فصلنامه مهندسی پزشکی زیستی
سال دوم شماره 2 (تابستان 1387)

This paper proposes a model to learn Farsi handwriting in different sizes based on human behavior. This model copies a human handwritten character with imitation. The imitation includes two stages of perception and action. During the perception, the information that is needed in order to generate the character is extracted from the original pattern and during the action, the model generates a character similar to the original one. To rewrite a given character, first it is decomposed into the consecutive strokes. Each stroke is approximated by several linear subdivisions. We considered the slopes and lengths of these subdivisions as the features of a given handwriting. The model learns to write a character by learning to reproduce these features. These features are descriptive of the human handwriting behavior. The learning process becomes complete when all points of the character's trajectory have distance less than a specified distance with the original trajectory. This specified distance describes visual attention and is defined as the attention width. Attention width demonstrates the human accuracy during the different trials of learning. In our model, visual attention is adaptive and decreases as the learning progresses. After the completion of learning, Farsi letters with different sizes can be generated using only memory. In order to evaluate the performance of the model, the correlation between the original and simulated characters is used. The simulation results showed good performance of the model between different Farsi characters.

Cognition of human postural responses can provide valuable insight on the control of stability. Researchers can use this finding to design rehabilitation exercises to improve the patients, balance. This study was done with the aim of conducting theoretical and experimental investigations on human response to tilting base plate in the sagittal plane. A four-segment model with three degrees of freedom was used as a biomechanical model of human body and its motion was studied in the sagittal plane. The postures of model were found by optimization technique such that the stability of model to be optimum. Zero moment point stability criterion was applied to find the optimum posture against the tilting base plate. To verify the theoretical results experimentally, the stability measure device was designed and manufactured. In several trials, the responses of ten male healthy persons standing on a tilting platform under perturbations were recorded by using the motion analysis system. Through data analysis, the response of each subject was surveyed and the experimental and theoretical results were compared. Both the experimental and theoretical results showed that the human central nervous system evokes the ankle strategy to keep its balance under tilting base plate conditions. A good coincident between the experimental results and theoretical predictions was observed, indicating that the model basis optimization method can be well relied upon to predict the human joints angle trajectories in response to base plate tilting.

In this study, the underlying dynamics of postural control system during quiet standing were investigated. Single-subject (SS) analysis was used as the statistical technique to compare the results. Center of pressure (COP) trajectories of 21 trials of a standing healthy subject and 24 trials of a cerebrovascular attacked (CVA) patient were considered in our analysis. Complexity, dimensionality and stability of postural balance control system were evaluated using the first local minimum of auto mutual information (AMI) function, correlation dimension (Dc) and largest lyapunov exponent (LLE), respectively. The results indicated higher time delays (higher determinism), lower correlation dimension (lower active dynamical degrees of freedom) and lower LLE (increase of local stability) in the postural steadiness time series of the CVA patient in compare with the normal subject. The results showed that these measures not only can be used as pathologic measures to distinguish healthy subjects from CVA patient but also provide us new openings to disclose the postural control mechanism during a quiet standing.

Human motor system is organized for execution of various motor tasks in different and flexible ways. This is mainly achieved by the way that CNS uses the redundancy in musculoskeletal system. The kinetic redundancy in human musculoskeletal systems is a significant property by which CNS achieves many complementary goals. Following the definition and role of uncontrolled manifold for movement kinematics, the kinetic redundancy concept is explored in mathematical terms. The null space of the kinetically redundant system when certain joint Moment and/or Stiffness are needed is derived and discussed. The mathematical methods have already been developed and applied to a simpler planar model. However in this paper, the aforementioned concepts were illustrated, using a 3-dimensional 3- degree of the freedom biomechanical model of spine with 18 anatomically oriented Hill-type-model muscle fascicles. The results can shed light to the interaction mechanisms in activation patterns of muscles, seen in various tasks and exertions and can provide a significant understanding for future studies and clinical practices related to low back disorders.

Speech production is one of the most complicated physiological systems including different subsystems. These subsystems must work together in a synchronous manner. One of the important sub-systems is the jaw. Although different models have suggested for jaw, no suitable model has been proposed yet to consider the interactions between muscles, bones and nervous system. In this paper, using Spring-Damper-Mass and a nonlinear concept, we introduced a novel model for jaw movement during speech production. Experimental data were used to estimate the model parameters. Computer simulation results showed that the model could generate the jaw movement patterns similar to those observed in physiological behavior. Generality and simplicity of the model are two model features useful for more investigation of the jaw movement in different tasks.

Patients with spinal cord injury in C5/C6 levels are capable of controlling the voluntary movements of the shoulder joints, but some muscles involved in the movement of the elbow joint are paralyzed in these patients. By using FES as well as an appropriate stimulation of the paralyzed muscles, the patients can be assisted with their essential daily living activities. One of the major problems of using FES for reanimation of the paralyzed arm is to provide voluntary commands for FES control. Kinematic synergy and muscle synergy are two main options in this regard. In this paper, these two command sources were evaluated and compared. Furthermore, a mixed method was proposed, which improves performance. Thus, the EMG and kinematical data during a set of activities of daily living (AOL) were recorded and processed. Precise investigations were carried out in order to determine the appropriate values for high-level neural network controller parameters. Next, six different neural network controller structures were trained by the EMG and/or kinematical data. Using this method, cross correlation between the estimation and measurement for all records was obtained as 94.76% for kinematic synergy and 98.08%, for muscle synergy. In the mixed method, these values were improved to 94.82% and 98.84% respectively. Furthermore, mixed method paved the way to improve the performance of low-level controller with estimating the desired kinematics for the distal joint and desired activity for the paralyzed muscle.

Hish rate classification of Electromyogram (EMG) signals for controlling of prosthetic hands is still a hot topic among the rehabilitation research titles. Specially, when the degree of freedom in artificial hands increases, the classification rate decreases dramatically. In this paper, a new five layer classifier based on Neuro-Fuzzy-Genetic structure was introduced to increase the classification accuracy of EMG signals. The proposed classifier has a self- organized structure, which adaptively creates new rules according to the input features and trains the fuzzy rule weights based on the back propagation method. Finally, the genetic algorithm (GA) was employed for the final tuning stage. In this study, six subjects were asked to perform 9 different movements and their EMG signals were caught during the tasks from the six different forearm muscles. In order to remove the noises, the signals were filtered. Then the integral absolute average (IAV), Cepstrum coefficients and Wavelet Packet Coefficients with entropy pruning were extracted from the filtered signals as features. We used principal components analysis (PCA) for dimensionality reduction (234 to 10). The dimensionality reduction by PCA simplifies the structure of the classifier and reduces the processing time for the pattern recognition. The proposed classifier was applied on the features and the results were led to higher than 96.7% classification rate for the 9 classes of movement. To make a comparison, support vector machine (SVM) was employed (76% classification rate for 9 classes) and the results showed a drastic supremacy of the proposed method.