Search

The study of optical illusion is an important method to elucidate the mechanism of visual perception. However, many details about the cause of optical illusions are still unclear. In this research, based on the characteristic of the physiological structure of the retina, we proposed an on-center receptive field model of the retina. Using this model, we simulated the distributions of the inducing field of some visual stimulus. Comparing to the past studies' results, the validity of the proposed model was proofed. Furthermore, we simulated the distributions of the inducing field of some typical illusions. The simulation results can explain these illusion phenomenon rationally. Therefore, it suggested that some of illusions are probably engendered by the distributions of the inducing field in the retina which generated by the illusions stimuli. The practicality of the proposed model was also verified.

The efficiency of novel tumor chemotherapeutics could be increased using targeted drug delivery by hyperthermia. In this paper, the 3D liposomal doxorubicin distribution in the tumor tissue enhanced by local hyperthermia was quantitatively studied in real time using laser confocal microscopy. Results showed that the thermally induced liposomal doxorubicin extravasation was non-uniform and more excessive in the peripheral region than that in the tumor center. The effect of the thermally targeted drug delivery was also investigated. On the 1st, 3rd day after the thermally targeted drug treatment, histological examination showed that many nucleolus were condensed and collapsed in the peripheral region. But, in the tumor center, there were no such changes found until the 3rd day. While on the 6th day, tumor cells in both the peripheral and center region were found necrotic. The enhancement of the nanoparticle anti-tumor drug effect was significant. A theoretical analysis of liposomal doxorubicin diffusion to the tumor cells in vivo was performed. Results showed that it took more than 40 hrs for the doxorubicin to get into the tumor cells in the center region from the periphery region. The theoretical results well explained the experimental observations.

This paper describes a model-based approach to the unobtrusive monitoring of elders in their home environment to assess their health, daily activities, and cognitive function. We present a semi-Markov model representation with automated learning to characterize individual elder's mobility in the home environment. The assessed mobility can be used to characterize the elder's speed of walking and can serve as one of the predictors of future cognitive functionality and the ability of elders to live independently in their home environment.

This paper describes experimental results involving the percentage cell lysis in SWLA-2 murine hybridomas produced by square wave electric field pulses of 100, 200, and 300 V across a 1 mm gap width in a standard cuvette. Pulse lengths were of 0.2 and 0.6 ms duration; 1, 2, or 3 pulses were applied with 100 ms time interval between pulses. Cells were cultured and separate samples examined at 48 hours to determine cell mortality. Nearly 90% cell mortality was produced by applying 3 pulses at of 0.6 ms duration at 300 V.

In the present study, the feasibility of applying synchrotron radiation for the detection of lung cancers was investigated. Lung cancer tissues grown in mice over different periods were observed with phase-contrast synchrotron X-ray chest imaging. Irregular and tortuous vessels appearing in early cancer angiogenesis were found in the periphery region of the implanted cancer. Structure difference was clearly shown between normal tissue and early stage solid cancer in micrometers. Results from this study indicate that synchrotron X-ray may open broad perspectives for imaging cancer development and progression noninvasively.

This work presents a new programming model for sensor networks embedded in the human body which is based on the concept of multi-programming application-oriented software. This model was conceived with a top-down approach of four layers and its main goal is to allow the healthcare professionals to program and to reconfigure the network locally or by the Internet. In order to evaluate this hypothesis, a benchmarking was executed in order to allow the assessment of the mean time spent in the programming of a multi-functional sensor node used for the measurement and transmission of the electrocardiogram.

Due to the possibility of detecting certain physiological conditions from thermal features of the skin surface acquired from infrared thermal imaging, the health conditions of a person can be revealed by analyzing the thermal signatures of his or her forearms regions in an infrared image. The assessment of hand's or arm's temperature distribution for clinical diagnosis or monitoring requires the confinement of region of interest (ROI) on the forearms regions. Hence, the purpose of this study is automatically to segment forearms regions in an infrared thermal image so that the clinicians can able to locate the interested regions and extract the skin temperature distributions with a high degree of reproducibility.

An integrated multi-method system to analyze the neuroelectric source parameters of electroencephalography (EEG) signal is presented. In order to handle the large-scale high dimension data efficiently and provide a real-time localizer in EEG inverse problem, an improved isometric mapping algorithm is used to find the low dimensional manifolds from high dimensional recorded EEG. Then, based on reduced dimension data, a single-scaling radial-basis wavelet network module is employed to determine the parameters of different type of EEG source models. In our simulation experiments, satisfactory results are obtained.

This paper presents our comparative study of the application of intensity based similarity measures to the problem of matching genomic structures in microscopic images of living cells. As part of our ongoing research, we present here for the first time evidence from experiments and simulations that show the benefit of using an iterative matching algorithm guided by an intensity based similarity measure. Our experimental results are compared against a gold standard and suggest the measures that work best in the presence of fluorescent decay and other problems inherent to time-lapse microscopy. This makes our approach widely applicable in the study of the dynamics of living cells with time-lapse microscopic imaging.

Parameters of heart rate turbulence (HRT) and blood pressure turbulence (BPT) reflect the baroreflex mediated transient acceleration-deceleration response of the sinus node triggered by a premature ventricular complex (PVC) and are known to be useful risk markers in different cardiac diseases. However, the analyses of HRT and BPT are based on the prevalence of PVC. In this study the spontaneous HRT and BPT after most premature normal beats (MPNB) was investigated in comparison to HRT and BPT after PVC. ECG and blood pressure were recorded from 91 patients with dilated cardiomyopathy (DCM) and 45 healthy subjects (REF). In 69% of DCM and 33% of REF data sets PVCs were present. HRT and BPT were determined for all patients with PVC as well as in all patients triggered by MPNB. Univariate statistical analysis of the comparison between DCM and REF revealed similar results for HRT/BPT after PVC and MPNB. In conclusion, HRT and BPT triggered by PVC might be substituted by spontaneous HRT and BPT after MPNB. In contrast to the HRT and BPT after PVC, an increased number of data sets can be included in the analysis and may lead to an enhancement in risk stratification in different cardiac diseases.

Synchronization of neuronal activity in the gamma-band has been shown to play an important role in higher cognitive functions, by grouping together the necessary information in different cortical areas to achieve a coherent perception. In the present work, we used a neural network of Wilson-Cowan oscillators to analyze the problem of binding and segmentation of high-level objects. Binding is achieved by implementing in the network the similarity and prior knowledge Gestalt rules. Similarity law is realized via topological maps within the network. Prior knowledge originates by means of a Hebbian rule of synaptic change; objects are memorized in the network with different strengths. Segmentation is realized via a global inhibitor which allows desynchronisation among multiple objects avoiding interference. Simulation results performed with a 40x40 neural grid, using three simultaneous input objects, show that the network is able to recognize and segment objects in several different conditions (different degrees of incompleteness or distortion of input patterns), exhibiting the higher reconstruction performances the higher the strength of object memory. The presented model represents an integrated approach for investigating the relationships among learning, memory, topological organization and gamma-band synchronization.

Building a system based on variants of disparate individual components/programs is usually a challenging task. The components/programs are not designed to communicate with each other but the whole system construction does require a seamless collaboration among them. In this paper, targeting at protein structure prediction, a pluggable application server framework is presented. The framework is capable of combining various existing programs into an efficient unit and the design is devoted to provide a model which is able to integrate heterogeneous components/programs into the system quickly without modifying their codes. Based on the model, different components can be plugged into the system with easy configuration, which would lead to a self-configurable and adaptive system. A protein structure prediction server implementation was developed by applying the design model and the real implementation emphasizes the efficiency and simplicity of the system construction. The method and model are generic and can be applied to other system design as well.

We propose a semantic tagger that provides high level concept information for phrases in clinical documents, which enriches medical information tracking system that support decision making or quality assurance of medical treatment. In this paper, we have tried to deal with patient records written by doctors rather than well-formed documents such as Medline abstracts. In addition, annotating clinical text on phrases semantically rather than syntactically has been attempted, which are at higher level granularity than words that have been the target for most tagging work.

Electrical impedance or admittance cardiography is a simple method for non-invasive, continuous measurement the stroke volume and cardiac output. For the electrical impedance cardiography, the band-electrodes array proposed by Kubicek et al has been widely used, and various spot-electrodes array have been experimented in search of a less uncomfortable and equally reliable electrodes array that is easier to attach. From the uniformity of current distribution on the thorax, we have reinvestigated focusing on the measurement of contour maps of static and pulsatile components of a regional area along the medial line on the frontal part of the thorax. Consequently, the appropriate electrodes locations for current injection were determined as the back of an ear and on the lower abdomen, while those for voltage pick-up was on the medial portion at the level of clavicle and on the portion above the xiphisternum. Preliminary comparison experiments between the cardiac output values obtained by the electrical impedance cardiography and by a pulse dye-densitometry showed a fairy good agreement.

Our goal is to simulate interactions with a computer model of the esophagus and the stomach, built from the Visible Human database, and reported in previous works. The optical distortion of the endoscope was simulated during navigation allowing the user a quantitative assessment of distortion when measures an injury. Another improvement to our model is the inclusion of abnormal anatomy, from two types, the first one is related to the color features of the disease and the second is a direct modification in the triangle structure of the mesh, with the goal to simulate blisters and injures in the model. The esophagus in its natural state presents radial collapsing, which was simulated using finite element methods. To this collapsed state we superposed the interaction of the triangle mesh with a model of the air pressure against the walls. The collapsed state allows to train the user in the insertion of the endoscope and to assess the effects of friction between the walls and the instrument.

This paper presents an analysis of the Arruda accessory pathway localization method (for patients suffering from Wolff-Parkinson-White syndrome) with suggestions to increase the overall performance. The Arruda method was tested on a total of 121 patients, and a 90% localization performance was reached. This was considered almost as performing result as the highest published (90%) by L. Boersma in 2002. After a deeper analysis of each decision point of Arruda localization method we considered that the lead AVF is not as relevant as other used leads (I, II, III, VI). The overall performance (90%) was slightly lower then the correct decision rate (91.67%) at the weakest decision element (AVF+) of the method. The vectorial space constructed from the most used leads (II, VI, AVF) is not orthogonal which can be a reason for weaker rate in case of AVF.

In this paper, we evaluate two segmentation methods on 15 brain tissue structures. One is narrow band level set method and the other is pattern classification method based on maximum a posteriori (MAP) probability framework. Two sets of experiments are conducted on 18 verified MRI brain data sets. Dice Similarity Index (DSI) is used to evaluate the closeness between our segmentation results and the gold standards, which were provided by experienced radiologists. The results for comparison of two methods are given and their potential applicability is discussed. Tissue structures such as left and right lateral ventricle have achieved over 70% DSI, while other structures such as third ventricle, caudate nucleus, globus pallidus, putamen and thalamus have achieved above 60% DSI.

An implantable neural recording front-end has been designed in two versions. The first is a multi-channel signal-conditioning ASIC for use with any neural recording probe technology. This ASIC was implemented in a commercial 0.5 mum CMOS process, includes 16 parallel amplifier channels, and measures 2.3 mm2 The amplifiers have a gain of 59.5 dB, a high cutoff frequency at 9.1 kHz and consume 75 microW per channel. The low cutoff frequency is independently tunable on each channel to accept or reject field potentials. This chip is small enough to be chronically packaged for experiments in awake behaving animals or it can be integrated into a fully implantable neural recording microsystem. The second version of the front-end is a neural recording probe with integrated signal conditioning circuitry on the back-end implemented in a 3 microm CMOS process. This version dissipates 142 microW and includes 64 to 8 site selection, 8 per-channel amplifiers each having a gain of 50.2 dB, a tunable low cutoff frequency, and a 7 kHz upper cutoff frequency. Real-time site impedance and circuit testing has been integrated in this design.

Using a noise covariance model based on a single Kronecker product of spatial and temporal covariance in the spatiotemporal analysis of MEG data was demonstrated to provide improvement in the results over that of the commonly used diagonal noise covariance model. In this paper we present a model that is a generalization of all of the above models. It describes models based on a single Kronecker product of spatial and temporal covariance as well as more complicated multi-pair models together with any intermediate form expressed as a sum of Kronecker products of spatial component matrices of reduced rank and their corresponding temporal covariance matrices. The model provides a framework for controlling the tradeoff between the described complexity of the background and computational demand for the analysis using this model. Ways to estimate the value of the parameter controlling this tradeoff are also discussed.

Studies of neural function that are carried out in different laboratories and that address different questions use a wide range of descriptors for data storage, depending on the laboratory and the individuals that input the data. A common approach to describe non-textual data that are referenced through a relational database is to use metadata descriptors. We have recently designed such a prototype system, but to maintain efficiency and a manageable metadata table, free formatted fields were designed as table entries. The database interface application utilizes an intelligent agent to improve integrity of operation. The purpose of this study was to investigate how reinforcement learning algorithms can assist the user in interacting with the database interface application that has been developed to improve the performance of the system.

Active Shape Model (ASM) is a popular statistical shape model for objects localization. It combines a Point Distribution Model (PDM) and local grey-level model of each landmark point to achieve reliable and accurate result in image interpretation. However, in 2D images, the one-dimensional profile grey-level models will lose information in other directions. In this paper, we modeled the grey-level information of the k x k square neighbor region around each point, hoping to take more information into account. Yet, further works should be done in order to compare the searching results obtained by square grey-level models with those by one-dimensional profile grey-level models.

Possible clues for sensory perception of an unexpected sudden change in floor level during human gait could be discrepancies in the time of heel contact, location of force application point (center of pressure, CoP) in the sole at the heel contact and subsequent stance period from the ones prescribed prior to the gait execution prepared for the normal level gait, if they exist. In order to get insight for this perception mechanism, a psychophysical experiment during gait on a walk way with a small, unexpected and sudden change in the floor level was performed, in which the subject was forced to answer among step-down, flat, and step-up immediately after the walk. During the gait, foot pressure distribution (FPD), electromyogram (EMG) of the ankle muscles, and gait trajectory (GT) were measured. The experiment could quantify accuracy of the perception. The floor level change dependent changes in FPD, CoP trajectory, EMG, GT were examined to suggest that these changes might be the clues for the perception.

Absolutely lossless compression technique has been recently proposed for medical image compression. It is a hybrid of lossless and lossy compression for medical images using difference image. It is a multi step process in which we have used lossy coding followed by difference image coding. The combination of two compression ratios is resultant. In this paper we analyze statistics associated with difference image. The main objective of this work is to exploit some statistical measures for a difference image to compress it losslessly. Difference image statistics plays a vital role in our technique to get maximum compression ratios. Proposed scheme is simple, computationally economical and can achieve higher compression ratios than existing standard lossless compression techniques and also meets the legal requirement of medical image archiving.

Monitoring of posture change in sagittal plane and walking speed is important for evaluate the effectiveness of rehabilitation program or brace. We have developed a wearable device for monitoring human activity. However, in the previous system, there still remain several drawbacks for practical use such as accuracy in angle measurement, cumbersome cable arrangements, and so on. In order to improve these practical drawbacks, a new sensor system was designed, and its availability was evaluated. The results demonstrated that the accuracy of this system showed superior to that of the previous, and this system appears to be a significant means for quantitative assessment of the patient's motion.

Statistical image analysis has emerged as a basic methodology in the study of many real phenomena, which can be represented by sequences of binary images. Recent techniques, such as total internal reflection fluorescent microscopy, allow us to image simultaneously two fluorescent-tagged proteins which are relevant in important cellular processes, such as endocytosis. Here, we model these biological pairs of image sequences as realizations of a 3D non-isotropic bivariate random set, in which one dimension corresponds to time. We analyze their second-order properties by means of the cross-covariance and the pair correlation function in order to study colocalization between proteins. Results show the proposed methodology allows us to study spatial dependencies in a formal and robust way.

We investigate the use of five dimension reduction and manifold learning techniques to estimate a 2D subspace of hand poses for the purpose of generating motion. Our aim is to uncover a 2D parameterization from optical motion capture data that allows for transformation sparse user input trajectories into desired hand movements. The use of shape descriptors for representing hand pose is additionally explored for dealing with occluded parts of the hand during data collection. We present early results from uncovering 2D parameterizations of power and precision grasps and their use to drive a physically simulated hand from 2D mouse input.

We propose a new similarity measure for medical image registration and make comparisons with two kinds of mutual information (MI) methods in the literature. This new measure is computed from the normal vector information (NVI) of the images instead of pixel intensity that the MI methods based. The NVI of an image is extracted from the relationship between pixels and computed from the normal vector of the pixel based on their local isosurface. The NVI method has been proved to be able to successfully register two-dimensional (2D) images. In this paper, we will apply it in three-dimensional (3D) medical images and employ the known-result datasets to quantitatively evaluate the performances of the NVI and MI methods. The visual assessment is employed for the unknown-result clinical image dataset registrations. The results show that the NVI method is ready to be affected by the salt-and-pepper noise; while when the random noise is removed, the NVI method performs no worse than the MI methods.

This paper suggests possibility for detecting recruitment of fast muscle fiber on increasing load tests by multiresolution analysis of surface electromyograms (sEMGs). Recruitment of the muscle fiber can be guessed from a graph of power ratio vs. % maximum voluntary contraction concerning the detail at each level, RPD(j). RPD(j)s are extracted parameters from multiresolution analysis of sEMGs. The sEMGs were recorded from the biceps brachii in five healthy males.

The objective of this study was to classify hysteroscopy images of the endometrium based on texture analysis for the early detection of gynaecological cancer. A total of 418 regions of interest (ROIs) were extracted (209 normal and 209 abnormal) from 40 subjects. Images were gamma corrected and were converted to gray scale. The following texture features were extracted: (i) statistical features, (ii) spatial gray level dependence matrices (SGLDM), and (iii) gray level difference statistics (GLDS). The PNN and SVM neural network classifiers were also investigated for classifying normal and abnormal ROIs. Results show that there is significant difference (using Wilcoxon rank sum test at a=0.05) between the texture features of normal and abnormal ROIs for both the gamma corrected and uncorrected images. Abnormal ROIs had lower gray scale median and homogeneity values, and higher entropy and contrast values when compared to the normal ROIs. The highest percentage of correct classifications score was 77% and was achieved for the SVM models trained with the SF and GLDS features. Concluding, texture features provide useful information differentiating between normal and abnormal ROIs of the endometrium.

Microfluidic channels were studied, in which the surface is modified from a solid/liquid interface into solid/liquid and air/liquid alternating interface, creating the equivalent of a superhydrophobic surface on the interior of the channel. The composite microchannel can be easily fabricated using embossing or cast molding of PDMS. The channels are stable under typical microfluidic conditions. For the most part, fluid flow behavior is not significantly changed; however, interesting mass transport effects can be observed in such channels under appropriate conditions. An application example of a microvalve based on the mass transport effect is demonstrated, showing advantages of simple design, fabrication, no moving part and zero dead volume.

In this paper, a new non-regularization method for positron emission tomography (PET) reconstruction is proposed. The proposed method is a feature-based method using Fourier-Wavelet basis. In order to obtain the reconstructions, we have to calculate the Fourier-Wavelet moment (FWM) from the measurements. To achieve this, iterative method is employed. The rotation invariance property of the proposed basis permits us to reduce computational cost. A row-action (RA) like fast convergent algorithm is used to further accelerate the convergence rate. In experiment, we compare the proposed method with some existing algorithms. The results show that our method offers good reconstruction quality compared to conventional MAP method.

In this paper, we propose a semi-automatic thalamus and thalamus nuclei segmentation algorithm from diffusion tensor magnetic resonance imaging (DT-MRI) based on the mean-shift algorithm. Comparing with existing thalamus segmentation algorithms which are mainly based on K-means algorithm, our mean-shift based algorithm is more flexible and adaptive. It does not assume a Gaussian distribution or a fixed number of clusters. Furthermore, the single parameter in the mean-shift based algorithm supports hierarchical clustering naturally.

The task of Electrocardiographic Imaging is an ill-posed inverse problem, requiring regularization. However, it has special features, firstly because it is a "non-stationary" inverse problem, and secondly because the inherent dynamical variety (e.g., epicardial breakthroughs, arrhythmias, ischemic changes) may preclude a fruitful nontrivial process model. Importantly, its structure places it in the category of "partial inverse problems" - a theory that arises from this setting. Surprising features of the resulting regularization methodology include the ability to fashion nontrivial regularization matrices in part (and sometimes entirely) from the data. There is evidence that these theoretical results can have significant practical benefits.

In this paper, we propose an architecture that extends the Object-Oriented Database (OODB) system architecture by adding domain specific additional layers to manage protein structure data. The two layers introduced above OODB are Protein-QL, domain-specific query language and Protein-OODB, a domain-specific data layer. This architecture is designed specifically for the protein domain, but it is the first step in building a general Bio-OODBMS for biological applications. Three internal data types are defined for the primary, secondary, and tertiary protein structures, respectively, to simplify queries in Protein-QL. This enables the domain scientists to easily formulate data requests. We use lambda-DB as the back-end database to implement Protein-QL. Queries in Protein-QL are compiled into OQL which are then executed against the database. In order to make the underlying OODB system (lambda-DB) more powerful, we introduce additional constraints to check the integrity of protein data.

Urine glucose level monitoring technique using near infrared spectroscopy in conjunction with the chemometric method was newly developed aiming for the use of home health care. The calibration models were obtained by the partial least square method and their validity were assessed using albumin added glucose solution and urine samples. From the results obtained, it was clearly demonstrated that the present method had a capability of predicting urine glucose level with reasonable accuracy (standard error of prediction; 22.3 mg/dl, correlation coefficient; 0.99) and appeared to be a useful means for long-term home health care.

This paper reports on a battery-powered telemetric command receiver for electronic site selection in multichannel neural recording applications. The receiver selects seven recording sites from a total of 28 available sites, according to four pre-defined site-selection patterns, via a 1-MHz, 5-Kb/s, amplitude-shift-keyed (ASK) link. The seven selected sites can then be wirelessly monitored via a 100-MHz frequency-modulated (FM) link. The receiver also performs power management to increase battery service lifetime. A bidirectional wireless recording microsystem incorporating this receiver is fabricated on a 4.6times4.6-mm(2) chip using the AMI 1.5 microm 2P2M n-well CMOS process. Design methodology and architecture of the receiver together with measurement results from its wireless analog front-end are presented.

In this study the evaluation of a Platelet-based Maximum Penalized Likelihood Estimation (MPLE) for denoising SPECT images was performed and compared with other denoising methods such as Wavelets or Butterworth filteration. Platelet-based MPLE factorization as a multiscale decomposition approach has been already proposed for better edges and surfaces representation due to Poisson noise and inherent smoothness of this kind of images. We applied this approach on both simulated and real SPECT images. For NEMA phantom images, the measured noise levels before (M(b)) and after (M(a)) denoising with Platelet-based MPLE approach were M(b)=2.1732, M(a)=0.1399. In patient study for 32 cardiac SPECT images, the difference between noise level and SNR before and after the approach were (M(b)=3.7607, SNR(b)=9.7762, M(a)=0.7374, SNR(a)=41.0848) respectively. Thus the Coefficient Variance (C.V) of SNR values for denoised images with this algorithm as compared with Butterworth filter, (145/33%) was found. For 32 brain SPECT images the Coefficient Variance of SNR values, (196/17%) was obtained. Our results shows that Platelet-based MPLE is a useful method for denoising SPECT images considering better homogenous image, improvements in SNR, better radioactive uptake in target organ and reduction of interfering activity from background radiation to compare to that of other conventional denoising methods.

The goal of our research is to develop a high-level controller to provide reference trajectories automatically to the low-level controller of a rehabilitation robotic device. The high-level controller, which is the supervisory controller, is an event driven, asynchronous discrete event system (DES), described by a finite state automaton. Extensive simulations are performed using the supervisory controller for a rehabilitation task. The results demonstrate the feasibility of the proposed supervisory controller.

This paper implements spectral analysis of scalp EEG recordings during a language naming and visualisation task. The method offers new frontier to explore spatio-temporal features of the organisation of conceptual knowledge in the intact brain. Our findings tallies with results reported in the literature using other techniques such as fMRI. The method introduced in this paper provides new perspective for understanding and possibly diagnosing category specific semantic deficits.

We propose a new approach for the protein tertiary structure prediction based on the concept of mini-threading. The method identifies useful fragments in Protein Data Bank (PDB) with variable lengths and retrieves spatial restraints. The multidimensional scaling method and least-squares minimization are used to build coarse-grain structural models. Our method uses the information in the PDB efficiently and the prediction time is in minutes when compared to hours and days required by existing methods.

Long hours of automobile driving under monotonous situations may cause the lowering of what we term a driver's activation state (DAS) or in other words the production of drowsiness, resulting in an increased risk of a traffic accident. There is therefore a need to create a newly advanced system focused on the DAS in-car, hopefully thus avoiding potentially dangerous situations. In order to develop such a system as a final goal, we have firstly set out to acquire such cardiovascular variables as beat-by-beat blood pressure (BP), RR interval from ECG and normalized pulse volume (NPV) used as a peripheral vascular tone of alpha-adrenergic sympathetic activity, during presentation to the driver of a screen movie simulating monotonous travel at constant speed on a test-course. Subsequently, we have investigated the reactivity in terms of the driver's cardiovascular hemodynamics. Through the successful monitoring of cardiovascular parameters during the movie presentation obtained in 11 healthy male subjects, the following results were obtained: The monotonous driving produces a statistically significant gradual rise in BP following drowsiness, which could be explained by enhancement of sympathetic activity using a time-frequency analysis of BP and RR. This finding strongly indicates that continuous driving in such monotonous situations can make a driver considerably stressful and thus may cause a gradual increase in BP, and that this gradual BP increase may be used as a possible index relevant to the DAS. This finding was also confirmed by the analysis of NPV, suggesting that the gradual increase in BP during the monotonous driving would be rather caused by a regulation of peripheral vasomotor constriction.

In the context of the growing proportion of seniors in the western world population and the efforts provided in home care services, we have developed a computer vision system for monitoring medication intake. The system detects automatically medication intake using a single low-cost webcam. Person detection and tracking over the video sequence is done using color-based techniques while the recognition of the medication intake activity is performed using our main contribution, a three-level scenario model. Experimental results in controlled conditions are shown and we discuss improvements to our system.

This article describes the design and development process of an eye tracking-based computer system that benefits from the intact ocular motor control of the completely locked-in patient to provide him or her with an alternative means of communication. A completely locked-in patient is an individual who has lost all types of motor control and communication ability with people in his or her environment. This developed system uses a head-mounted web camera to capture real-time images of the patient's eye. These images are then passed to a program, developed using Matlab, which processes them and computes the coordinates of the pupil position. The program then sends commands to an interactive JAVA-based interface, which provides the patient with a matrix of pictograms representing the most essential daily communication activities. When a pictogram is activated (clicked), the system plays back an audible statement, recorded in any language, reflecting the desired activity. Ten healthy adult volunteers, free from any musculoskeletal or neurological disorders, participated in the validation of the system. Validation results revealed a system accuracy of 96.11+/-5.58 % and repeatability of 94.44+/-2.51 %. The rehabilitative system developed in this project offers the locked-in patient, of any social class, the ability of simple yet effective communication. The advantages of this system over existing systems are low cost, low processing power, ease of operation, little training requirements, minimal disturbance to the patient, and ease of customization to any mother tongue.

We have developed an instrumented pillbox, called a MedTracker, which allows monitoring of medication adherence on a continuous basis. This device improves on existing systems by providing mobility, frequent and automatic data collection, more detailed information about non-adherence and medication errors, and the familiar interface of a 7-day drug store pillbox. We report on the design of the MedTracker, and on the results of a field trial in 39 homes to evaluate the device.

Faced with the growing population of seniors, Western societies need to think about new technologies to ensure the safety of elderly people at home. Computer vision provides a good solution for healthcare systems because it allows a specific analysis of people behavior. Moreover, a system based on video surveillance is particularly well adapted to detect falls. We present a new method to detect falls using a single camera. Our approach is based on the 3D trajectory of the head, which allows us to distinguish falls from normal activities using 3D velocities.

A thin-film cochlear electrode array has been developed for a cochlear prosthesis to achieve improved sound perception and position accuracy. The array is fabricated using a bulk-silicon micromachining process that allows parylene deposition and patterning at wafer level, followed by a wet silicon release etch that is compatible with the use of boron etch-stops. The process is capable of realizing arrays with substrates stressed to hug the modiolar wall in the rest state and whose stiffness can be adjusted over a wide range. Built-in tip and curvature sensors respond to tip contact and bending-induced shank stress, respectively during in-vitro and in-vivo implants. The process is also compatible with the integration of parylene ribbon cables for lead transfer to an implanted electronics package.

In this paper, we present a comprehensive neural network based modeling and validation framework for reverse engineering gene regulatory interactions. We employ two approaches, Gene Set Stochastic Sampling and Sensitivity Analysis, to infer these interactions. We first apply these methods to a simulated artificial dataset to ensure their correctness and accuracy. True biological interactions are then modeled by analyzing a rat hippocampus development dataset. Finally, we present a thorough computational methodology to test the validity and robustness of the inferred regulations through novel assemblies of relevant testing datasets.

In this study, we discuss the use of support vector machine (SVM) learning to classify heart rate signals. Each signal is represented by an attribute vector containing a set of statistical measures for the respective signal. At first, the SVM classifier is trained by data (attribute vectors) with known ground truth. Then, the classifier learnt parameters can be used for the categorization of new signals not belonging to the training set. We have experimented with both real and artificial signals and the SVM classifier performs very well even with signals exhibiting very low signal to noise ratio which is not the case for other standard methods proposed by the literature.

When a brief current pulse is incident on excitable cells in cardiac and other nervous tissue, a change in phase of the cell's response is usually observed. In cardiac tissue, the cells are bared to external stimulation of generally positive currents, which depolarize the cells. In this paper an overview of the application of the phase resetting technique (PRT) in several cardiac models is presented. We discuss the effects of external stimuli in several cardiac cell models and we provide the phase transition curves (PTCs) resulted from the application of PRT with the Zhang et al. sinoatrial node model.

This paper reports the design, fabrication and simulation of a novel array of micromachined electrodes with different lengths for use in auditory nerve prostheses. A 10 x 10 array is created in a 1 mm2 on silicon with bulk micromachining technology. The electrode lengths vary from 200 microm to 400 microm. This design could provide access to most fascicles within the auditory nerve and thus allow stimulation of a broad tonotopic range within the nerve fiber. A multichannel ultra-low power circuit is also designed, fabricated, and tested for neural signal recording. The chip is fabricated in 0.5 microm AMI CMOS technology with a die size of 1.5 mm x 1.5 mm. The total power consumption is less than 100 microW.