Automated Computer Vision Assessment of Hypomimia in Parkinson Disease: Proof-of-Principle Pilot Study.
Parkinson disease
computer vision
hypomimia
telemedicine
Journal
Journal of medical Internet research
ISSN: 1438-8871
Titre abrégé: J Med Internet Res
Pays: Canada
ID NLM: 100959882
Informations de publication
Date de publication:
22 02 2021
22 02 2021
Historique:
received:
04
06
2020
accepted:
18
12
2020
revised:
30
07
2020
entrez:
22
2
2021
pubmed:
23
2
2021
medline:
18
5
2021
Statut:
epublish
Résumé
Facial expressions require the complex coordination of 43 different facial muscles. Parkinson disease (PD) affects facial musculature leading to "hypomimia" or "masked facies." We aimed to determine whether modern computer vision techniques can be applied to detect masked facies and quantify drug states in PD. We trained a convolutional neural network on images extracted from videos of 107 self-identified people with PD, along with 1595 videos of controls, in order to detect PD hypomimia cues. This trained model was applied to clinical interviews of 35 PD patients in their on and off drug motor states, and seven journalist interviews of the actor Alan Alda obtained before and after he was diagnosed with PD. The algorithm achieved a test set area under the receiver operating characteristic curve of 0.71 on 54 subjects to detect PD hypomimia, compared to a value of 0.75 for trained neurologists using the United Parkinson Disease Rating Scale-III Facial Expression score. Additionally, the model accuracy to classify the on and off drug states in the clinical samples was 63% (22/35), in contrast to an accuracy of 46% (16/35) when using clinical rater scores. Finally, each of Alan Alda's seven interviews were successfully classified as occurring before (versus after) his diagnosis, with 100% accuracy (7/7). This proof-of-principle pilot study demonstrated that computer vision holds promise as a valuable tool for PD hypomimia and for monitoring a patient's motor state in an objective and noninvasive way, particularly given the increasing importance of telemedicine.
Sections du résumé
BACKGROUND
Facial expressions require the complex coordination of 43 different facial muscles. Parkinson disease (PD) affects facial musculature leading to "hypomimia" or "masked facies."
OBJECTIVE
We aimed to determine whether modern computer vision techniques can be applied to detect masked facies and quantify drug states in PD.
METHODS
We trained a convolutional neural network on images extracted from videos of 107 self-identified people with PD, along with 1595 videos of controls, in order to detect PD hypomimia cues. This trained model was applied to clinical interviews of 35 PD patients in their on and off drug motor states, and seven journalist interviews of the actor Alan Alda obtained before and after he was diagnosed with PD.
RESULTS
The algorithm achieved a test set area under the receiver operating characteristic curve of 0.71 on 54 subjects to detect PD hypomimia, compared to a value of 0.75 for trained neurologists using the United Parkinson Disease Rating Scale-III Facial Expression score. Additionally, the model accuracy to classify the on and off drug states in the clinical samples was 63% (22/35), in contrast to an accuracy of 46% (16/35) when using clinical rater scores. Finally, each of Alan Alda's seven interviews were successfully classified as occurring before (versus after) his diagnosis, with 100% accuracy (7/7).
CONCLUSIONS
This proof-of-principle pilot study demonstrated that computer vision holds promise as a valuable tool for PD hypomimia and for monitoring a patient's motor state in an objective and noninvasive way, particularly given the increasing importance of telemedicine.
Identifiants
pubmed: 33616535
pii: v23i2e21037
doi: 10.2196/21037
pmc: PMC7939934
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
e21037Informations de copyright
©Avner Abrami, Steven Gunzler, Camilla Kilbane, Rachel Ostrand, Bryan Ho, Guillermo Cecchi. Originally published in the Journal of Medical Internet Research (http://www.jmir.org), 22.02.2021.
Références
Psychol Bull. 1984 Jan;95(1):52-77
pubmed: 6242437
NPJ Digit Med. 2020 Jan 15;3:5
pubmed: 31970290
JMIR Hum Factors. 2017 Apr 24;4(2):e11
pubmed: 28438724
Neuropsychology. 2019 Nov;33(8):1163-1173
pubmed: 31478721
Cogn Behav Neurol. 2019 Dec;32(4):247-255
pubmed: 31800485
Sci Rep. 2020 Apr 30;10(1):7377
pubmed: 32355166
J Neurol Neurosurg Psychiatry. 2013 Jun;84(6):681-5
pubmed: 23236012
Pharmacopsychiatria. 1982 Nov;15(6):192-6
pubmed: 7156182
J Int Neuropsychol Soc. 2006 Nov;12(6):765-73
pubmed: 17064440
Cogent Psychol. 2017;4:
pubmed: 29607351
JMIR Mhealth Uhealth. 2015 Mar 26;3(1):e29
pubmed: 25830687
Neuropsychiatr Dis Treat. 2013;9:1137-44
pubmed: 23966784
J Neurosci Methods. 2020 Feb 1;331:108524
pubmed: 31747554
Philos Trans R Soc Lond B Biol Sci. 2009 Dec 12;364(1535):3453-8
pubmed: 19884140
J Neurol Neurosurg Psychiatry. 2019 Jun;90(6):704-711
pubmed: 30455406
Parkinsonism Relat Disord. 2014 Apr;20(4):370-5
pubmed: 24508573
J Neural Transm (Vienna). 2018 Dec;125(12):1819-1827
pubmed: 30343335
Craniomaxillofac Trauma Reconstr. 2015 Mar;8(1):1-13
pubmed: 25709748
Comput Math Methods Med. 2014;2014:427826
pubmed: 25478003
J Med Internet Res. 2019 Apr 12;21(4):e11109
pubmed: 30977734
NPJ Digit Med. 2020 Jan 17;3:6
pubmed: 31970291
Front Psychol. 2016 Jun 07;7:780
pubmed: 27375505
J Neurol Neurosurg Psychiatry. 1988 Mar;51(3):362-6
pubmed: 3361329
J Neurol Sci. 2014 Dec 15;347(1-2):332-6
pubmed: 25467144
JMIR Res Protoc. 2015 Mar 06;4(1):e30
pubmed: 25803512
J Neurosci Methods. 2017 Apr 1;281:7-20
pubmed: 28223023
Sensors (Basel). 2018 Nov 16;18(11):
pubmed: 30453518