Search

We are developing a prototype monolithic scintillation camera with optical sensors on the entrance surface (SES) for use with statistically-estimated depth-of-interaction in a continuous scintillator. We opt to use Geiger-Müller mode avalanche photodiodes (GM-APDs) for the SES camera since they possess many desirable properties; for the intended application (SES and PET/MR imaging), they offer a thin attenuation profile and an operational insensitivity to large magnetic fields. However, one issue that must be addressed in using GM-APDs in an RF environment (as in MR scanners) is the thermal dissipation that can occur in this semiconductor material.Signals of GM-APDs are strongly dependent on junction temperature. Consequently, we are developing a temperature-controlled GM-APD-based PET camera whose monitored temperature can be used to dynamically account for the temperature dependence of the output signals. Presently, we aim to characterize the output-signal dependence on temperature and bias for a GM-APD-based scintillation camera.We've examined two GM-APDs, a Zecotek prototype MAPD-3N, and a SensL commercial SPMArray2. The dominant effect of temperature on gain that we observe results from a linear dependence of breakdown voltage on temperature (0.071 V/°C and 0.024 V/°C, respectively); at 2.3 V excess bias (voltage above breakdown) the resulting change in gain with temperature (without adjusting bias voltage) is -8.5% per °C for the MAPD-3N and -1.5 % per °C for the SPMArray2. For fixed excess bias, change in dark current with temperature varied widely, decreasing by 25% to 40% as temperature was changed from 20 °C to 10 °C and again by 20% to 35% going from 10 °C to 0 °C. Finally, using two MAPD-3N to read out a pair of 3.5-by-3.5-by-20 mm(3) Zecotek LFS-3 scintillators in coincidence, we observe a decrease from 1.7 nsec to 1.5 nsec in coincidence-time resolution as we lowered temperature from 23 °C to 10 °C.

MUNE methods estimate the typical motor unit (MU). The typical MU amplitude is divided into the compound muscle action potential (CMAP) amplitude to obtain the MUNE. A novel method is proposed in which the CMAP is modeled as a linear least squares (LLS) fit of the MU waveforms. The sum of the coefficients in the model is the motor unit number estimate. Each coefficient represents the count of the different MU estimates. The MU waveforms are clustered into groups with similar morphology and a representative waveform is calculated for each cluster. These representative waveforms are used in the LLS model.

A bond graph model of entire cardiovascular system is developed in this paper. Although relying on the concepts of previously developed lumped-parameter models, this new model benefits from some advantages of general system theory which allows the modeler to include as many details as necessary. Specifically, inclusion of pace making mechanism has become much easier. The pace making mechanism is not, however, included in this stage of model development. Modular sources of effor are, instead, predicted in the model to genrate synthetic signals close to what occurs in reality. The model is simulated and results of simulation are compared with widely available data of normal cardiovascular system's pressure-flow performance. It has been shown that they are of a good agreement and the model could successfully simuulate performance of the human cardiovascular system.

This paper describes our experiences in the simulation, implementation and application of a flat panel detector based cone beam computed tomography (CT) imaging system for dedicated 3-D breast imaging. In our simulation study, the breast was analytically modeled as a cylinder of breast tissue loosely molded into cylindrical shape with embedded soft tissue masses and calcifications. Attenuation coefficients for various types of breast tissue, soft tissue masses and calcifications were estimated for various kVp's to generate simulated image signals. Projection images were computed to incorporate x-ray attenuation, geometric magnification, x-ray detection, detector blurring, image pixelization and digitization. Based on the x-ray kVp/filtration used, transmittance through the phantom, detective quantum efficiency (DQE), exposure level, and imaging geometry, the photon fluence was estimated and used to compute the quantum noise level on a pixel-by-pixel basis for various dose levels at the isocenter. This estimated noise level was then used with a random number generator to generate and add a fluctuation component to the noiseless transmitted image signal. The noise carrying projection images were then convolved with a Gaussian-like kernel, computed from measured 1-D line spread function (LSF) to simulate detector blurring. Additional 2-D Gaussian filtering was applied to the projection images and tested for improving the detection of soft tissue masses and calcifications in the reconstructed images. Reconstruction was performed using the Feldkamp filtered backprojection algorithm. All simulations were performed on a 24 PC (2.4 GHz Dual-Xeon CPU) cluster with MPI parallel programming.

We have introduced well-defined nanopillar arrays of polyethylene glycol (PEG) as a platform for studying the adhesion and growth of cultured cardiomyocytes. The nanopillar arrays were fabricated by using a simple molding technique involving the placement of a patterned polyurethane acrylate mold on top of a drop-dispensed ultraviolet (UV) curable PEG polymer followed by UV exposure and mold removal. The adhesion and growth of cardiomyocytes turned out be guided by an external nanotopography, which has been characterized in terms of cell morphology and cytoskeletal arrangement. In particular, the nanopillars provided guiding posts to both elongating filopodia and expanding lamellipodia. Interestingly, the 3D growth of cardiomyocytes was mediated by the increased hydrophobicity of the nanostructured PEG substrate, indicating that the cell adhesion and growth is very sensitive to the nanotopography. The precise nanostructures of PEG-based polymer with controlled geometrical features presented in this study not only open opportunities for understanding and tailoring cell adhesion and growth, but could serve as a template for better tissue engineering by controlling cellular activities at the molecular level.

Prevention of pressure ulcer (PU) has been hampered for a long time by limitations of present risk assessment methodologies for carers. This paper introduces a novel solution of a portable bioimpedance spectrometer (BIS), which is designed to detect in vivo subtle changes in electrical properties of tissues at the portentous onset of PU and thereby establish a reliable assessment method for early detection of PU. The principle of the modified tetrapolar method applied in BIS and system designing methodology are explained thoroughly in this paper. The BIS system can measure complex impedance of human tissues over a wide range of frequencies from 10Hz to 1MHz based on direct digital synthesis (DDS) technique. Different from conventional bioimpedance measuring systems based on phase sensitive detection, the new BIS system shows its integrity in theoretical analysis and simplicity in practical implementation. As an objective, fast, straightforward, and inexpensive technique that captures fundamentally different information than subjective assessment and optical judgment, the new BIS may bring to clinical use for risk assessment of PU in the future.

Performance analysis is important for evaluating the system effectivity. In this paper, the queueing theory and discrete event simulation technology are applied to a performance study of a home ECG telemonitoring system. The data from a four-month trial is used to establish the input stream model of the medical service system. The input model includes the customer inter-arrival time distribution and the service time distribution of the server. With the input model, a queueing model is introduced to simulate the ECG monitoring system. The performance simulation results indicate the feasibility of the model. The results are helpful on designing an optimal service system with given properties of input stream and service procedures.

To date, most interests in simulation studies with whole heart models are in body surface or epicardial potentials in order to compare the results with measurements in experimental and clinical practices. The focus of this study is paid on intracardial potentials as they are comparable with measurements during the invasive Electrophysiological Study (EPS) and the catheter ablation, which are becoming a dominant means in the therapy of arrhythmias. In this paper, simulations are implemented based on anisotropic computer heart models and a simulation system previously developed. The heart model consists of about 50,000 discrete cell elements with 1.5 mm spatial resolution. An algorithm in computing intracardial potentials inside the four cavities of the heart is developed. The HRA (High Right Atrium), HIS (His Bundle), RVA (Right Ventricle Apex), and CS (Coronary Sinus) potentials are shown and compared between a normal heart model and a heart model with the Wolff-Parkinson-White (WPW) Syndrome. By employing the intracardial potential parameters and morphology, the position of the accessory pathway between the atria and ventricula in the WPW computer heart model can be qualitatively located. It is concluded that the simulated intracardial potentials based on an anisotropic whole heart model can well approximate the realistic intracardial potentials.

Neural scientists have long been eager to find the subcortical pathways between different parts of cortex in human brain. By far, Diffusion Weighted-Magnetic Resonance Imaging (DW-MRI) is the only noninvasive technique to achieve this task in vivo. It applies a bipolar gradient to reveal diffusion of water molecules. With the assumption that the directions of neural fibers parallel to the directions in which water molecules diffuse, we can use the DW-MRI to track the neural fibers. Various signal processing techniques have been developed in extracting the directional information via DW-MRI. In this study, four techniques that are currently being used but have not been rigorously compared for performance are evaluated via custom-designed synthetic data and phantoms. By combining different techniques in accordance with their specialties, a better approach to track finer subcortical pathways in vivo can be derived and used to verify theories in human brain mapping.

The left-ventricle ejection fraction is an important cardiac-efficiency measure that is regularly used in cardiology. Standard estimations are based on time-consuming geometrical analysis and modelling, which requires experienced cardiologists. Alternative methods are very invasive due to the need for cardiac catheterization. In this paper we present and study a minimally-invasive indicator dilution technique for ejection fraction quantification that has recently been developed. It is based on a peripheral injection of an ultrasound contrast agent bolus. Left-atrium and left-ventricle acoustic intensities are recorded versus time by transthoracic echocardiography during contrast bolus passage. The measured curves are corrected for attenuation distortion, filtered to suppress the measurement noise, and processed by an adaptive Wiener deconvolution algorithm for the estimation of the left-ventricle impulse response. The estimated impulse response is interpolated by a mono-compartment exponential model for the ejection fraction assessment. An adaptive search of the interval for the model fitting is also included. The feasibility of the method is tested on 52 measurements in patients with left-ventricle ejection fractions between 10% and 80%. The results are promising and show a 0.83 correlation coefficient with echographic biplane ejection fraction measurements.

To improve the equality of life, we must prevent the falls in both healthy elderly and patients with the cerebrovascular diseases. Wearable accelerometer was applied to monitor. In this paper, we introduced two different clinical applications. On is fall detector and the other is monitoring device for screening test. 1) We have developed body-worn accelerometer with data loggers and monitored the daily of life in patient with Parkinson disease. The patients wore the device and monitored falls while walking and standing. As a result, we could obtain fall times for a long period. 2) The ability of walking and standing have been evaluated by Timed up & go test. We used telemetry with accelerometer. The stability of walking could be evaluated by the acceleration signals. The simple body-won device can be useful for fall study.

Nevoscope is an optical imaging device to obtain images of skin-lesions through back-scattered diffused light using trans-illumination to determine abnormalities such as skin-cancers. Due to the nature of the process of transillumination, images obtained from Nevoscope suffer from a dark-field effect. We proposed a two-step algorithm to restore Nevoscope images under the trans-illumination mode. First, a background attenuation matrix was extracted from an image of normal skin close to the suspicious abnormality. This matrix was then applied to the trans-illuminated image to correct background. Second, the background corrected image was decomposed using wavelet transform. By adjusting the wavelet coefficients judiciously, a denoised and feature enhanced image was obtained.

We describe the model and implementation of the hierarchical hybrid method for controlling of the lower-arm (pronation/supination and elbow flexion/extension) in humans with disabilities. The control follows the strategy found in ablebodied humans where the movement is planned based on the task and the most distal part of the arm; yet, the command starts from the most proximal segment. The controller uses a black box of the movement and relies on temporal and spatial synergies. The driving signals are the shoulder flexion/extension velocity and acceleration, the outputs are four stimulation patterns for the control of elbow flexion/extension and pronation/supination. The operation is discrete at the voluntary and coordination levels, and continuous at the actuator level. The repertoire of movement that were considered was limited to a set of typical daily activities within the normal workspace in the sitting position only. The main application of this control is the therapeutic electrical stimulation in post-stroke hemiplegic patients.

We investigated the source localization performance of the Laplacian weighted minimum norm (LWMN) estimate technique in a realistic geometry (RG) head model in the present study. We simulated current sources at different brain regions with various noise levels. The present results show there is no obvious depth dependency on the three-dimensional (3D) source estimation. The average source localization error over all simulated cases is about 10 mm. The tangential sources exhibit larger localization errors than the radial sources when they are close to the epicortical surface. The localization error will increase when the noise level increases. The LWMN technique was applied to source imaging of motor potentials induced by finger movement in a human subject. Both activities in the motor and premotor cortex, which are related to the execution and coordinating of the finger movement, were reconstructed by the LWMN technique. The present study suggests that LWMN has great ability in 3D sources imaging.

Hospitals and manufacturers are designing and deploying the IEEE 802.x wireless technologies in medical devices to promote patient mobility and flexible facility use. There is little information, however, on the reliability or ultimate safety of connecting multiple wireless life-critical medical devices from multiple vendors using commercial 802.11a, 802.11b, 802.11g or pre-802.11n devices. It is believed that 802.11-type devices can introduce unintended life-threatening risks unless delivery of critical patient alarms to central monitoring systems and/or clinical personnel is assured by proper use of 802.11e Quality of Service (QoS) methods. Petri net tools can be used to simulate all possible states and transitions between devices and/or systems in a wireless device network, and can identify failure modes in advance. Colored Petri Net (CPN) tools are ideal, in fact, as they allow tracking and controlling each message in a network based on pre-selected criteria. This paper describes a research project using CPN to simulate and validate alarm integrity in a small multi-modality wireless patient monitoring system. A 20-monitor wireless patient monitoring network is created in two versions: one with non-prioritized 802.x CSM protocols and the second with simulated Quality of Service (QoS) capabilities similar to 802.11e (i.e., the second network allows message priority management.) In the standard 802.x network, dangerous heart arrhythmia and pulse oximetry alarms could not be reliably and rapidly communicated, but the second network's QoS priority management reduced that risk significantly.

Advanced array processing techniques are becoming an indispensable requirement for integrating the rapid developments in wireless high-density electronic interfaces to the central nervous system with computational neuroscience. This work aims at describing a systems approach for latency reduction in telemetry-linked brain machine interfaces to enable real-time transmission of high volumes of neural data. We show that the tradeoff between transmission bit rate and processing complexity requires a smart processing mechanism to strip the redundancy and extract the useful information early in the data stream. The results presented demonstrate that space-time processing offers tremendous savings in communication costs compared to on-chip spike detection followed by off-chip classification. They also demonstrate that the performance asymptotically approaches that of on-chip spike detection and sorting. Detailed performance evaluation is described.

Image latencies occur when a virtual environment (VE) responds to hand movements of its operator. A study has been conducted to investigate the effects of, and interactions among, hand movement related image latencies, target distance, and target width on discrete manual control tasks in a VE. In the presence of latencies, completion times of hand movement tasks have been found to obey Fitts' law. Significant interactions were found between the effect of latency and target width, but not between the effect of latency and target distance. Results suggest that, in the presence of hand movement-related image latencies, the effects of target width and target distance should not be analyzed as a combined single effect of ;Index-of-Difficulty'.

Image segmentation is a fundamental image process technique and plays an essential role in ultrasound image analysis. In this article, we propose an algorithm for image segmentation which is based on the random neural network (RNN) and features extracted by a bank of Gabor filters. With the scientists' work, it is revealed that Gabor functions act as some functions of human vision. And the RNN model proposed by Gelenbe is closer to biophysical reality and mathematically more tractable, in which signals in the form of impulses are transmitted with a certain probability. The segmentation algorithm based on these two techniques provide a good distinguish and classification capability for textures in the image. Furthermore, a strategy which is named as quartered segmentation strategy is also presented here in order to reduce the computation and speed up our approach. The presented algorithm is tested on an image produced by using Brodatz album and an ultrasound image, and the results are promising.

Analysis of electroencephalogram (EEG) requires a framework that facilitates handling the uncertainties associated with the varying brain dynamics and the presence of noise. Recently, the type-2 fuzzy logic systems (T2 FLSs) have been found effective in modeling uncertain data. This paper examines the potential of the T2 FLS methodology in devising an EEG-based brain-computer interface (BCI). In particular, a T2 FLS has been designed to classify imaginary left and right hand movements based on time-frequency information extracted from the EEG with the short time Fourier transform (STFT). Robustness of the method has also been verified in the presence of additive noise. The performance of the classifier is quantified with the classification accuracy (CA). The T2 fuzzy classifier has been proven to outperform its type-1 (T1) counterpart on all data sets recorded from three subjects examined. It has also compared favorably to the well known classifier based on linear discriminant analysis (LDA).

The contrast in dielectric properties between normal breast tissues and malignant tumors is significant. Based on this fact, detection of breast cancer with microwaves has been proposed. Here, a cross-Vivaldi antenna capable of measuring cross-polarization is investigated for tumor detection. Observation of cross-polarized reflections reduces reflections from surfaces such as skin, permitting more reliable detection of small tumors. Simulation results indicate that the antenna has the potential to detect breast tumors using this approach.

The aim of this study is to investigate the upper limit of protection afforded by various protective equipment and techniques that used in combination and used individually. A multi-element, nonlinear mathematical model, which was previously used to produce cardiovascular response to +Gz stress and protection afforded by anti-G suits, was modified by incorporating pulmonary-respiration model, as well as detailed vessel models of abdomen and lower extremities into previous model. Using the model, hemodynamic responses of cardiovascular system to +Gz and to different combination of protection measures were simulated. Simulation results indicate that when appropriate counter-pressure is exerted to lower body, the combination of extended coverage anti-G suit with appropriate positive pressure breathing could provide effortless protection around +9Gz. For higher no-staining protection to +12 Gz, tilt-back seat with angles greater than 55 should be used. It also suggests that for anti-G techniques that enhance +Gz tolerance by means of elevating intrathoracic pressure or increasing blood volume, such as ECGS, PBG, and AGSM, their combined protection may be expressed as an additive sum of their individual effects. However, the combined protection of tilt-back seat with other anti-G techniques is higher than the simple addition.

Traditional voice activity detection algorithms are mostly threshold-based or statistical model-based. All those methods are absent of the ability to react quickly to variations of environments. This paper describes an incremental SVM (Support Vector Machine) method for speech activity detection. The proposed incremental procedure makes it adaptive to variation of environments and the special construction of incremental training data set decreases computing consumption effectively. Experiments results demonstrated its higher end point detection accuracy. Further work will be focused on decreasing computing consumption and importing multi-class SVM classifiers.

Protein sequence alignments reveal the evolutionary information between homologous sequences. Traditional sequence alignment methods only use sequence information and the structure information from template is ignored. Recently, Kleinjung et al. developed a contact-based sequence alignment method that used the structural information from side-chain contacts. Alignment scores are provided by the CAO (Contact Accepted mutatiOn) substitution matrices. Kleinjung et al. devised an approximate dynamic programming algorithm for protein sequence alignment, on the assumption that the distance between the contacting residues during evolution has been conserved. However, such assumption is not suitable for insertion/deletion events during evolution. In this paper, the contact-based simulated annealing alignment method has been proposed, which can find the optimal alignment solution between two protein sequences without any restriction. The alignment score is calculated by the sequence-based scores and the weighted contact-based scores. A new parameter, the contact-penalty r, has been introduced. When the contacting residue in the template aligns with gap in the query sequence, the total alignment score is decreased by a contact-penalty. All the parameters including relative weight w of CAO scores versus Blosum62 scores, matrix constant c for CAO scores, gap-open penalty p, gap-extension penalty q and contact-penalty r are re-optimized by genetic algorithm. Testing on the Homstrad database shows that the accuracy of this method is 85.4%, which is higher than that of Kleinjung's method by about 3.6 percent. Such method can be useful in many biological problems such as protein remote homology detection, comparative modeling and fold recognition.

Wireless Medical Device Network (WMDN) deployment is occurring to facilitate ambulatory patient care, increase safer and more intelligent diagnostic and therapeutic capabilities, and improve flexible patient bed configuration that matches census requirements. Patient safety risks exist from delayed or lost WMDN alarm and data streams, however, and non-proprietary Verification and Validation (V2) techniques do not exist. Single-vendor and heterogeneous multi-vendor deployment must be considered, as well as safe and reliable coexistence with other IS technologies sharing similar network components. V2 of even a homogeneous single vendor, single device WMDNS is very complex for several reasons including: absence of industry standards or regulations, unconstrained mobility of patients and devices, and rapid changes in the underlying wireless network modalities. This project will evaluate and recommend appropriate best V2 practices from fields like software and systems engineering to improve a hospitals ability to properly implement and manage the emerging WMDN opportunities and to prevent patient injuries or other serious problems.

Wearable sensing networks offer a multitude of benefits for the user. One field that can benefit from such technologies is in-home robotics. Much work has been dedicated to the general area of interaction between robots and humans, and more specifically to gesture recognition. We propose a wearable monitoring network composed of conductive fabrics that can more easily facilitate robot interaction. It creates additional interaction by visual indication, and electronically by way of a local PC. This system creates a more natural human-robot interface and makes feasible many new forms of gesture recognition. Finally, we can address safety concerns; the garment gives us a method of locating the human and minimizing the possibility for robots to strike the user. In this paper, we lay out the architecture for such a system, and perform some of the initial characterization.

the electrical impedance and the insertion force were monitored during the implantation of the microwire electrodes. The data reveals distinct phases in the force and impedance profiles corresponding to the different mediums the electrodes pass through. The impedance is shown to be linearly related to the pressure exerted by the electrode on the dura (p < 0.05, R 2 >0.8). While both insertion force and the impedance could be used to determine when the electrode touches and breaks through the brain surface, the later, especially the phase angle of the impedance, is shown to be more consistent and sensitive. It is proposed that the electrode impedance be used to provide more objective and accurate estimation of 1) when the electrode touches the brain; 2) the load distribution among the wires of the microwire array, and 3) the mechanical and immunological interactions between the electrode and the tissue in-vivo. The method can potentially improve the accuracy of electrode placement, and enable the development of better surgical procedures and apparatus. The principle could also be applied to other situations, such as endoscopic and minimum invasive surgeries.

High-level upper-limb amputations result in prostheses with many degrees-of-freedom to be controlled, with very few control signals. A novel method for the control of myoelectric upper limb prostheses has been developed. By transferring the residual nerves to spare muscles in or near the residual limb, additional myoelectric control signals are created that allow the simultaneous control of multiple degrees-of-freedom in the prostheses. Since the nerve function correlates physiologically to the prosthetic function, operation is more natural and thus easier than current control paradigms. This surgical procedure and subsequent prosthetic fitting have been successfully completed on one shoulder disarticulation and one transhumeral amputee. The shoulder disarticulation amputee has also been fit with a unique 6 motor system, allowing him to control 3 motors (hand, elbow, and humeral rotator) with the use of 6 myoelectric signals; there was marked increase in functional range-of-motion. He was able to control multiple joints simultaneously and could perform tasks that he could not do before, including reaching out to pick up objects.

This paper describes a low-profile three-dimensional silicon/parylene microelectrode array as basis for practical neural prostheses for use in the central nervous system. The circuit areas of the silicon probes, containing mixed-signal CMOS circuitry for neural stimulation/recording, can be folded over to reduce the overall height of the microassembled array above the cortical surface. The low- profile structure is implemented using multiple gold beams spaced by orthogonal silicon braces. An integrated silicon/parylene batch process is introduced to encapsulate these interconnects and achieve high yield.

This paper describes an approach to fine scale non-rigid registration for mapping patterns of tissue volume loss in serial MRI studies of the brain. Specifically it addresses the important confound of diffuse tissue contrast changes which can influence local sub-voxel estimates of volume change. Such changes can be induced by neurodegenerative or neurodevelopmental processes, which not only modify apparent tissue volume, but also modify tissue integrity and its resulting MRI contrast parameters. We derive an approach to the voxel-wise maximization of regional mutual information (RMI) and use this to drive a viscous fluid deformation model between images. This provides a topology preserving map of local changes in volume between time points that is robust to regional changes in tissue contrast. Comparisons with current methodology are included showing that the approach provides a significant reduction in errors when tissue contrast varies locally between MRI acquisitions.

In this paper, it is shown that the main source of mechanical energy of cardiovascular (CV) system i.e., rhythmic heart contraction is transformed to the oscillations of the CV walls and blood flow, and finally CV acoustical waves. These waves propagate through both blood flow (hemodynamical pathways) and tissues (viscoelastical pathways) toward the skin. Nonetheless, the CV walls could be assumed as the source of acoustical waves, since they act as the interface between blood flows and other tissues including skin. After obtaining the approximate accelerations of CV walls from pressure-flow (PF) models, we also needed to model the viscoelastical pathways until the skin. Some improvements on PF models were fulfilled to present small variations of blood pressure such as dicrotic notch. The turbulence occurrence was also noticed to and conceptually modeled. The total homomorphic model could conceptually show the relations of CV sounds with CV characterizations and tissue specifications. Thus, it could be helpful to assess CV system in order to diagnose CV diseases via CV sounds. The CV sounds recorded from the skin of any place (e.g., chest or arm) could be simulated via this model, if the hemodynamical and viscoelastical parameters especially for the region under that place are obtained.

Biofeedback and a variety of brain-computer-interface methods imply as very first stages the obtainment of control of selected signals e.g. the related generating psycho-physiological processes. The basic mechanism in the learning phase is the operant conditioning, which represent a complex behavioral method consisting of several components. One of the most important components is the setting and adjustment of thresholds for the triggering of corresponding rewarding options. An adaptive threshold optimization method, for the training based on average values is presented. The procedure is derivated from the sequential test from Wald. The application of the sequential tests in the learning/training process allows a threshold adaptation corresponding to the abilities of the particular person and to the learning success.

In order to test the human color vision objectively, we have developed the technique using the sweep parameter VEP technique (Norcia and Tyler, 1985) with an iso-luminant chromatic grating is effective to detect the defective color vision (Momose and Saito, EMBC2002). In this study, the wavelet transform using complex Morlet wavelet was examined to detect the dynamical spectrum of VEP in color vision defectives. Vertical sinusoidal iso-luminant chromatic gratings (6 Hz, 2 c/deg) presented on a color monitor were used as stimuli. The two colors for the gratings were selected from the colors on the dichromatic iso-chromatic lines. Steady-state VEPs were recorded during a continuous decrease of chromatic contrast at 10%/s. VEP components at 12 Hz and 16 Hz were evaluated as the response signal and background EEG (noise), respectively. For deuteranopic iso-chromatic colors, the signal to noise ratio (SNR) of deuteranope and deteranomalia were significantly lower than those of normals (P<0.01), indicating that this technique is effective for objectively detecting of the color vision defectives.

Diagnosis, treatment planning and prognosis may be improved by combining functional data with anatomic location via PET/CT dual-modality image registration. The aim of this work is to reveal the spatial accuracy, reliability and speed of registration techniques used in GeminiTM PET/CT Imaging System, as a part of quality assurance program for clinical and research applications. A multi-layer alignment bearing source device was used to validate PET/CT intrinsic alignment, as well as a Rando a Man phantom. A series of known transformations were performed to simulate different misalignment between PET and CT images. Two semi-automated registration techniques used to correct misalignment were assessed and compared quantitatively by measuring absolute distances between the centroids of corresponding fiducial markers in the registered volumes. Comparisons between independent and cross fiducial localization indicate that fused PET/CT is superior to visual correlation of PET and CT individually in identifying registration errors. The experimental and theoretical findings confirm that PET/CT hybrid system can produce robust intrinsic image alignment and accurate PET/CT image registration with careful user interaction. The maximum spatial error in our experiments is below 4 mm, and is better than the spatial resolution of the PET scanner used. This is considered to be sufficient for most PET/CT applications.

For expanding application of virtual reality, such as rehabilitation engineering, concerns of cybersicknes should be cleared. We have investigated changes in autonomic regulations under real cycling and virtual mountain biking video with the first-person viewpoint. The results showed that the dominant sensory stimuli affected autonomic regulation with different process. The different process will lead to the hints for preventing cybersickness.

The anterior cruciate ligament (ACL) functions as a mechanical stabilizer in the tibiofemoral joint. Over 250,000 Americans each year suffer ACL ruptures and tears, making the ACL the most commonly injured knee ligament. Methods which permit the in situ monitoring of changes in ACL graft mechanical properties during healing are needed. A long term goal in ACL reconstruction is to regenerate the ACL-bone interface. To this end, an understanding of mechanical properties of the ligament-bone interface is needed. However, experimental determination has been difficult due the small length scale (<1 mm) involved and limited resolution of standard techniques. The current study uses elastography to characterize the functional properties of the ACL and the ACL-bone interface under applied load. In a first experiment, bovine joints were excised, cast in an agar gel matrix and externally compressed. In a second experiment, tibiofemoral joints were mounted on a MTS 858 Bionix Testing System. The ACL was loaded at different strain rates and tested to failure while RF data was collected at 5 MHz. For both tensile and compression testing, axial elastograms between successive RF frames were generated using cross-correlation and recorrelation techniques. When the ACL-bone complex was tested in the tibial alignment on the MTS system, compressive strains were found to dominate at the tibial insertion. Compressive strains were observed in the ligament proper when the transducer beam was aligned with respect to the insertion during loading.

Continuous flow blood pumps, such as axial flow and centrifugal pumps, have been gaining interest as circulatory devices for total artificial hearts (TAHs) and a biventricular assist device (BVAD) because of their smaller size and simpler structure compared to pulsatile pumps. However, continuous flow pumps are more prone to suction of the left ventricle than pulsatile pumps are. Sudden increases in flow rate to meet changes in physiological demand, especially in the left pump, often cause ventricle suction. In this study, a control algorithm to prevent suction from occurring in the left ventricle by controlling the rotational speed of the right pump, instead of reducing the cardiac output of the left pump, was developed and investigated. The method was tested in acute animal experiments with calves. The results indicate that this proposed method is capable of preventing suction and could simultaneously maintain circulatory control. A key advantage of this control system is that flow rates can be maximized while avoiding ventricle suction conditions particularly when the circulatory system is unstable such as in a the first few days after operation.

The limitation of current prostheses is not the devices themselves but rather the lack of sufficient independent control sources. A system capable of reading intra muscular EMG signals would greatly increase the number control sources available for prosthesis control. We are developing a multichannel/multifunction prosthetic hand/arm controller system capable of receiving and processing signals from up to sixteen implanted bipolar differential electromyographic (EMG) electrodes. An external prosthesis controller will decipher user intent from telemetry sent over a transcutaneous magnetic link by the implanted electrodes. The same link will provide power for the implanted electrodes. This paper describes some of the technical aspects of the implant and telemetry design.

For better understanding of genetic mechanisms underlying clinical observations, we often want to determine which genes and clinical traits are interrelated. We introduce a computational method that can find co-clusters or groups of genes and clinical parameters that are believed to be closely related to each other based upon given empirical information. The proposed method was tested with data from an Acute Myelogenous Leukemia (AML) study and identified statistically significant co-clusters of genes and clinical traits. The validation of our results with Gene Ontology (GO) as well as the literature suggest that the proposed method can provide biologically meaningful co-clusters of genes and traits.