Shigeru Kiryu1, Koichiro Yasaka2, Hiroyuki Akai2, Yasuhiro Nakata3, Yusuke Sugomori4, Seigo Hara4, Maria Seo4, Osamu Abe5, Kuni Ohtomo6. 1. Department of Radiology, International University of Health and Welfare Hospital, 537-3, Iguchi, Nasushiobara, Tochigi, 329-2763, Japan. kiryu-tky@umin.ac.jp. 2. Department of Radiology, The Institute of Medical Science, The University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo, 108-8639, Japan. 3. Department of Neuroradiology, Tokyo Metropolitan Neurological Hospital, 2-6-1 Musashidai, Fuchu, Tokyo, 183-0042, Japan. 4. MICIN, Inc., Nihon Building 13F, 2-6-2 Otemachi, Chiyoda-ku, Tokyo, 100-0004, Japan. 5. Department of Radiology, Graduate School of Medicine, The University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo, 113-0033, Japan. 6. International University of Health and Welfare, 2600-1 kitakanamaru, Otawara, Tochigi, 324-8501, Japan.
Abstract
OBJECTIVES: To evaluate the diagnostic performance of deep learning with the convolutional neural networks (CNN) to distinguish each representative parkinsonian disorder using MRI. METHODS: This clinical retrospective study was approved by the institutional review board, and the requirement for written informed consent was waived. Midsagittal T1-weighted MRI of a total of 419 subjects (125 Parkinson's disease (PD), 98 progressive supranuclear palsy (PSP), and 54 multiple system atrophy with predominant parkinsonian features (MSA-P) patients, and 142 normal subjects) between January 2012 and April 2016 was retrospectively assessed. To deal with the overfitting problem of deep learning, all subjects were randomly divided into training (85%) and validation (15%) data sets with the same proportions of each disease and normal subjects. We trained the CNN to distinguish each parkinsonian disorder using single midsagittal T1-weighted MRI with a training group to minimize the differences between predicted output probabilities and the clinical diagnoses; then, we adopted the trained CNN to the validation data set. Subjects were classified into each parkinsonian disorder or normal condition according to the final diagnosis of the trained CNN, and we assessed the diagnostic performance of the CNN. RESULTS: The accuracies of diagnostic performances regarding PD, PSP, MSA-P, and normal subjects were 96.8, 93.7, 95.2, and 98.4%, respectively. The areas under the receiver operating characteristic curves for distinguishing each condition from others (PD, PSP, MSA-P, and normal subjects) were 0.995, 0.982, 0.990, and 1.000, respectively. CONCLUSION: Deep learning with CNN enables highly accurate discrimination of parkinsonian disorders using MRI. KEY POINTS: • Deep learning convolution neural network achieves differential diagnosis of PD, PSP, MSA-P, and normal controls with an accuracy of 96.8, 93.7, 95.2, and 98.4%, respectively. • The areas under the curves for distinguishing between PD, PSP, MSA-P, and normality were 0.995, 0.982, 0.990, and 1.000, respectively. • CNN may learn important features that humans not notice, and has a possibility to perform previously impossible diagnoses.
OBJECTIVES: To evaluate the diagnostic performance of deep learning with the convolutional neural networks (CNN) to distinguish each representative parkinsonian disorder using MRI. METHODS: This clinical retrospective study was approved by the institutional review board, and the requirement for written informed consent was waived. Midsagittal T1-weighted MRI of a total of 419 subjects (125 Parkinson's disease (PD), 98 progressive supranuclear palsy (PSP), and 54 multiple system atrophy with predominant parkinsonian features (MSA-P) patients, and 142 normal subjects) between January 2012 and April 2016 was retrospectively assessed. To deal with the overfitting problem of deep learning, all subjects were randomly divided into training (85%) and validation (15%) data sets with the same proportions of each disease and normal subjects. We trained the CNN to distinguish each parkinsonian disorder using single midsagittal T1-weighted MRI with a training group to minimize the differences between predicted output probabilities and the clinical diagnoses; then, we adopted the trained CNN to the validation data set. Subjects were classified into each parkinsonian disorder or normal condition according to the final diagnosis of the trained CNN, and we assessed the diagnostic performance of the CNN. RESULTS: The accuracies of diagnostic performances regarding PD, PSP, MSA-P, and normal subjects were 96.8, 93.7, 95.2, and 98.4%, respectively. The areas under the receiver operating characteristic curves for distinguishing each condition from others (PD, PSP, MSA-P, and normal subjects) were 0.995, 0.982, 0.990, and 1.000, respectively. CONCLUSION: Deep learning with CNN enables highly accurate discrimination of parkinsonian disorders using MRI. KEY POINTS: • Deep learning convolution neural network achieves differential diagnosis of PD, PSP, MSA-P, and normal controls with an accuracy of 96.8, 93.7, 95.2, and 98.4%, respectively. • The areas under the curves for distinguishing between PD, PSP, MSA-P, and normality were 0.995, 0.982, 0.990, and 1.000, respectively. • CNN may learn important features that humans not notice, and has a possibility to perform previously impossible diagnoses.
Entities:
Keywords:
Artificial intelligence; Deep learning; Magnetic resonance imaging; Parkinson disease; ROC curve
Authors: Kai Boelmans; Nils Christian Bodammer; Bogdana Suchorska; Jörn Kaufmann; Georg Ebersbach; Hans-Jochen Heinze; Ludwig Niehaus Journal: Parkinsonism Relat Disord Date: 2010-06-22 Impact factor: 4.891
Authors: I Litvan; Y Agid; D Calne; G Campbell; B Dubois; R C Duvoisin; C G Goetz; L I Golbe; J Grafman; J H Growdon; M Hallett; J Jankovic; N P Quinn; E Tolosa; D S Zee Journal: Neurology Date: 1996-07 Impact factor: 9.910
Authors: M Cosottini; R Ceravolo; L Faggioni; G Lazzarotti; M C Michelassi; U Bonuccelli; L Murri; C Bartolozzi Journal: Acta Neurol Scand Date: 2007-07 Impact factor: 3.209
Authors: S Gilman; G K Wenning; P A Low; D J Brooks; C J Mathias; J Q Trojanowski; N W Wood; C Colosimo; A Dürr; C J Fowler; H Kaufmann; T Klockgether; A Lees; W Poewe; N Quinn; T Revesz; D Robertson; P Sandroni; K Seppi; M Vidailhet Journal: Neurology Date: 2008-08-26 Impact factor: 9.910
Authors: Jennifer G Goldman; Ian O Bledsoe; Doug Merkitch; Vy Dinh; Bryan Bernard; Glenn T Stebbins Journal: Neurology Date: 2017-02-24 Impact factor: 9.910
Authors: Gary H Chang; David T Felson; Shangran Qiu; Ali Guermazi; Terence D Capellini; Vijaya B Kolachalama Journal: Eur Radiol Date: 2020-02-13 Impact factor: 5.315
Authors: Dmitry Cherezov; Rahul Paul; Nikolai Fetisov; Robert J Gillies; Matthew B Schabath; Dmitry B Goldgof; Lawrence O Hall Journal: Tomography Date: 2020-06