Multiply sectioned Bayesian networks for neuromuscular diagnosis.

A prototype neuromuscular diagnostic system (PAINULIM) that diagnoses painful or impaired upper limbs has been developed based on Bayesian networks. This paper presents nonmathematically the major knowledge representation issues that arose in the development of PAINULIM. Motivated by the computational overhead of large application domains, and the desire to provide a user with an interface that gives a focused display of a subdomain of current interest, we built PAINULIM using the idea of multiply sectioned Bayesian networks. A preliminary evaluation of PAINULIM with 76 patients has demonstrated good clinical performance.

Quality control in nerve conduction studies with coupled knowledge-based system approach.

Contemporary equipment used for nerve conduction studies is usually capable of computerized measurement of latency, amplitude, duration, and area of nerve and muscle action potentials and resulting conduction velocities. Abnormalities can be due to technical error or disease. Identification of technical error is a major element of quality control in electromyography, and artificial intelligence could be useful for this purpose. We have developed a coupled knowledge-based prototype system (QUALICON) to assess the correctness of recording and stimulating characteristics in routine conduction studies. QUALICON extracts numeric features from CMAPs or SNAPs, which are translated into symbolic form to drive a Bayesian network. The network uses high-level knowledge to infer the quality of stimulating and recording electrode placement as well as polarity and stimulus strength making recommendations as to the likely technical error when abnormal potentials are detected. A preliminary assessment shows that QUALICON performs as well as manual assessment performed by professionals.

Iterative methods for solving the Gabor expansion: considerations of convergence.

J.G. Daugman's (1988) neural network solution to the Gabor expansion of an image is reformulated as a steepest descent implementation. Nonlinear optimization theory is then applied to select an appropriate convergence factor. Two quasi-Newton-based nonlinear optimization techniques are applied to improve the convergence for certain types of lattice.

Practical digital filters for reducing EMG artefact in EEG seizure recordings.

In long-term scalp EEG monitoring of epileptic patients it is virtually impossible, in the present state of the technology, to avoid movement-related artefacts. These often obscure EEG information about the location of the seizure focus. One important example of such artefact is EMG activity. Its removal or suppression is sometimes enough to make otherwise useless EEG traces readable. Different methods of filtering have been applied towards that end. We routinely use a 15 Hz setting on our polygraph in obtaining EEG seizure printouts. We have recently examined digital filters which attenuate EMG beyond what is possible with the 15 Hz filter. A concern has been that the filters are practical in that they run in real time on a simple microprocessor and cause a minimum of confusion between smoothed artefact and actual brain activity.

Optimal edge detector design I: parameter selection and noise effects.

This paper comments on the optimality of the Laplacian of a Gaussian edge detection filter which localizes edges through zero crossings in the filtered image. The arguments of both Marr and Hildreth, and Dickey and Shanmugam are reviewed to establish that the filter is optimal in the sense of maximizing output image energy near edge features. This filter's principal advantage over other edge detectors is that its response is user-adjustable through selection of a single parameter, the Gaussian standard deviation. However, no clear method for the selection of this parameter has been provided. The problem is addressed here by applying the filter to two ideal periodic edge models blurred by a Gaussian distribution point-spread function. The observed response to the edge spacing and blur standard deviation is then translated into a filter parameter design procedure. The problems of optimum filter performance in the presence of additive Gaussian noise are then addressed. The problem of selecting the sampled filter's coefficient word size is dealt with in a companion paper.

Optimal Edge Detector Design II: Coefficient Quantization.

A companion paper describes the design and implementation of the Laplacian of a Gaussian edge detection filter which localizes edges through zero crossings in the filtered image. Accuracy in the presence of noise has been found to be proportional to the square root of the filter's standard deviation. Digital implementation of any continuous filter requires sampling and coefficient quantization. The sampled filter was examined, but a method is proposed here for selection of a minimum coefficient word size for direct-form implementation to satisfy in-band rejection bounds.

Automation of the seizure investigation unit at the University of British Columbia Health Sciences Centre Hospital.

At the University of British Columbia Health Sciences Centre Hospital we have constructed an automated Seizure Investigation Unit for long term monitoring of epileptic patients. A central component of this system is a new device, developed at the University of British Columbia, which prevents video and EEG records of seizures recorded on video tape from being over-recorded. Computer technology is relied upon to a considerable degree in our unit. Computers are seen, in this phase of development of the SIU, as a means of making patient monitoring less dependent on supervision. They can help to redirect human energy towards complex analysis rather than time consuming and simple monitoring tasks.

Effects of detector wobble motion on image noise in positron emission tomography.

The effects of detector wobble motion on image variance is described analytically and verified by simulation studies. Simulation results in the form of variance images are presented for various phantoms, sampled under both uniform and wobble-motion conditions. It is shown that under most measurement conditions in a typical PET study, the image precision is degraded due to the nonuniform projection variance introduced by the wobble motion. Using the wobble parameters of the UBC-TRIUMF development of PETT VI, the rms noise is found to be increased by 9 percent due to wobbling.

Lensless matched filter: operating principle, sensitivity to spectrum shift, and third-order holographic aberrations.

Synthesizing a Fresnel zone plate inside a Vander Lugt matched filter (MF) results in a lensless matched filter (LLMF) that is space-invariant from the point of view of vignetting apertures. In practice the unfocused output components of the LLMF do not degrade its SNR. A LLMF displacement appears as a combined shift of its filter and lens components. The third-order holographic aberrations of the MF and LLMF, and particularly their root-mean-square astigmatisms, are compared via computer-generated plots.