Search

Doppler ultrasound (DU) is used to detect venous gas emboli (VGE) post dive as a marker of decompression stress for diving physiology research as well as new decompression procedure validation to minimize decompression sickness risk. In this paper, we propose the first deep learning model for VGE grading in DU audio recordings.A database of real-world data was assembled and labeled for the purpose of developing the algorithm, totaling 274 recordings comprising both subclavian and precordial measurements. Synthetic data was also generated by acquiring baseline DU signals from human volunteers and superimposing laboratory-acquired DU signals of bubbles flowing in a tissue mimicking material. A novel squeeze-and-excitation deep learning model was designed to effectively classify recordings on the 5-class Spencer scoring system used by trained human raters.On the real-data test set, we show that synthetic data pretraining achieves average ordinal accuracy of 84.9% for precordial and 90.4% for subclavian DU which is a 24.6% and 26.2% increase over training with real-data and time-series augmentation only. The weighted kappa coefficients of agreement between the model and human ground truth were 0.74 and 0.69 for precordial and subclavian respectively, indicating substantial agreement similar to human inter-rater agreement for this type of data.The present work demonstrates the first application of deep-learning for DU VGE grading using a combination of synthetic and real-world data.The proposed method can contribute to accelerating DU analysis for decompression research.

This paper presents a multi-objective deep learning method for the purpose of achieving image fusion or pansharpening in remote sensing. This method uses a Denoising Autoencoder (DA). Two terms are...

This paper presents a multispectral image fusion method for remote sensing applications by using an improved estimation of detail maps. This estimation is achieved by considering a linear combinati...

This paper presents a deep learning-based pansharpening method for fusion of panchromatic and multispectral images in remote sensing applications. This method can be categorized as a component substitution method in which a convolutional autoencoder network is trained to generate original panchromatic images from their spatially degraded versions. Low resolution multispectral images are then fed into the trained convolutional autoencoder network to generate estimated high resolution multispectral images. The fusion is achieved by injecting the detail map of each spectral band into the corresponding estimated high resolution multispectral bands. Full reference and no-reference metrics are computed for the images of three satellite datasets. These measures are compared with the existing fusion methods whose codes are publicly available. The results obtained indicate the effectiveness of the developed deep learning-based method for multispectral image fusion.

This paper presents a generative model method for multispectral image fusion in remote sensing which involves training without supervision. This method eases the supervision of learning and also uses a multi-objective loss function to achieve image fusion. The loss function used incorporates both spectral and spatial distortions. Two discriminators are designed to minimize the spectral and spatial distortions of the generative output. Extensive experimentations are conducted using three public domain datasets. The comparison results across four reduced-resolution and three full-resolution objective metrics show the superiority of the developed method over several recently developed methods.

This paper presents a review of multi-objective deep learning methods that have been introduced in the literature for speech denoising. After stating an overview of conventional, single objective deep learning, and hybrid or combined conventional and deep learning methods, a review of the mathematical framework of the multi-objective deep learning methods for speech denoising is provided. A representative method from each speech denoising category, whose codes are publicly available, is selected and a comparison is carried out by considering the same public domain dataset and four widely used objective metrics. The comparison results indicate the effectiveness of the multi-objective method compared with the other methods, in particular when the signal-to-noise ratio is low. Possible future improvements that can be achieved are also mentioned.

This paper provides a deep learning-based approximation of the MultiSpectral Band Intensity component by considering the joint multiplication of adjacent spectral channels. This calculation is conducted as part of a component substitution approach for the fusion of PANchromatic and MultiSpectral images in remote sensing. After calculating the band-dependent intensity elements, a deep learning model is trained to learn the nonlinear relationship between the PAN image and its nonlinear intensity elements. Low Resolution MultiSpectral bands are then fed into a trained network to achieve a high resolution MultiSpectral band estimation. Experiments performed on three datasets indicate that the established deep learning estimation methodology offers better performance compared to current approaches based on a number of objective metrics.

In this paper, we propose a new three-dimensional fractional-order chaotic system with four nonlinear terms. The continuous fractional-order system is solved as a discrete map by employing Adomian Decomposition Method (ADM). Corresponding phase portraits for commensurate system are investigated and all Lyapunov characteristic exponents are calculated based on the QR decomposition method. It is found that the proposed fractional-order system exhibits chaotic behavior with fractional order less than three. Moreover, the problem of master-slave synchronization for commensurate system is considered and Lyapunov-based and modified versions of Laplacian-based synchronization methods are employed for this purpose. Simulation results for synchronization of the new system with these methods are presented. Effectiveness of these synchronization methods are compared against each other based on different performance criteria which shows the superiority of the modified Laplacian-based

Existing deep learning-based speech denoising approaches require clean speech signals to be available for training. This paper presents a deep learning-based approach to improve speech denoising in real-world audio environments by not requiring the availability of clean speech signals as reference in training mode. A fully convolutional neural network is trained by using two noisy realizations of the same speech signal, one used as the input and the other as the target of the network. Two noisy realizations of the same speech signal are generated by using a mid-side stereo microphone. Extensive experimentations are conducted to show the superiority of the developed deep speech denoising approach over the conventional supervised deep speech denoising approach based on four commonly used performance metrics as well as a subjective testing.

A new three-dimensional (3D) chaotic system is proposed with four nonlinear terms which include two quadratic terms. To analyze the dynamical properties of the new system, mathematical tools such as Lyapunov exponents (LEs), Kaplan–York dimensions, observability constants, and bifurcation diagram have been exploited. The results of these calculations verify the specific features of the new system and further determine the effect of different system parameters on its dynamics. The proposed system has been experimentally implemented as an analog circuit which practically confirms its predicted chaotic behavior. Moreover, the problem of master–slave synchronization of the proposed chaotic system is considered. To solve this problem, we propose a new method for designing a nonfragile Takagi–Sugeno (T–S) fuzzy static output feedback synchronizing controller for a general chaotic T–S system and applied the method to the proposed system. Some practical advantages are achieved employing the new nonlinear controller as well as using system output data instead of the full-state data and considering gain variations because of the uncertainty in values of practical components used in implementation the controller. Then, the designed controller has been realized using analog devices to synchronize two circuits with the proposed chaotic dynamics. Experimental results show that the proposed nonfragile controller successfully synchronizes the chaotic circuits even with inexact analog devices.

In this paper, a novel image fusion method for remote sensing applications is proposed. In order to estimate the primitive detail map, the band coefficients of multispectral images are computed using least squares method. To refine the detail map and better image fusion, an adaptive method is proposed to compute the weights of linear combinations of panchromatic (Pan) and multispectral (MS) gradients. The injected weights are calculated using particle swarm optimization (PSO). Two data sets obtained by WorldView-3 and QuickBird satellites are employed for testing the assessment and comparing with state-of-art methods.

Present work describes a promising method in image fusion remote sensing applications. Due to intrinsic properties of deep neural networks (DNN) in image reconstruction, a novel pansharpening method presents based on multi resolution analysis (MRA) framework. First, a low resolution Panchromatic (LR Pan) image is constructed using its high resolution (HR) version. Then, the relationship between LR/HR Pan images are used to reconstruct the HR Multispectral (MS) image utilizing the LR MS. For our work, two datasets are considered and for each of them, the effect of several parameters such as window size, overlapping percentage and number of training samples on spectral distortion are considered. After training DNN, the LR MS image is given to the trained network as input to obtain MS image with better spatial details and finally the fused image obtains using MRA framework. Comparison with state of art methods, the proposed method has better results from objective and visual perspectives.