Adaptive filtering of evoked potentials with radial-basis-function neural network prefilter

Evoked potentials (EPs) are time-varying signals typically buried in relatively large background noise. To extract the EP more effectively from noise, we had previously developed an approach using an adaptive signal enhancer (ASE) (Chen et al., 1995). ASE requires a proper reference input signal for its optimal performance. Ensemble- and moving window-averages were formerly used with good results. In this paper, we present a new method to provide even more effective reference inputs for the ASE. Specifically, a Gaussian radial basis function neural network (RBFNN) was used to preprocess raw EP signals before serving as the reference input. Since the RBFNN has built-in nonlinear activation functions that enable it to closely fit any function mapping, the output of RBFNN can effectively track the signal variations of EP. Results confirmed the superior performance of ASE with RBFNN over the previous method.     

Learning contextual relationships in mammograms using a hierarchical pyramid neural network

This paper describes a pattern recognition architecture, which we term hierarchical pyramid/neural network (HPNN), that learns to exploit image structure at multiple resolutions for detecting clinically significant features in digital/digitized mammograms. The HPNN architecture consists of a hierarchy of neural networks, each network receiving feature inputs at a given scale as well as features constructed by networks lower in the hierarchy. Networks are trained using a novel error function for the supervised learning of image search/detection tasks when the position of the objects to be found is uncertain or ill defined. We have evaluated the HPNN's ability to eliminate false positive (FP) regions of interest generated by the University of Chicago's (UofC) Computer-aided diagnosis (CAD) systems for microcalcification and mass detection. Results show that the HPNN architecture, trained using the uncertain object position (UOP) error function, reduces the FP rate of a mammographic CAD system by approximately 50% without significant loss in sensitivity. Investigation into the types of FPs that the HPNN eliminates suggests that the pattern recognizer is automatically learning and exploiting contextual information. Clinical utility is demonstrated through the evaluation of an integrated system in a clinical reader study. We conclude that the HPNN architecture learns contextual relationships between features at multiple scales and integrates these features for detecting microcalcifications and breast masses.     

Adaptation of spatiotemporal mechanisms by increment and decrement stimuli

Sawtooth modulation has been used in the past to examine visual sensitivity to luminance increments and decrements. The threshold elevation caused by adaptation depends on the spatial profile of the stimulus field and the polarities of the adaptation and test stimuli. We hypothesized that the adaptation effects reflect a change in the sensitivity of the spatiotemporal channels that detect the stimuli. We used a 2-deg disk centered in a larger surround field. Five levels of contrast between the test field and surround were investigated: equiluminant, three intermediate levels, and dark. At each contrast, observers adapted for 5 s to 2-Hz sawtooth modulation (rapid-on or rapid-off). Immediately after adaptation, thresholds were measured for detection of a single cycle of either a rapid-on or a rapid-off waveform. Varying the contrast of the surround affected observers' sensitivity to the polarity of the sawtooth stimulus to the extent that the pattern of sensitivity with the equiluminant surround was the opposite of that with the dark surround. To examine temporal factors, we measured thresholds for slow (500-ms ramps) and fast (8.3-ms pulses) test stimuli. The adaptation effect was preserved with the ramp stimuli but not with the pulse stimuli. Blurring the edge between the test and surround fields in the equiluminant surround condition raised thresholds for all sawtooth test stimuli, suggesting that spatiotemporal channels sensitive to high spatial frequencies and low temporal frequencies facilitate detection in that condition. These findings suggest that adaptation to sawtooth modulation can differentially effect the sensitivity of ON and OFF pathways, but the relative desensitization of each pathway depends on an interaction with the adaptation state of spatiotemporal channels that are involved in detection.     

Toughness in synthetic and biological multilayered systems

Toughness in hard biological tissues is associated with fibrous or lamellar structures that deflect or stop growing cracks. In some cases, such as nacreous shell, protein interlayers absorb much of the crack energy. In other tissues, such as tooth enamel, the toughness derives from the mineral microstructure, and the small amount of residual protein apparently has little effect. There have been a number of efforts to make tough synthetic materials using layered structures. In this work, freeform fabrication has been used to make layered structures with a view to introducing similar toughness into brittle materials. Results are presented for epoxy-glass composites with glass fabric interlayers, porous alumina back-filled with aluminium metal, and layered glass-ceramic/silver materials.     

The thermochemical structure and evolution of Earth's mantle: constraints and numerical models

Geochemical observations place several constraints on geophysical processes in the mantle, including a requirement to maintain several distinct reservoirs. Geophysical constraints limit plausible physical locations of these reservoirs to a thin basal layer, isolated deep 'piles' of material under large-scale mantle upwellings, high-viscosity blobs/plums or thin strips throughout the mantle, or some combination of these. A numerical model capable of simulating the thermochemical evolution of the mantle is introduced. Preliminary simulations are more differentiated than Earth but display some of the proposed thermochemical processes, including the generation of a high-mu mantle reservoir by recycling of crust, and the generation of a high-(3)He/(4)He reservoir by recycling of residuum, although the resulting high-(3)He/(4)He material tends to aggregate near the top, where mid-ocean-ridge melting should sample it. If primitive material exists as a dense basal layer, it must be much denser than subducted crust in order to retain its primitive (e.g. high-(3)He) signature. Much progress is expected in the near future.     

Fourier-Bessel field calculation and tuning of a CW annular array

A 1-D Fourier-Bessel series method for computing and tuning the linear lossless field of flat continuous wave (CW) annular arrays is given and discussed with both numerical simulation and experimental verification. The technique provides a new method for modelling and manipulating the propagated field by linking the quantized surface pressure profile to a set of limited diffraction Bessel beams propagating into the medium. In the limit, these become a known set of nondiffracting Bessel beams satisfying the lossless linear wave equation, which allow us to derive a linear matrix formulation for the field in terms of the ring pressures on the transducer surface. Tuning (beamforming) of the field then follows by formulating a least squares design with respect to the transducer ring pressures. Results are presented in the context of a 10-ring annular array operating at 2.5 MHz in water.     

Nondestructive subharmonic imaging

Ultrasound contrast agent microbubbles are intravascular agents that can be used to estimate blood perfusion. Blood perfusion may be estimated by destroying the bubbles in a vascular bed and observing the refresh of contrast agents back into the vascular bed. Contrast agents can be readily destroyed by traditional imaging techniques. The design of a nondestructive imaging technique is necessary for the accurate quantification of contrast agent refresh. In this work, subharmonic imaging is investigated as a method for nondestructive imaging with the contrast agent microbubble MP1950 (Mallinckrodt, Inc., St. Louis, MO). Optical observation during insonation, in conjunction with a modified Rayleigh-Plesset (R-P) analysis, provides insight into the mechanisms of and parameters required for subharmonic frequency generation. Subharmonic imaging with a transmission frequency that is the same as the resonant frequency of the bubble is shown to require a minimum pressure of insonation that is greater than the experimentally-observed bubble destruction threshold. Subharmonic imaging with a transmission frequency that is twice the resonant frequency of the bubble produces a subharmonic frequency response while minimizing bubble instability. Optimization is performed using optical experimental analysis and R-P analysis     

Remodeling of fibrinogen by endothelial cells in dependence on fibronectin matrix assembly. Effect of substratum wettability

The endothelization of cardiovascular implants is desirable to improve their blood compatibility. The capacity of the endothelial cells to attach, migrate, proliferate and function on the implant surface depends on the presence of matrix proteins such as fibronectin (FN) and fibrinogen (FNG). In this study, we show that the deposition of fibrinogen into extracellular matrix-like structures by human umbilical vein endothelial cells (HUVEC) is dependent on FN matrix formation. We found further that the process of organization of both adsorbed and soluble FN and FNG is dependent on the wettability of materials since it was observed only on a hydrophilic and not on a hydrophobic model surface. ₃ integrin was involved in the process of cell attachment to adsorbed FNG, while the mechanism of FNG fibrillogenesis required the activity of the ₁ integrin. Studies of EC morphology showed the predominant peripheral organization of actin filaments and the formation of distinct leading and trailing cell edges suggesting a motile phenotype of cells when they are seeded on FNG. In summary, we concluded that adsorbed fibrinogen may enhance the motility of HUVEC and that soluble FNG requires FN matrix assembly to be organized in fibrilar structures.     

Evaluation of a program to motivate impaired driving offenders to install ignition interlocks

Approximately 30,000 alcohol ignition interlocks, which are designed to prevent the operation of a vehicle if the driver has been drinking, are in use in the US and Canada. Ignition interlock programs are also being initiated in Sweden and Australia. The best-controlled studies that are currently available suggest that ignition interlocks are effective in reducing impaired driving recidivism while on the vehicle. However, in the US, the practical effectiveness of these devices is limited because only a small number of offenders are willing to install them in order to drive legally. This paper reports on a study of a court policy that created a strong incentive for impaired driving offenders to install interlocks by making traditional penalties, such as jail or electronically monitored house arrest, the alternative to participation in an interlock program. Comparison of the recidivism rates of offenders subject to this policy with offenders in similar, nearby courts, not using interlocks, indicated that the policy was producing substantial reductions in DUI recidivism.