Research on feature extraction algorithm for plantar pressure image and gait analysis in stroke patients

https://doi.org/10.1016/j.jvcir.2018.12.017Get rights and content

Abstract

The plantar pressure image is an important tool for gait analysis. It has important applications in evaluating the recovery of stroke patients after operation and formulating the rehabilitation training program. It is one of the key technologies of gait analysis to extract foot feature parameters from static/dynamic plantar pressure images. This article deals with the noise in the original image through the piecewise linear grayscale transformation, the time domain mean filter and the maximum value filter, then determine the position of the feet in the image by the foot localization algorithm based on the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) and the K-means clustering method. Finally, the plantar pressure feature parameters were extracted according to the positioned images. Based on the above feature parameter extraction algorithm, the plantar pressure feature parameters of 20 healthy subjects and 20 S patients with relative recovery period (2–6 months after the onset) were compared, showing a statistically significant difference (P < 0.001). Based on the above data, gait characteristics of stroke patients were further analyzed.

Introduction

Plantar pressure refers to the pressure field between the foot and the supporting surface in the daily exercise [1]. The plantar pressure can be transformed into pressure image through the pressure sensor array, thus reflecting the change of the plantar pressure. Through the analysis of the plantar pressure image, it can reflect the gait characteristics, exercise habits, and exertion methods of the subjects. [2]. The analysis of the plantar pressure images is widely used in the study of gait rehabilitation in stroke patients [3], rehabilitation robots [4], etc.

The extraction of gait feature parameters from the plantar pressure image is the key technology of gait analysis. Research institutions in various countries have studied the extraction of different features in pressure images. Amit Kumar Vimal et al. in India [5] studied the calculation of the center of pressure (COP) based on a new type of flexible insole wearable pressure sensor, and the error is lower than the standard acquisition system (Zebris© force platform). Li et al. in China [6] analyze the plantar pressure image of diabetic patients based on the FootScan7.0 system. They preprocessed the pressure image based on the watershed transformation, and then segmented the image by the clustering method based on the convolutional neural network to obtain the characteristic region of interest. The study can be applied to the customization of diabetic shoe lasts.

In this paper, a gait feature extraction algorithm is proposed. Through pressure image preprocessing and foot localization algorithm, the static/dynamic plantar pressure feature parameters in gait characteristics are accurately calculated. Based on the above algorithm, the plantar pressure feature parameters of healthy people and stroke patients were statistically compared, and the gait characteristics of patients with stroke were analyzed.

Section snippets

Genetic algorithm

The preprocessing of pressure images includes the piecewise linear grayscale transformation, the time domain mean filter and the maximum value filter.

The plantar pressure acquisition system used in this paper is composed of a resistance pressure sensor array [7], and its conductivity is linear with the pressure received, as shown in Fig. 1. The change of the detected pressure can be converted to the change of the resistance value, and then transformed into a pressure numerical matrix through a

The foot localization algorithm

Identifying the position of the footprints in the plantar pressure image is a key step in gait analysis. It includes two parts of work: On the one hand, it is necessary to identify the respective pressure area blocks constituting the footprints, and on the other hand, it is necessary to merge the pressure area blocks into the foot shape of the left and right foot respectively.

Gait feature extraction

Gait characteristic parameters are used to quantitatively analyze human gait, which can reflect human foot structure, exercise habits and foot disease types and degrees, including static characteristic parameters and dynamic characteristic parameters.

Experiments

In this paper, 20 male healthy subjects (age 47.3 ± 6.5 years old, height 172.3 ± 7.1 cm, weight 70.9 ± 6.6 kg) and 20 male stroke patients with relative recovery period (2–6 months after the onset, right limb hemiplegia, age 55.6 ± 8.2 years old, height 170.1 ± 5.2 cm, weight 73 ± 6.9 kg) were collected.

The plantar pressure acquisition system used in this paper is shown in Fig. 10. The pressure sensor array is composed of piezoresistive sensing unit array, the sensing unit size is 5 mm × 5 mm.

Results and discussion

The static/dynamic gait parameters of healthy subjects and stroke people were calculated as shown in Table 1.

The gait feature parameters of the healthy and stroke subjects were compared with the SPSS software and the Independent Sample T-test was used for statistical analysis. For LUD:P > 0.05, there was no statistical difference. For LLR RUD RLR kLR kILR kSupport kSwing, P < 0.001, the feature parameters show significant statistical differences. Further analysis of the feature parameters can

Conclusions

In this paper, a gait feature extraction algorithm is proposed. First, deals with the noise in the original image through the piecewise linear grayscale transformation, the time domain mean filter and the maximum value filter, then determine the position of the feet in the image by the foot localization algorithm based on the DBSCAN and the K-means clustering method. Finally, the plantar pressure feature parameters were extracted according to the positioned images. Based on the above algorithm,

Declarations

Ethical Approval and Consent to participate: Approved.

Consent for publication: Approved.

Availability of supporting data: We can provide the data.

Competing interests

There is no potential competing interests in our paper. And all authors have seen the manuscript and approved to submit to your journal. We confirm that the content of the manuscript has not been published or submitted for publication elsewhere.

Funding

The project has been supported by the Special Fund for the Development of Shenzhen (China) Strategic New Industry (JCYJ20170818085946418) and the Shenzhen (China) Science and Technology Research and Development Fund (JCYJ20170306092000960).

Author’s contributions

MW completed the design of the hardware system and the research of the image processing algorithm. XAW provides guidance for the implementation of the project. ZCF and FC carried out the collection and screening of experimental data and preprocess the data, and preprocess the data. SXZ and CP completed the construction and the test of the hardware system.

Author details

Mo Wang: A405, Key Laboratory of Integrated Micro-systems Science and Engineering Applications, Shenzhen Graduate School of Peking University, Shenzhen University Town, Nanshan, Shenzhen, Guangdong, P.R.China.

Xin’an Wang: A408, Key Laboratory of Integrated Micro-systems Science and Engineering Applications, Shenzhen Graduate School of Peking University, Shenzhen University Town, Nanshan, Shenzhen, Guangdong, P.R.China.

Acknowledgements

The authors thank the editor and anonymous reviewers for their helpful comments and valuable suggestions. I want to thank all of our team members for their efforts in the project.

Mo Wang received the B.S. degree in Electronic Science and Engineering from Jilin University, Jilin, China in 2011. From 2012 to now, he is pursuing the Ph.D. degree in the school of electronic and computer engineering in Peking University Shenzhen Graduate School, Guangdong, China. His interests include research on motion analysis based on multi-sensors and rehabilitation medical electronics.

References (17)

There are more references available in the full text version of this article.

Cited by (0)

Mo Wang received the B.S. degree in Electronic Science and Engineering from Jilin University, Jilin, China in 2011. From 2012 to now, he is pursuing the Ph.D. degree in the school of electronic and computer engineering in Peking University Shenzhen Graduate School, Guangdong, China. His interests include research on motion analysis based on multi-sensors and rehabilitation medical electronics.

Xin’an Wang received the B.S. degree in computer science from Wuhan University, Wuhan, China, in 1983, and the M.S. and Ph.D. degrees in microelectronics from Shanxi Microelectronics Institute, Xi’an, China, in 1989 and 1992, respectively. He is currently a Professor with the School of Electronics Engineering and Computer Science, Peking University Shenzhen Graduate School, Beijing, China. He is currently with the School of Electronic and Computer Engineering, Peking University, Shenzhen Campus. His research interests are focused on life health and medical electronics.

Zhuochen Fan received the B.S. degree in College of Communication Engineering in Chongqing University in 2017. He is currently a graduate student in the school of electronic and computer engineering in Peking University Shenzhen Graduate School, Guangdong, China. His interests include data analysis and algorithm research in medical field.

Fei Chen received the B.S. degree from Jiangnan University, and he is currently a PhD student in the Department of Chinese and Bilingual Studies, the Hong Kong Polytechnic University. His current research interests include rehabilitation science and speech therapy.

Sixu Zhang received the B.S. degree in Electronic Science and Engineering from Jilin University, Jilin, China in 2017. He is currently a graduate student in the school of electronic and computer engineering in Peking University Shenzhen Graduate School, Guangdong, China. His interests include intelligent hardware design.

Chen Peng received the B.S. degree in College of Electronic Science and Applied Physics from Hefei University of Technology, Anhui, China in 2017. He is currently a graduate student in the school of electronic and computer engineering in Peking University Shenzhen Graduate School, Guangdong, China. His interests include kinematics analysis and medical electronics.

This article is part of the Special Issue on TIUSM.

View full text