Vision‐based gait analysis system utilizing deep learning algorithms in idiopathic normal‐pressure hydrocephalus patients. (31st December 2021)

• Record Type: Journal Article
• Authors: Kang, Kyunghun; Jeong, Sungmoon; Yu, Hosang; Park, Jaechan
• Is Part Of: Alzheimer's & dementia. Volume 17(2021)Supplement 5
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-12-31
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.053139
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Idiopathic normal‐pressure hydrocephalus (INPH) is a rare neurological disorder. It is an idiopathic adult‐onset syndrome involving nonobstructive enlargement of cerebral ventricles, and it is known by its symptoms of cognitive impairment, gait disturbance, and urinary dysfunction. While INPH can present with any of these classic clinical symptoms in varying degrees, the most frequent and important INPH clinical feature is gait disturbance. A vision‐based gait analysis method using monocular videos was proposed to estimate temporo‐spatial gait parameters by leveraging deep learning algorithms. This study aimed to validate vision‐based gait analysis using GAITRite as the reference system and analyze relationships between Frontal Assessment Battery (FAB) scores and gait variability measured by vision‐based gait analysis in INPH patients. Method: Gait data from 46 patients were simultaneously collected from the vision‐based system utilizing deep learning algorithms and the GAITRite system. Result: There was a strong correlation in 11 gait parameters between our vision‐based gait analysis method and the GAITRite gait analysis system. Our results also demonstrated excellent agreement between the two measurement systems for all parameters except stride time variability after the cerebrospinal fluid tap test. Our data showed that stride time and stride length variability measured by the vision‐based gait analysis system were correlated with FAB scores. Conclusion: Vision‐based gait analysis utilizing deep learning algorithms can provide comparable data to GAITRite when assessing gait dysfunction in INPH. Frontal lobe functions may be associated with gait variability measurements using vision‐based gait analysis for INPH patients.