A novel post-processing scheme for two-dimensional electrical impedance tomography based on artificial neural networks.

OBJECTIVE: Electrical Impedance Tomography (EIT) is a powerful non-invasive technique for imaging applications. The goal is to estimate the electrical properties of living tissues by measuring the potential at the boundary of the domain. Being safe with respect to patient health, non-invasive, and having no known hazards, EIT is an attractive and promising technology. However, it suffers from a particular technical difficulty, which consists of solving a nonlinear inverse problem in real time. Several nonlinear approaches have been proposed as a replacement for the linear solver, but in practice very few are capable of stable, high-quality, and real-time EIT imaging because of their very low robustness to errors and inaccurate modeling, or because they require considerable computational effort. METHODS: In this paper, a post-processing technique based on an artificial neural network (ANN) is proposed to obtain a nonlinear solution to the inverse problem, starting from a linear solution. While common reconstruction methods based on ANNs estimate the solution directly from the measured data, the method proposed here enhances the solution obtained from a linear solver. CONCLUSION: Applying a linear reconstruction algorithm before applying an ANN reduces the effects of noise and modeling errors. Hence, this approach significantly reduces the error associated with solving 2D inverse problems using machine-learning-based algorithms. SIGNIFICANCE: This work presents radical enhancements in the stability of nonlinear methods for biomedical EIT applications.

A Post-Processing Method for Three-Dimensional Electrical Impedance Tomography.

Electrical impedance tomography is a modern biomedical imaging method. Its goal is to image the electrical properties of human tissues. This approach is safe for the patient's health, is non-invasive and has no known hazards. However, the approach suffers from low accuracy. Linear inverse solvers are commonly used in medical applications, as they are strongly robust to noise. However, linear methods can give only an approximation of the solution that corresponds to a linear perturbation from an initial estimate. This paper proposes a novel reconstruction process. After applying a linear solver, the conductivity distribution is post-processed with a nonlinear algorithm, with the aim of reproducing the abrupt change in conductivity at the boundaries between tissues or organs. The results are used to compare the proposed method with three other widely used methods. The proposed method offers higher quality images and a higher robustness to noise, and significantly reduces the error associated with image reconstruction.

On the influence of spread constant in radial basis networks for electrical impedance tomography.

Electrical impedance tomography (EIT) is a non-invasive imaging technique. The main task of this work is to solve a non-linear inverse problem, for which several techniques have been suggested, but none of which gives a very high degree of accuracy. This paper introduces a novel approach, based on radial basis function (RBF) artificial neural networks (ANNs), to solve this problem, and uses several ANNs to obtain the best solution to the EIT inverse problem. ANNs have the potential to directly estimate the solution of the inverse problem with a high degree of accuracy. While different radial basis neural networks do not always perform well on different problems, they usually give good results on some specific problems. This paper evidences a strong correlation between the area of the target and the spread constant of the RBF network that gives the best reconstruction. A solution to automatically estimate the size of the target and pick the best neural network directly from voltage measurements is presented, making the reconstruction process automatic. By automatically selecting the best ANN for each specific set of voltage measurements, the proposed solution gives a more accurate reconstruction of both small and large targets.

Noise reduction using wavelet thresholding of multitaper estimators and geometric approach to spectral subtraction for speech coding strategy.

OBJECTIVES: Noise reduction using wavelet thresholding of multitaper estimators (WTME) and geometric approach to spectral subtraction (GASS) can improve speech quality of noisy sound for speech coding strategy. This study used Perceptual Evaluation of Speech Quality (PESQ) to assess the performance of the WTME and GASS for speech coding strategy. METHODS: This study included 25 Mandarin sentences as test materials. Environmental noises including the air-conditioner, cafeteria and multi-talker were artificially added to test materials at signal to noise ratio (SNR) of -5, 0, 5, and 10 dB. HiRes 120 vocoder WTME and GASS noise reduction process were used in this study to generate sound outputs. The sound outputs were measured by the PESQ to evaluate sound quality. RESULTS: Two figures and three tables were used to assess the speech quality of the sound output of the WTME and GASS. CONCLUSION: There is no significant difference between the overall performance of sound quality in both methods, but the geometric approach to spectral subtraction method is slightly better than the wavelet thresholding of multitaper estimators.

Conditions for generating virtual channels in cochlear prosthesis systems.

Simultaneous electrical stimulation of neighboring electrodes in cochlear prosthesis systems generates channel interaction. However, intermediate channels, or virtual channels between the neighboring electrodes can be created through controlled channel interaction. This effect may be exploited for sending new information to the hearing nerves by stimulating in a suitable manner. The actual stimulation sites are therefore not limited to the number of electrodes. Clinical experiments, however, show that virtual channels are not always perceived. In this paper, electrical simulation with finite element analysis on a half turn human cochlea model is adopted to model the virtual channel effect, and the conditions for generating virtual channels are discussed. Five input current ratios (100/0, 70/30, 50/50, 30/70, 0/100) are applied to generate virtual channels. Three electrode arrays parameters are taken into consideration: distance between electrode contact and modiolus, spacing between adjacent electrode contacts and scale of electrode contact size. By observing the activating function contours, the virtual channel patterns and performances can be measured and examined. The results showed that a broad excitation pattern is necessary to produce the kind of electrode interaction that can form distinct virtual channels.