Browse by author
Lookup NU author(s): Dr Sam Stuart, Dr Alan Godfrey, Professor Lynn RochesterORCiD, Dr Lisa AlcockORCiD
Full text for this publication is not currently held within this repository. Alternative links are provided below where available.
INTRODUCTION: Tracking eye-movements when walking allows inferences to be made about underlying cognitive and visual processes that may influence gait, particularly in ageing and Parkinson’s disease (PD) where such processes are commonly impaired [1]. However, very few studies are concerned with the context of eye-movements (i.e. the location of fixations). This is largely due to such analysis requiring time-consuming manual frame-by-frame inspection of eye-tracker videos [2], which can be subjective because it does not use algorithm-derived objective eye-movement outcomes. Therefore there is potential for a lack of consistency between raters.This study aimed to; 1) modify a previously developed eye-movement objective measurement algorithm [3] to provide still images of fixation locations; 2) develop a classification system for manual fixation location analysis of mobile eye-tracking data obtained when walking; and 3) assess inter-rater reliability of the proposed classification system.METHODS: An infra-red mobile eye-tracker (Dikablis, Ergoneers) recorded eye-movements during walking in healthy older adult controls (HC) (n=5) and people with Parkinson’s disease (n=5). Raw eye-tracker video data was pre-processed to eliminate tracking errors. Fixations were identified using a previously validated algorithm [3], which was adapted to provide still images of fixation locations. Fixation locations were then manually classified by two raters according to a classification system comprising of pre-defined areas within the measurement field of view. Intra-class correlation coefficients (ICC2,1) were used to determine inter-rater reliability.RESULTS: The algorithm successfully provided a total of 116 still images for the start of each fixation identified, allowing manual classification to be performed. Inter-rater reliability for classifying fixation location was high for both PD (ICC2,1=0.97, 95% agreement) and HC (ICC2,1=0.93, 91% agreement) groups, which indicated that the classification system was reliable.CONCLUSION: This study developed a reliable semi-automated contextual analysis method for eye-tracking during dynamic gait studies in HC and people with PD. Future studies could adapt this methodology for use within various laboratory or real-world eye-tracking during walking studies, which would establish a time-effective but rigorous methodological approach. With improvements in the spatial resolution of eye-trackers, future studies may also be able to fully automate this process using image detection algorithms.REFERENCES[1] Stuart (2016) Neurosci & Biobehav Rev, 62, pp.76-88. [2] Vitorio (2014) Neurosci, 277, pp.273-280. [3] Stuart et al. (2014) IEEE EMBC.
Author(s): Stuart S, Hunt D, Nell J, Godfrey A, Rochester L, Alcock L
Publication type: Conference Proceedings (inc. Abstract)
Publication status: Published
Conference Name: World Congress of the International Society for Posture and Gait Research
Year of Conference: 2017
Print publication date: 26/06/2017
Online publication date: 26/06/2017
Acceptance date: 26/06/2017
