Dong Ah Lee1, Ho-Joon Lee2, Hyung Chan Kim1, Kang Min Park3. 1. Department of Neurology, Haeundae Paik Hospital, Inje University College of Medicine, Haeundae-ro 875, Haeundae-gu, Busan, 48108, Korea. 2. Department of Radiology, Haeundae Paik Hospital, Inje University College of Medicine, Busan, Republic of Korea. 3. Department of Neurology, Haeundae Paik Hospital, Inje University College of Medicine, Haeundae-ro 875, Haeundae-gu, Busan, 48108, Korea. smilepkm@hanmail.net.
Abstract
PURPOSE: We evaluated the feasibility of machine learning analysis using diffusion tensor imaging (DTI) parameters to identify patients with idiopathic rapid eye movement (REM) sleep behavior disorder (RBD). We hypothesized that patients with idiopathic RBD could be identified via machine learning analysis based on DTI. METHODS: We enrolled 20 patients with newly diagnosed idiopathic RBD at a tertiary hospital. We also included 20 healthy subjects as a control group. All of the subjects underwent DTI. We obtained the conventional DTI parameters and structural connectomic profiles from the DTI. We investigated the differences in conventional DTI measures and structural connectomic profiles between patients with idiopathic RBD and healthy controls. We then used machine learning analysis using a support vector machine (SVM) algorithm to identify patients with idiopathic RBD using conventional DTI and structural connectomic profiles. RESULTS: Several regions showed significant differences in conventional DTI measures and structural connectomic profiles between patients with idiopathic RBD and healthy controls. The SVM classifier based on conventional DTI measures revealed an accuracy of 87.5% and an area under the curve of 0.900 to identify patients with idiopathic RBD. Another SVM classifier based on structural connectomic profiles yielded an accuracy of 75.0% and an area under the curve of 0.833. CONCLUSION: Our findings demonstrate the feasibility of machine learning analysis based on DTI to identify patients with idiopathic RBD. The conventional DTI parameters might be more important than the structural connectomic profiles in identifying patients with idiopathic RBD.
PURPOSE: We evaluated the feasibility of machine learning analysis using diffusion tensor imaging (DTI) parameters to identify patients with idiopathic rapid eye movement (REM) sleep behavior disorder (RBD). We hypothesized that patients with idiopathic RBD could be identified via machine learning analysis based on DTI. METHODS: We enrolled 20 patients with newly diagnosed idiopathic RBD at a tertiary hospital. We also included 20 healthy subjects as a control group. All of the subjects underwent DTI. We obtained the conventional DTI parameters and structural connectomic profiles from the DTI. We investigated the differences in conventional DTI measures and structural connectomic profiles between patients with idiopathic RBD and healthy controls. We then used machine learning analysis using a support vector machine (SVM) algorithm to identify patients with idiopathic RBD using conventional DTI and structural connectomic profiles. RESULTS: Several regions showed significant differences in conventional DTI measures and structural connectomic profiles between patients with idiopathic RBD and healthy controls. The SVM classifier based on conventional DTI measures revealed an accuracy of 87.5% and an area under the curve of 0.900 to identify patients with idiopathic RBD. Another SVM classifier based on structural connectomic profiles yielded an accuracy of 75.0% and an area under the curve of 0.833. CONCLUSION: Our findings demonstrate the feasibility of machine learning analysis based on DTI to identify patients with idiopathic RBD. The conventional DTI parameters might be more important than the structural connectomic profiles in identifying patients with idiopathic RBD.
Authors: Mohamed Elfil; Nada Ahmed; Amritha Alapati; Rucha Bahekar; Mohamed Kandil; Christine Kim; Sara Schaefer; Sule Tinaz; Amar S Patel; John M de Figueiredo; Elan D Louis; Brian B Koo Journal: J Neurol Date: 2019-12-04 Impact factor: 4.849
Authors: Marcus M Unger; Marcus Belke; Katja Menzler; Johannes T Heverhagen; Boris Keil; Karin Stiasny-Kolster; Felix Rosenow; Nico J Diederich; Geert Mayer; Jens C Möller; Wolfgang H Oertel; Susanne Knake Journal: Sleep Date: 2010-06 Impact factor: 5.849
Authors: Yves Dauvilliers; Carlos H Schenck; Ronald B Postuma; Alex Iranzo; Pierre-Herve Luppi; Giuseppe Plazzi; Jacques Montplaisir; Bradley Boeve Journal: Nat Rev Dis Primers Date: 2018-08-30 Impact factor: 52.329
Authors: Bong Soo Park; Yoo Jin Lee; Jin-Han Park; Il Hwan Kim; Si Hyung Park; Ho-Joon Lee; Kang Min Park Journal: Brain Behav Date: 2018-05-23 Impact factor: 2.708
Authors: Sami Nikkonen; Henri Korkalainen; Samu Kainulainen; Sami Myllymaa; Akseli Leino; Laura Kalevo; Arie Oksenberg; Timo Leppänen; Juha Töyräs Journal: Sleep Date: 2020-12-14 Impact factor: 5.849
Authors: Mary B Makarious; Hampton L Leonard; Dan Vitale; Hirotaka Iwaki; Lana Sargent; Anant Dadu; Ivo Violich; Elizabeth Hutchins; David Saffo; Sara Bandres-Ciga; Jonggeol Jeff Kim; Yeajin Song; Melina Maleknia; Matt Bookman; Willy Nojopranoto; Roy H Campbell; Sayed Hadi Hashemi; Juan A Botia; John F Carter; David W Craig; Kendall Van Keuren-Jensen; Huw R Morris; John A Hardy; Cornelis Blauwendraat; Andrew B Singleton; Faraz Faghri; Mike A Nalls Journal: NPJ Parkinsons Dis Date: 2022-04-01