Impact of the Marker Set Configuration on the Accuracy of Gait Event Detection in Healthy and Pathological Subjects

Abstract

For interpreting outcomes of clinical gait analysis, an accurate estimation of gait events, such as initial contact (IC) and toe-off (TO), is essential. Numerous algorithms to automatically identify timing of gait events have been developed based on various marker set configurations as input. However, a systematic overview of the effect of the marker selection on the accuracy of estimating gait event timing is lacking. Therefore, we aim to evaluate (1) if the marker selection influences the accuracy of kinematic algorithms for estimating gait event timings and (2) what the best marker location is to ensure the highest event timing accuracy across various gait patterns. 104 individuals with cerebral palsy (16.0 ± 8.6 years) and 31 typically developing controls (age 20.6 ± 7.8) performed clinical gait analysis, and were divided into two out of eight groups based on the orientation of their foot, in sagittal and frontal plane at mid-stance. 3D marker trajectories of 11 foot/ankle markers were used to estimate the gait event timings (IC, TO) using five commonly used kinematic algorithms. Heatmaps, for IC and TO timing per group were created showing the median detection error, compared to detection using vertical ground reaction forces, for each marker. Our findings indicate that median detection errors can be kept within 7 ms for IC and 13 ms for TO when optimizing the choice of marker and detection algorithm toward foot orientation in midstance. Our results highlight that the use of markers located on the midfoot is robust for detecting gait events across different gait patterns.

A deep-learning approach for automatically detecting gait-events based on foot-marker kinematics in children with cerebral palsy—Which markers work best for which gait patterns?

“…On the same test-set, the best performing kinematic algorithm was capable of identifying 95% of IC and events within 33ms and 22ms [4], whereas our approach these events to within 20ms and 33ms, respectively (Fig 4). When optimised towards specific gait patterns, 95% of the events were detected within 7ms and 12ms for IC and TO respectively [22]. While our approach showed varying performance per subgroup, the kinematic algorithms only showed performance differences when optimised on a particular subgroup.…”

“…The high variability in heel-toe progression known to be present in MF, could be the reason for the MF group showing the worst performance for both IC and TO detection. Additionally, our dataset only classified MF indirectly-while, HS and FF were clearly defined based on their dorsi-flexion angles during IC (section Methods), all cases which did not correspond to the criteria for HS or FF were classified as MF [22]. The fact that we chose subgroups based on IC could also be one reason why the TO detection showed a higher inter-group variability than the IC detection.…”

“…Judging by the differences in performance on the individual gait groups, it is likely that specialized networks would perform better than the generic network trained in this study. For kinematic algorithms, the selection of optimum input markers for each gait group improved the event detection from 80ms to 7ms for IC and from 143ms to 12ms for TO detection [22]. In clinical practice, such a differentiation of networks would mean that every subject would…”

“…However, many pathologies are associated with modified walking patterns, such as crouch-gait in CP or toe-walking in idiopathic toe-walkers [28,29]. While some studies have investigated developing kinematic algorithms based on OMCS suited to specific gait patterns with pathological groups [18][19][20][21][22], and demonstrated promising results, there is only a handful of studies comparing how the tailored kinematic algorithms perform on other gait patterns [4,10,12]. In clinical practice, wide various gait patterns are assessed daily.…”

Uneven terrain treadmill walking in younger and older adults

“…We used custom MATLAB scripts for all data processing. We computed heel strike events for each foot as the time points when the ground reaction forces reached 20 N, similar to [27,28]. For each gait cycle, we calculated the step duration as the time difference between heel strikes of opposing feet.…”