فصلنامه مهندسی پزشکی زیستی
سال یکم شماره 4 (زمستان 1386)

Episodic memory is the explicit recollection of incidents occurred at a particular time and place in One’s Personal Past. In This Study, Detection of Episodic Memory Activity In Event Related Potentials (ERPs) was done. ERPs were recorded while the subjects made old/new recognition judgments on the new unstudied meaningless pictures and the old pictures which had been presented at the study phase. In order to extract the features correlated with the episodic memory activity, time and time-frequency features were extracted from ERPs. Wavelet method was implemented for feature extraction in time-­frequency. Independent sample test has was for detection of the separable degree the between old/new ERPs. Furthermore, by using stepwise linear discriminate analysis, ERP signals were classified to old and new classes. Ultimately for better classification between old/new ERPs, Multilayer Perceptron was implemented, and for best feature selection, genetic algorithm was used. In the best results, by using time domain features extracted from Pz channel, 100% accuracy in the training and test data was obtained.

Dynamic signature verification based on temporal features are more precise than the static methods because in addition to position information of the drawing pattern, it uses local and global features extracted from velocity, acceleration, pressure and pen angle signals, while static methods only use image information. In this study, we segmented the signature patterns using the basic role of velocity in the control process of skilled movements and then the function features were extracted. In order to signal the matching evaluation, we applied five generalized functions and five weighting strategies for score level fusion. The results showed that the correlation criterion had the minimum error. The experiments on the database, consisting of persons of Persian, Chinese and English, showed that the skilled forgeries obtained an equal error rate (EER) of 0.87% and 1.24% for the user and universal thresholds, respectively.

Two different types of computer modeling, i.e., the elastic and hyperelastic plane strain models were employed and compared with each other. Using finite element analysis, we determined a suitable model for describing the biomechanical behavior of the brain, especially the deformation and displacement of the brain ventricles. The CT-Scan of an epidural hematoma patient was modeled using both approaches. Then, by varying the mechanical parameters of the tissue (i.e., C10, C01, E, and v) and the internal ventricular pressure, the displacement rate of the corresponding points in the ventricles was simulated. Finally, the results of the simulation were compared with those of the actual ventricles, and then, the data set with the least amount of error was identified. For various types of loadings and with different pressure gradients, the results of the simulation show that if the effect of an increase in the internal pressure of the ventricles is neglected, it will lead to unrealistic results. Particularly, in unidirectional strain loading with a pressure gradient of zero (AP= 0), the walls of the ventricle adjacent to the hematoma will collapse completely. The best results were obtained for the elastic model where ΔP = 9.4 mmHg (1.25 kPa) and for the hyperelastic model where ΔP = 7.5 mmHg (1.00 kPa). These findings are consistent with the clinical conditions of the patient. In the plane strain biomechanical modeling, for unidirectional strain loading (conditions which are similar to the application of navigation systems in surgeries), neglecting the geometry and the variation of the internal pressure of the ventricles will not lead to acceptable results. Taking into account the above­mentioned parameters in describing the mechanical behavior of the brain (for epidural hematoma lesions), the elastic model (88.7% average relative accuracy) brings about better results compared with those of the hyperclastic model (86.9% average relative accuracy).

A 3D anatomically accurate finite element model of the human first cervical vertebra (atlas), including cortical and cancellous bones, was developed in ANSYS 9 based on CT-scan images. The main objective was to investigate the effect of cancellous bone on the value and distribution of maximum and average Von Mises stress in Atlas. The results showed that the material property of cancellous bone has no significant effect on the location of maximum stress and the pattern of average stress distribution in anterior arch, the junction of posterior arch and lateral mass and the groove of the posterior arch. Although the presence of cancellous bone in the model yielded higher values for the maximum and average stresses. The boundary condition had a considerable effect on this increase. Altering the material property of cancellous bone under neutral and hyperextension loading configurations, affected the average stress only in cancellous bone in the lateral mass, but change in the material property of cortical bone resulted in average stress change both in the cortical and cancellous bones, and in the lateral displacement of the lateral mass as well. The interconnected effects of changing the material properties of these two bone tissues were also studied.

The extraction of the fetal electrocardiogram (FECG) from the skin electrode signals recorded of the mother's body is a problem of concern to signal processing. Blind signal separation (BSS) technique that separates some signals from their combinations without acknowledgments about transmission channel, is a fundamental method for solving this problem. The most proposed BSS algorithm for separation of fetal electrocardiogram (FECG) and mother electrocardiogram (MECG) relies on the independence of these signals (ICA). This paper introduces a novel technique for the cases that signals are correlated with each other, i.e. considering a real assumption. The method uses Wold decomposition principle for extracting the desired and proper information from the predictable part of the measured data, and exploits approaches based on the second-order statistics to estimate source signals. Simulation results are showed the effectiveness of the method for separation of electrocardiogram signals.

One of the most accurate techniques for human identification is based on the uniqueness of the retinal blood vessels pattern. In this paper, we present a new approach for human identification using retina image. This approach is insensitive to rotation, scaling and translation. The Fourier-Mellin transform coefficients and moments of the retinal image were used to extract the suitable features. To compensate the rotational effects caused by different relative positions of the retina scanner with respect to the eye, a rotation compensator was designed. For retinal image interpretation, the optic disc location was considered as a fixed and reference point. For its localization, the Haar wavelet and the Snakes model were used. The experimental results demonstrated an error rate close to zero for the proposed method.

The optimal doses of Doxorubicin and Cyclophosphamide (AC) regimen in pre-operation Neoadjuvant chemotherapy for the patients suffering from stage III breast cancer were investigated. The major benefit of Neoadjuvant chemotherapy is that it can shrink large cancers so that they are small enough to be removed by lumpectomy instead of mastectomy. The optimal regimen designed in this paper was based on the special conditions that every patient had been treated by her/his own physician and the resistance of tumor cells. With respect to these regimen that can achieve non equivalent doses of drug in treatment times for neoadjuvant chemotherapy. The purpose of treating the patients with cancer in neoadjuvant chemotherapy could be either destroying the cancer cells or preserving the normal cell populations in the best way, or different cases between these two situations. In this article, by solving a cost function involved with the dynamics of both cancer cells and normal cells - using the appropriate weighting coefficients suggested by the treating physician- the optimal doses of AC drugs for the patients suffering from breast cancer at stage III were computed by the proposed optimal controller.