EMD-BASED FEATURE EXTRACTION AND SELECTION FOR SUBJECTS CLASSIFICATION 

Get Complete Project Material File(s) Now! »

The main components of the postural system

The tonic postural system has two types of entries. The input related to external in-formation and those related to interior information. The sensory input provides the observed information, the body orientation, as well as the external environment status. The postural system takes into account the information related to the position of each body segment with respect to others as well as the whole body position with respect to its environment. In the following subsections, the components of the tonic postural system are presented. (1) the cephalic sensors including the ocular sensor and vesti-bular system ; (2) the primary sensory organ of equilibrium (the foot) ; (3) the central regulation system and motor response.

Ocular sensor

The ocular sensor provides two di erent types of information. The rst type is purely visual information, when the retinal picture is transmitted to the central nervous system. The second type of information is related to the tension of the external oculomotor muscles (Figure 2.1).
The retina provides information about the position and the movement of the body in space. This is achieved thanks to the retina sensory receptors which transmits foveal and peripheral visual information. The foveal vision allows the identi cation of objects and provides the main directions, i.e. the vertical and horizontal ones. The peripheral vision gives information about the situation of the subject with respect to its environment. This type of vision is involved in the dynamic equilibrium [13].

Vestibular system

The vestibular system is one from the essential sensory systems that provide and main-tain the human body stability. It is located in the inner ear as a 3D motion detector (Figure 2.2). The inner ear consists of two distinguished parts, the rst one has a neu-rosensory canal for the hearing function. The second part is the vestibule which is the responsible of the equilibrium function. The vestibule has three semicircular canals po-sitioned at right angles to each other in the superior, posterior and horizontal positions. These canals are very sensitive during body movements and detect any displacement in the three planes of the space. The rst canal detects the displacements on the horizontal plane, the second one detects the displacements on the frontal plane, and the last one, the displacements on the sagittal plane. These canals are attached to the utricle which communicates with the saccule. These two organs controls the positions of the head, in the horizontal and vertical planes respectively [5, 15, 16].
The three canals are lled with a uid called endolymph. When the head rotates, the endolymphatic uid within the concerned canal lags behind because of inertia, and exerts pressure that de ects the cupula in the opposite direction. This de ection stimulates the hair cells by bending their stereocilia in the opposite direction. The receptor then sends impulses to the brain about movement from the speci c canal that is stimulated. When the vestibular organs on both sides of the head are functioning properly, they send symmetrical impulses to the brain (Figure 2.2).

The base of support : the foot

The foot is an important organ for the postural stability process, which is also the contact area between the human body and the ground. It informs the postural system on the geometry of the body support zone on the ground and also on the characteristics of the reaction force acting on that zone. The foot is equipped with multiple sensory receptors at di erent levels : cutaneous, joint, tendon and muscle. The proprioception of the foot is about four times higher than that of the leg. By transmitting the ground reaction force to the body, the foot accurately adjusts the posture of the human body. Indeed, plantar soles continuously indicate the di erential pressure between the two plantar vaults. In consequence, the feet generates its own internal forces and adapts its compliance [16].
Each foot consists of three support points which constitute three arches : The internal arch is normally the most hollow goes from the rst metatarsal head to the support center of the calcaneus (Figure 2.3). The external arch is much less hollow and goes from the fth metatarsal head to the support center of the calcaneus. The anterior arch is relatively at and goes from the head of the rst metatarsal head of the fth metatarsal.
The feet position on the ground and their symmetrical arches determine the support polygon. This polygon is constituted by the surface of the foot on the ground. Normally, the projection of the body center of gravity passes through the center of the polygon in the static posture. In contrast, a projection appearing outside the polygon, induces a balance problem.

Central regulation system

Central regulation is based on the actions of superiors centers. The nerve impulses lead to cortical and subcortical structures. Their integrative action allows control of all com-ponents of the tonic postural system through intermediary re exes. The control of sta-bilizing look is possible thanks to vestibulo-ocular and visual-oculomotor re exes. The vestibulo-spinal and vestibular-ocular-cervical re exes allow the overall control and the maintenance of posture by their action on the myotatic re ex [18].

Motor response

The whole skeleton-Musculature provides a chain of articulated segments. The form of these segments, the functional distribution of muscles, and the degrees of freedom of the various joints especially in the lower limbs for the standing posture cause e ective movements to maintain their position and thus body stability [19]. The muscles are the main e ectors of corporal movement. The adaptation of postural disturbances involves the following muscles : At the level of the posterior compartment of the leg innervated by the tibial nerve, the soleus is a primary agonist in standing position. The muscles of the anterior compartment innervated by the deep peroneal nerve are involved in the dorsal exion of the ankle with the main agonist anterior tibialis. Among the muscles acting on the knee exion, there is the semitendinosus, which has as primary antagonist, the right thigh and the vast quadriceps (medial and lateral).
According to the directions in which postural adjustments are required, the osteoarti-cular biomechanics and muscular system use speci c strategies such as, ankle strategy, hip strategy and stepping strategy.

READ  Epidemiologic classification of HPV types associated with cervical cancer

The primary strategies to maintain stability

Generally, the postural control system uses three types of movement strategies (Figure 2.4) [18]. The rst strategy considers the whole body to be an inverted pendulum that moves around the ankle to maintain the equilibrium during quiet stance (ankle strategy). This strategy is used for small disturbances to maintain the body center of mass (COM) within the feet polygon.
The second strategy uses fast movements when the body exerts torque at the hip in order to generate an appropriate center of mass position to avoid falling (hips strategy). This strategy is used for rapid external disturbances and for small support surfaces.
The third strategy consists of moving the feet (stepping) to bring back the human body to a stable position (stepping strategy) [20{22]. It is used when the person has a high risk of falling.

Tools for evaluating postural stability

In fact, any person is not able to remain perfectly standing without movements of low amplitudes. To assess the performance of the human postural system, it is necessary to assess the movements of the center of mass of human body in static and/or in dynamic situations. Consequently, in almost static situations, precise systems are necessary to detect these weak oscillations of the human body. The rst stabilometric measurements were recorded by Karl Vierordt in 1960 (Figure 2.5) [24]. The equipment used by Vierordt to record the postural sway was rudimentary : a feather attached to the top of a helmet scratching a sheet coated by black carbon, and attached to the ceiling. Only the envelope of the drawing made through this feather worn by the person for each of the following conditions : (a) with eyes open, (a’) with eyes closed ; (b) the right leg being the support ; (c) sitting with eyes open, (c’) sitting with eyes closed ; and (d) standing on the only right foot. In the literature, there are several methods to measure the human body displacements in quiet standing. Three categories of methods can be identi ed :
1. Video-based methods [25, 26],
3. Inertial sensors-based methods [27].

Table of contents :

1 GENERAL INTRODUCTION 
2 THE HUMAN POSTURAL STABILITY ANALYSIS 
2.1 Introduction
2.2 The human postural system
2.2.1 Denition
2.2.2 The main components of the postural system
2.2.2.1 Ocular sensor
2.2.2.2 Vestibular system
2.2.2.3 The base of support : the foot
2.2.2.4 Central regulation system
2.2.2.5 Motor response
2.2.3 The primary strategies to maintain stability
2.3 Tools for evaluating postural stability
2.3.1 Postural recording systems
2.3.1.1 Video-based methods
2.3.1.2 Body-worn inertial sensors-based methods
2.3.1.3 Force platform-bsed methods : the stabilometer
2.3.2 Protocols for COP displacements recordings
2.3.3 Clinical stabilometry standardization
2.4 Postural stability analysis techniques
2.5 Conclusion
3 EMD-BASED APPROACH FOR POSTURE ANALYSIS 
3.1 Introduction
3.2 Stabilometric data acquisition protocol
3.3 Empirical Mode Decomposition and its variant Ensemble Empirical Mode decomposition
3.3.1 EMD basics
3.3.2 Sifting process
3.3.3 Stopping criteria
3.3.4 Ensemble Empirical Model Decomposition
3.4 Stabilogram-diusion analysis
3.4.1 Brownian motion
3.4.2 Mean square displacement
3.5 EMD-based approach for posture analysis
3.5.1 Diusion curves modeling
3.6 Results and discussions
3.6.1 Balance analysis : classical approach
3.6.2 Balance analysis using EMD
3.6.2.1 Gain analysis
3.6.2.2 CP analysis
3.7 Conclusion
4 EMD-BASED FEATURE EXTRACTION AND SELECTION FOR SUBJECTS CLASSIFICATION 
4.1 Introduction
4.2 Classication techniques
4.2.1 K Nearest Neighbors
4.2.2 Classication and regression tree (CART)
4.2.3 Random Forest
4.2.4 Support Vector Machine
4.3 Feature extraction and selection for subjects classication
4.3.1 Feature extraction
4.4 Experimental results
4.4.1 Performance evaluation
4.4.2 Results and discussions
4.4.2.1 Obtained results using data collected from all conditions
4.4.2.2 Obtained results using data collected from each condition (IMFs data)
4.4.2.3 Obtained results using data collected from each condition (Raw data)
4.5 Conclusion
5 HMM-BASED CLASSIFICATION APPROACH 
5.1 Introduction
5.2 Hidden Markov Models
5.2.1 Introduction
5.2.2 Markov Chain
5.2.3 Discrete HMM
5.2.4 Gaussian HMM
5.3 HMM-based classication approach
5.4 Results and discussions
5.5 Conclusion
6 HMMREGRESSION-BASED APPROACH FOR AUTOMATIC SEG- MENTATION OF STABILOMETRIC SIGNALS 
6.1 Introduction
6.2 Hidden Markov Model Regression
6.2.1 Simple Hidden Markov Model Regression
6.2.1.1 Parameter estimation
6.2.2 Multiple Hidden Markov Model Regression
6.2.2.1 Parameter estimation
6.3 HMM Regression-based approach for automatic segmentation of stabilometric signals
6.4 Results and discussions
6.4.1 Segmentation based on feet and visual conditions of healthy subjects
6.4.2 Segmentation based on feet and visual conditions of PD subjects .
6.5 Conclusion
7 GENERAL CONCLUSION AND PERSPECTIVES 
7.1 Conclusion
7.2 Perspectives
Bibliographie 

GET THE COMPLETE PROJECT

Related Posts