr 15;83(8):648-656
pubmed: 29275841
Magn Reson Med. 1986 Aug;3(4):604-18
pubmed: 3747821
Neuroimage. 2006 Oct 15;33(1):115-26
pubmed: 16860573
Neuroimage. 2009 Aug;47 Suppl 2:T44-52
pubmed: 19362595
IEEE Trans Med Imaging. 2009 Jun;28(6):822-37
pubmed: 19068424
Neurosurgery. 2010 Dec;67(6):1745-56; discussion 1756
pubmed: 21107206
J Neurosurg. 1996 Feb;84(2):203-14
pubmed: 8592222
IEEE Trans Med Imaging. 2016 May;35(5):1299-1312
pubmed: 26978662
Nat Neurosci. 2003 Jul;6(7):750-7
pubmed: 12808459
Acta Neurochir Suppl (Wien). 1993;58:39-44
pubmed: 8109299
Neuroimage. 2000 Dec;12(6):601-16
pubmed: 11112393
Neuroimage. 2015 Jul 15;115:117-37
pubmed: 25936807
Neuroimage. 2019 Jul 1;194:272-282
pubmed: 30894331
Magn Reson Med. 2015 May;73(5):1786-94
pubmed: 24889754
Brain Imaging Behav. 2014 Dec;8(4):611-20
pubmed: 24421067
Magn Reson Med. 2012 Jan;67(1):118-26
pubmed: 21656553
IEEE Trans Med Imaging. 2016 May;35(5):1252-1261
pubmed: 27046893
N Engl J Med. 2013 Aug 15;369(7):640-8
pubmed: 23944301
AJNR Am J Neuroradiol. 2000 Sep;21(8):1397-401
pubmed: 11003270
Med Image Comput Comput Assist Interv. 2013;16(Pt 2):246-53
pubmed: 24579147
Med Image Comput Comput Assist Interv. 2014;17(Pt 1):520-7
pubmed: 25333158
Semin Neurol. 2013 Apr;33(2):83-90
pubmed: 23888393
Brain Struct Funct. 2017 Jul;222(5):2203-2216
pubmed: 27888345
Neuroimage. 2018 Dec;183:314-326
pubmed: 30121337
AJNR Am J Neuroradiol. 2012 Dec;33(11):2110-6
pubmed: 22700756
Neuroimage. 2010 Feb 1;49(3):2053-62
pubmed: 19853042
Front Neuroinform. 2019 Apr 24;13:30
pubmed: 31068797
Hum Brain Mapp. 2020 Apr 1;41(5):1351-1361
pubmed: 31785046
Brain Stimul. 2012 Oct;5(4):625-33
pubmed: 22405744
Magn Reson Med. 2004 Sep;52(3):612-8
pubmed: 15334582
Neuroimage. 2011 Jun 1;56(3):907-22
pubmed: 21352927
Front Neurosci. 2016 Jun 10;10:264
pubmed: 27375422
Mov Disord. 2016 Apr;31(4):547-54
pubmed: 26846525
IEEE Trans Med Imaging. 2006 Nov;25(11):1451-61
pubmed: 17117774
Brain Struct Funct. 2015;220(3):1619-35
pubmed: 24659254
Physiol Rev. 1988 Jul;68(3):649-742
pubmed: 2839857
Neuron. 2002 Jan 31;33(3):341-55
pubmed: 11832223
Med Image Anal. 2018 Aug;48:177-186
pubmed: 29935442
Magn Reson Imaging. 2020 Jan;65:114-128
pubmed: 31629074
Mult Scler. 2020 Jul;26(8):987-992
pubmed: 30730233
Eur J Neurosci. 2010 Jun;31(12):2292-307
pubmed: 20550571
Proc IEEE Int Symp Biomed Imaging. 2013;:852-855
pubmed: 24145869
Acta Neuropathol. 1991;81(3):261-8
pubmed: 1711755
Neuroimage. 2014 Jan 1;84:534-45
pubmed: 24018302
Neurosci Biobehav Rev. 2015 Jul;54:29-37
pubmed: 25862940
Neurology. 2013 Jan 8;80(2):210-9
pubmed: 23296131
Neuroimage. 2017 Sep;158:430-440
pubmed: 28669906
Neuroimage. 2012 Aug 15;62(2):774-81
pubmed: 22248573
IEEE Trans Med Imaging. 2018 Aug;37(8):1865-1876
pubmed: 29994439
Neuropsychiatr Dis Treat. 2017 Nov 27;13:2875-2890
pubmed: 29225466
J Neurosurg. 2004 Jul;101(1):43-7
pubmed: 15255250