Feature extraction via KPCA for classification of gait patterns

Wu, Jingxian; Wang, Jue; Liu, Li

doi:10.1016/j.humov.2007.01.015

Cited by 93 publications

(91 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The previous studies reported classification rates of 89.6% based on kinematic variables (Wu et al, 2006) and of 91.0% based on the combination of kinematic and spatio-temporal variables (Wu et al, 2007). Based on the proposed approach in the current paper, a classification rate of 95.8% (Fig.…”

Section: Discussionmentioning

confidence: 53%

“…Previous studies used pattern classification methods to differentiate gait patterns of young-elderly groups based on such kinematic variables (Wu et al, 2006) or the combination of kinematic and spatio-temporal variables (Wu et al, 2007) with classification rates of 89.6% and 91%, respectively. While these classification rates indicate the ability of pattern classification to differentiate the group gait patterns, a possible loss of information may have been introduced by the methods that were applied to the data.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Marker-based classification of young–elderly gait pattern differences via direct PCA feature extraction and SVMs

Eskofier

Federolf

Kugler

et al. 2013

Computer Methods in Biomechanics and Biomedical Engineering

View full text Add to dashboard Cite

This file was dowloaded from the institutional repository Brage NIH -brage.bibsys.no/nih Eskofier, B. M., Federolf, P., Kugler, P. F., Nigg, B. (2013 AbstractThe classification of gait patterns has great potential as a diagnostic tool, for example for injury diagnostic or to identify at-risk gait in the elderly. This paper has the purpose of presenting a method for classifying gait pattern group differences using the complete spatial and temporal information of the segment motion quantified by markers. The obtained classification rates are compared to previous studies using conventional classification features.For our analysis, 37 three-dimensional marker trajectories were collected from each of our 24 young and 24 elderly female subjects while they were walking on a treadmill. Principal component analysis was performed on these trajectories to retain the spatial and temporal information in the markers. Using a SupportVector Machine with linear kernel, a classification rate of 95.8% was obtained.This classification approach also allowed visualization of the contribution of individual markers to group differentiation in position and time.The approach made no specific assumptions and did not require prior knowledge of specific time points in the gait cycle. It is therefore directly applicable for group classification tasks in any study involving marker measurements.

show abstract

Section: Discussionmentioning

confidence: 53%

Section: Introductionmentioning

confidence: 99%

Marker-based classification of young–elderly gait pattern differences via direct PCA feature extraction and SVMs

Eskofier

Federolf

Kugler

et al. 2013

Computer Methods in Biomechanics and Biomedical Engineering

View full text Add to dashboard Cite

show abstract

“…As the second method has the disadvantage that the evaluation of features is dependent upon classifier type, two classifiers, the SVM with a linear kernel (Janssen et al, 2011;Wu, Wang, & Liu, 2007) and the LDA (Lee, Roan, Smith, & Lockhart, 2009), were used to determine if gender, age, and exercise conditions were classifiable based on the PC scores. A tenfold cross validation method was applied to obtain classification rates from the classifiers for gender and age, while a leave-one-out cross validation method was used for the PFP group.…”

Section: Resultsmentioning

confidence: 99%

Do intermediate- and higher-order principal components contain useful information to detect subtle changes in lower extremity biomechanics during running?

Phinyomark

Hettinga

Osis

et al. 2015

Human Movement Science

View full text Add to dashboard Cite

“…In this study, 36 of the most representative joint parameters were extracted from the joint kinematic analysis according to a previous study [14]. The extracted joint parameters for each lower body joint are shown in Tables 3, 4 and 5, respectively.…”

Section: Joint Parameters and The Calculation Of Jnimentioning

confidence: 99%

“…Therefore, several previous studies have focused on comparing only a limited number of specific gait characteristics [5][6][7][8][9] or obtaining reduced variables using multivariate statistic techniques [10] to evaluate gait pathologies more objectively [11][12][13][14][15][16][17]. In particular, these indices have been shown to be clinically meaningful indicators not only for a more general representation of a subject's overall gait pathologies but also for more objective gait analysis.…”

Section: Introductionmentioning

confidence: 99%

A joint normalcy index to evaluate patients with gait pathologies in the functional aspects of joint mobility

et al. 2010

View full text Add to dashboard Cite

Gait analysis using 3D motion capture systems provides joint kinematic and kinetic analysis results such as joint relative angles and moments that can be use used to evaluate the degrees of pathological gait patterns. However, the complex data produced using these 3D motion capture systems can only analyzed by experts, because the gait analysis is highly coupled to the kinematics of each joint. Therefore, severalSeveral previous studies using gait analysis have relied on the data compression technique to represent gait deviation from the average normal profiles as a single value. Even though it is important to evaluate gait pathologies at the joint level, all these previous studies have just used a single value to evaluate the pathological gait pattern. Using just one variable for evaluation of a gait is limited in terms of determining which joint movement patterns are getting better during rehabilitation. Therefore, in this study, a method suitable for evaluating gait deviation during a gait was developed to provide three indices for the hip, knee and ankle joints. In addition, to validate the proposed method in clinical cases, experimental tests were conducted on thirty thirty-six normal walkers and six patients with cerebral palsy. Furthermore, to validate the proposed method in regards to rehabilitation, experimental tests were conducted on three classified walking groups with imposed ankle equinus constraints. The JNI for the hip joint, knee joint and ankle were 8.78 (±3.70), 2.92 (±3.25) and 8.79 (±4.38), respectively, in the normal walking group. However, these values were significantly different for the pathological walking group with cerebral palsy. The JNI of the hip joint, knee joint and ankle joint were 203.73 (±171.59), 81.23 (±52.13) and 248.39 (±149.99), respectively, for this group. There were also differences between any two of the three classified groups with imposed ankle equinus constraints. In particular, the JNI of the ankle joint was statistically different at the p<0.01 level, and this parameter clearly increased as the degree of the imposed ankle equinus was increased. These results demonstrate that the proposed JNI can be used as a scalar factor to evaluate the angular deviation of each joint in normal and patient groups. In addition, this approach can be adapted to evaluate rehabilitation and pre/post surgery.

show abstract

Feature extraction via KPCA for classification of gait patterns

Cited by 93 publications

References 22 publications

Marker-based classification of young–elderly gait pattern differences via direct PCA feature extraction and SVMs

Marker-based classification of young–elderly gait pattern differences via direct PCA feature extraction and SVMs

Do intermediate- and higher-order principal components contain useful information to detect subtle changes in lower extremity biomechanics during running?

A joint normalcy index to evaluate patients with gait pathologies in the functional aspects of joint mobility

Contact Info

Product

Resources

About