Prediction of wave-front sensor slope measurements with artificial neural networks.

For adaptive optical systems to compensate for atmospheric turbulence effects, the wave-front perturbation must be measured with a wave-front sensor (WFS) and corrected with a deformable mirror. One limitation in this process is the time delay between the measurement of the aberrated wave front and implementation of the proper correction. Statistical techniques exist for predicting the atmospheric aberrations at the time of correction based on the present and past measured wave fronts. However, for the statistical techniques to be effective, key parameters of the atmosphere and the adaptive optical system must be known. These parameters include the Fried coherence length r(0), the atmospheric wind-speed profile, and the WFS slope measurement error. Neural networks provide nonlinear solutions to adaptive optical problems while offering the possibility to function under changing seeing conditions without actual knowledge of the current state of the key parameters. We address the use of neural networks for WFS slope measurement prediction with only the noisy WFS measurements as inputs. Where appropriate, we compare with classical statistical-based methods to determine if neural networks offer true benefits in performance.

Selecting optimal experiments for multiple output multilayer perceptrons.

Where should a researcher conduct experiments to provide training data for a multilayer perceptron? This question is investigated, and a statistical method for selecting optimal experimental design points for multiple output multilayer perceptrons is introduced. Multiple class discrimination problems are examined using a framework in which the multilayer perceptron is viewed as a multivariate nonlinear regression model. Following a Bayesian formulation for the case where the variance-covariance matrix of the responses is unknown, a selection criterion is developed. This criterion is based on the volume of the joint confidence ellipsoid for the weights in a multilayer perceptron. An example is used to demonstrate the superiority of optimally selected design points over randomly chosen points, as well as points chosen in a grid pattern. Simplification of the basic criterion is offered through the use of Hadamard matrices to produce uncorrelated outputs.

Computer-aided breast cancer detection and diagnosis of masses using difference of Gaussians and derivative-based feature saliency.

A new model-based vision (MBV) algorithm is developed to find regions of interest (ROI's) corresponding to masses in digitized mammograms and to classify the masses as malignant/benign. The MBV algorithm is comprised of five modules to structurally identify suspicious ROI's, eliminate false positives, and classify the remaining as malignant or benign. The focus of attention module uses a difference of Gaussians (DoG) filter to highlight suspicious regions in the mammogram. The index module uses tests to reduce the number of nonmalignant regions from 8.39 to 2.36 per full breast image. Size, shape, contrast, and Laws texture features are used to develop the prediction module's mass models. Derivative-based feature saliency techniques are used to determine the best features for classification. Nine features are chosen to define the malignant/benign models. The feature extraction module obtains these features from all suspicious ROI's. The matching module classifies the regions using a multilayer perceptron neural network architecture to obtain an overall classification accuracy of 100% for the segmented malignant masses with a false-positive rate of 1.8 per full breast image. This system has a sensitivity of 92% for locating malignant ROI's. The database contains 272 images (12 b, 100 microm) with 36 malignant and 53 benign mass images. The results demonstrate that the MBV approach provides a structured order of integrating complex stages into a system for radiologists.

Processing wave-front-sensor slope measurements using artificial neural networks.

For adaptive-optics systems to compensate for atmospheric turbulence effects, the wave-front perturbation must be measured with a wave front sensor (WPS), and key parameters of the atmosphere and the adaptive-optics system must be known. Two parameters of particular interest include the Fried coherence length r(0) and the WPS slope measurement error. Statistics-based optimal techniques, such as the minimum variance phase reconstructor, have been developed to improve the imaging performance of adaptive-optics systems. However, these statistics-based models rely on knowledge of the current state of the key parameters. Neural networks provide nonlinear solutions to adaptive-optics problems while offering the possibility of adapting to changing seeing conditions. We address the use of neural networks for three tasks: (l) to reduce the WPS slope measurement error, (2) to estimate the Fried coherence length r(0), and (3) to estimate the variance of the WPS slope measurement error. All of these tasks are accomplished by using only the noisy WPS measurements as input. Where appropriate, we compare our method with classical statistics-based methods to determine if neural networks offer true benefits in performance. Although a statistics-based method is found to perform better than a neural network in reducing WPS slope measurement error, neural networks perform better in estimating the variance of the WPS slope measurement error, and both methods perform well in estimating r(0).

Use of artificial neural networks for Hartmann-sensor lenslet centroid estimation.

For adaptive optical systems to compensate for atmospheric-turbulence effects, the wave-front perturbation must be measured with a wave-front sensor (WFS). A Hartmann WFS typically divides the optical aperture into subapertures and then measures the slope of the wave front within each subaperture. Hartmann WFS slope measurements are based on estimating the location of the centroid of the image that is formed from a guide star within each subaperture. Conventional techniques for centroid estimation involve the use of a linear estimator and conversion tables. Neural networks provide nonlinear solutions to this problem. We address the use of neural networks for estimating the location of the centroid from the subaperture image. We find that neural networks provide more accurate estimates over a larger dynamic range and with less variance than do the conventional linear centroid estimator.

Artificial neural networks for early detection and diagnosis of cancer.

Why use neural networks? The reasons commonly cited in the literature for using artificial neural networks for any problem are many and varied. They learn from experience. They work where other algorithms fail. They generalize from the training examples to perform well on independent test data. They reduce the number of false alarms without increasing significantly the number of false negatives. They are fast and are easier to use than conventional statistical techniques, especially when multiple prognostic factors are needed for a given problem. These factors have been overly promoted for the neural techniques. The common theme of this paper is that artificial neural networks have proven to be an interesting and useful alternate processing strategy. Artificial neural techniques, however, are not magical solutions with mystical abilities that work without good engineering. With good understanding of their capabilities and limitations they can be applied productively to problems in early detection and diagnosis of cancer. The specific cancer applications which will be used to demonstrate current work in artificial neural networks for cancer detection and diagnosis are breast cancer, liver cancer and lung cancer.

Optical wavelet transforms from computer-generated holography.

An optical implementation of a wavelet transform is presented. Optical Haar wavelets are created by the use of computer-generated holography. Two different holographic techniques are explored: (1) interferogram and (2) detour-phase. A discrete representation of a continuous wavelet transform is obtained by the optical correlation of an image with a Haar mother wavelet. Experimental results are compared with their digital simulations.

The multilayer perceptron as an approximation to a Bayes optimal discriminant function.

The multilayer perceptron, when trained as a classifier using backpropagation, is shown to approximate the Bayes optimal discriminant function. The result is demonstrated for both the two-class problem and multiple classes. It is shown that the outputs of the multilayer perceptron approximate the a posteriori probability functions of the classes being trained. The proof applies to any number of layers and any type of unit activation function, linear or nonlinear.