Larry E. Banta

The positron emission mammography/tomography breast imaging and biopsy system (PEM/PET): design, construction and phantom-based measurements

Tomographic breast imaging techniques can potentially improve detection and diagnosis of cancer in women with radiodense and/or fibrocystic breasts. We have developed a high-resolution positron emission mammography/tomography imaging and biopsy device (called PEM/PET) to detect and guide the biopsy of suspicious breast lesions. PET images are acquired to detect suspicious focal uptake of the radiotracer and guide biopsy of the area. Limited-angle PEM images could then be used to verify the biopsy needle position prior to tissue sampling. The PEM/PET scanner consists of two sets of rotating planar detector heads. Each detector consists of a 4 x 3 array of Hamamatsu H8500 flat panel position sensitive photomultipliers (PSPMTs) coupled to a 96 x 72 array of 2 x 2 x 15 mm(3) LYSO detector elements (pitch = 2.1 mm). Image reconstruction is performed with a three-dimensional, ordered set expectation maximization (OSEM) algorithm parallelized to run on a multi-processor computer system. The reconstructed field of view (FOV) is 15 x 15 x 15 cm(3). Initial phantom-based testing of the device is focusing upon its PET imaging capabilities. Specifically, spatial resolution and detection sensitivity were assessed. The results from these measurements yielded a spatial resolution at the center of the FOV of 2.01 +/- 0.09 mm (radial), 2.04 +/- 0.08 mm (tangential) and 1.84 +/- 0.07 mm (axial). At a radius of 7 cm from the center of the scanner, the results were 2.11 +/- 0.08 mm (radial), 2.16 +/- 0.07 mm (tangential) and 1.87 +/- 0.08 mm (axial). Maximum system detection sensitivity of the scanner is 488.9 kcps microCi(-1) ml(-1) (6.88%). These promising findings indicate that PEM/PET may be an effective system for the detection and diagnosis of breast cancer.

Parameter Incremental Learning Algorithm for Neural Networks

In this paper, a novel stochastic (or online) training algorithm for neural networks, named parameter incremental learning (PIL) algorithm, is proposed and developed. The main idea of the PIL strategy is that the learning algorithm should not only adapt to the newly presented input-output training pattern by adjusting parameters, but also preserve the prior results. A general PIL algorithm for feedforward neural networks is accordingly presented as the first-order approximate solution to an optimization problem, where the performance index is the combination of proper measures of preservation and adaptation. The PIL algorithms for the multilayer perceptron (MLP) are subsequently derived. Numerical studies show that for all the three benchmark problems used in this paper the PIL algorithm for MLP is measurably superior to the standard online backpropagation (BP) algorithm and the stochastic diagonal Levenberg-Marquardt (SDLM) algorithm in terms of the convergence speed and accuracy. Other appealing features of the PIL algorithm are that it is computationally as simple as the BP algorithm, and as easy to use as the BP algorithm. It, therefore, can be applied, with better performance, to any situations where the standard online BP algorithm is applicable

Estimation of limestone particle mass from 2D images

Abstract The strength of asphalt pavements is largely determined by the distribution of particle sizes (gradation) and shapes in the aggregate used in the mixture. Currently, these parameters are determined by manual sampling. The manual method is time-consuming and cannot provide real-time feedback for process control purposes. In this paper, an approach for predicting particle mass based on 2D electronic images is described. Crushed limestone aggregates, similar to those used in asphalt pavement mixtures were placed on a light table and imaged using a CCD video camera and framegrabber. The images were processed to separate touching and overlapping particles, define the edges of the particles and to calculate certain features of the particle silhouettes, such as area, centroid and shape-related features. Several dimensionless parameters were defined, based on the image features. A multiple linear regression model was created, using the dimensionless parameters as regressor variables to predict particle mass. Regressor coefficients were found by fitting to a sample of 501 particles ranging in size from 4.75 mm

Sensor fusion for mining robots

The use of robots for work in hazardous or unpleasant environments is one factor driving the demand for machines of ever-increasing autonomy and intelligence. Such machines are required to sense and interpret situations, plan strategies, and execute tasks with nearly absolute reliability. Negotiation of complex environments requires the use of a variety of different sensor types and the interpretation of conflicting or missing data, diagnosis of faulty sensors, and the ability to reconfigure a system to work with a partially inoperative sensor suite. This paper focuses on the issues of integration of information from disparate sensor types in the presence of noise and uncertainty. The application is a mobile robot called the autonomous navigation testbed being used at West Virginia University for research in mining robot applications. This paper describes both traditional control techniques and neural network-based methods being used to interpret data from a variety of sensors on the mobile testbed. >

A self tuning navigation algorithm

The author has to developed a system for improving the dead reckoning navigation accuracy for a mobile robot. He reports on the development of a parameter estimation scheme for correcting modeling errors due to wear, misalignment, or calibration degradation. A least-squares algorithm is used to give a suboptimal but robust estimate of the bias parameters. Simulations show marked improvement of the navigation accuracy of the robot. A test bed is being constructed at West Virginia University to verify and refine the technique.<<ETX>>

Mode-based navigation for autonomous mine vehicles

The development of an autonomous mobile robot for use in underground mines is described. The navigation scheme combines elements of both hierarchical control and reactive or subsumptive-type control. The robot navigates by sensing the environment and selecting a navigational mode that is appropriate to the circumstances and to the robots mission. Examples of navigational modes are wall following, collision avoidance and homing. The modes are implemented in modules formed by combinations of neural network processors and conventional control algorithms. The overall control system architecture and the navigational strategies of the experimental robot vehicle are also described. >

Image profile area calculation based on circular sample measurement calibration

A practical approach of measurement calibration is presented for obtaining the true area of the photographed objects projected in the 2-D image scene. The calibration is performed using three circular samples with given diameters. The process is first to obtain the ratio mm/pixel in two orthogonal directions, and then use the obtained ratios with the total number of pixels scanned within projected area of the object of interest to compute the desired area. Compared the optically measured areas with their corresponding true areas, the results show that the proposed method is quite encouraging and the relevant application also proves the approach adequately accurate.

Design and fabrication of an assistive device for arm rehabilitation using twisted string system

In this research the twisting cable mechanism was employed to design and develop a light elbow assistive robot. A particular type of fishing line was experimentally tested for use as the strand material, and the behavior was compared to the proposed model. A correction parameter called effective diameter was derived to adapt the model to the experimental data. To ensure the consistency of the model, hysteresis of two-strand cables was tested. A curve was fitted to the experimental data and the most linear range was selected to be used in the mechanism. Moreover, a single degree of freedom elbow mechanism for flexion and extension was designed and constructed. Ultimately, to ensure the applicability of the mechanism, an ordinary Activity of Daily Living (ADL) was used and the angle of the twisted strand actuator motor as a function of the motor rotation was computed.

Multivariable Robust Control of a Simulated Hybrid Solid Oxide Fuel Cell Gas Turbine Plant

This paper presents a systematic approach to the multivariable robust control of a hybrid fuel cell gas turbine plant. The hybrid configuration under investigation comprises a physical simulation of a 300kW fuel cell coupled to a 120kW auxiliary power unit single spool gas turbine. The facility provides for the testing and simulation of different fuel cell models that in turn help identify the key issues encountered in the transient operation of such systems. An empirical model of the facility consisting of a simulated fuel cell cathode volume and balance of plant components is derived via frequency response data. Through the modulation of various airflow bypass valves within the hybrid configuration, Bode plots are used to derive key input/output interactions in Transfer Function format. A multivariate system is then built from individual transfer functions, creating a matrix that serves as the nominal plant in an H-Infinity robust control algorithm. The controller’s main objective is to track and maintain hybrid operational constraints in the fuel cell’s cathode airflow, and the turbo machinery states of temperature and speed, under transient disturbances. This algorithm is then tested on a Simulink/MatLab platform for various perturbations of load and fuel cell heat effluence.

Determination of an Empirical Transfer Function of a Solid Oxide Fuel Cell Gas Turbine Hybrid System Via Frequency Response Analysis

This paper presents the study of the effect variations in the heat effluence from a solid oxide fuel cell (SOFC) has on a gas turbine hybrid configuration. The SOFC is simulated through hardware at the U.S. Department of Energy, National Energy Technology Laboratory (NETL). The gas turbine, compressor, recuperative heat exchanger, and other balance of plant components are represented by actual hardware in the Hybrid Performance Test Facility at NETL. Fuel cell heat exhaust is represented by a combustor that is activated by a fuel cell model that computes energy release for various sensed system states System structure is derived by means of frequency response data generated by the sinusoidal oscillation of the combustor fuel valve over a range of frequencies covering three orders of magnitude. System delay and order are obtained from Bode plots of the magnitude and phase relationships between input and output parameters. Transfer functions for mass flow, temperature, pressure, and other states of interest are derived as a function of fuel valve flow, representative of fuel cell thermal efftuent The Bode plots can validate existing analytical transfer functions, provide steady state error detection, give a stability margin criterion for the fuel valve input, estimate system bandwidth, identify any nonminimum phase system behavior, pinpoint unstable frequencies, and serve as an element of a piecewise transfer function in the development of an overall transfer function matrix covering all system inputs and outputs of interest. Further loop shaping techniques and state space representation can be applied to this matrix in a multivariate control algorithm.