RESUMEN

Peripheral nerve blocking (PNB) is a standard procedure to support regional anesthesia. Still, correct localization of the nerve's structure is needed to avoid adverse effects; thereby, ultrasound images are used as an aid approach. In addition, image-based automatic nerve segmentation from deep learning methods has been proposed to mitigate attenuation and speckle noise ultrasonography issues. Notwithstanding, complex architectures highlight the region of interest lacking suitable data interpretability concerning the learned features from raw instances. Here, a kernel-based deep learning enhancement is introduced for nerve structure segmentation. In a nutshell, a random Fourier features-based approach was utilized to complement three well-known semantic segmentation architectures, e.g., fully convolutional network, U-net, and ResUnet. Moreover, two ultrasound image datasets for PNB were tested. Obtained results show that our kernel-based approach provides a better generalization capability from image segmentation-based assessments on different nerve structures. Further, for data interpretability, a semantic segmentation extension of the GradCam++ for class-activation mapping was used to reveal relevant learned features separating between nerve and background. Thus, our proposal favors both straightforward (shallow) and complex architectures (deeper neural networks).

Asunto(s)

RESUMEN

Pressure is one of the essential variables to give information about engine condition and monitoring. Direct recording of this signal is complex and invasive, while angular velocity can be measured. Nonetheless, the challenge is to predict the cylinder pressure using the shaft kinematics accurately. In this paper, a time-delay neural network (TDNN), interpreted as a finite pulse response (FIR) filter, is proposed to estimate the in-cylinder pressure of a single-cylinder internal combustion engine (ICE) from fluctuations in shaft angular velocity. The experiments are conducted over data obtained from an ICE operating in 12 different states by changing the angular velocity and load. The TDNN's delay is adjusted to get the highest possible correlation-based score. Our methodology can predict pressure with an R2 >0.9, avoiding complicated pre-processing steps.

RESUMEN

Transfer entropy (TE) is a model-free effective connectivity measure based on information theory. It has been increasingly used in neuroscience because of its ability to detect unknown non-linear interactions, which makes it well suited for exploratory brain effective connectivity analyses. Like all information theoretic quantities, TE is defined regarding the probability distributions of the system under study, which in practice are unknown and must be estimated from data. Commonly used methods for TE estimation rely on a local approximation of the probability distributions from nearest neighbor distances, or on symbolization schemes that then allow the probabilities to be estimated from the symbols' relative frequencies. However, probability estimation is a challenging problem, and avoiding this intermediate step in TE computation is desirable. In this work, we propose a novel TE estimator using functionals defined on positive definite and infinitely divisible kernels matrices that approximate Renyi's entropy measures of order α. Our data-driven approach estimates TE directly from data, sidestepping the need for probability distribution estimation. Also, the proposed estimator encompasses the well-known definition of TE as a sum of Shannon entropies in the limiting case when α → 1. We tested our proposal on a simulation framework consisting of two linear models, based on autoregressive approaches and a linear coupling function, respectively, and on the public electroencephalogram (EEG) database BCI Competition IV, obtained under a motor imagery paradigm. For the synthetic data, the proposed kernel-based TE estimation method satisfactorily identifies the causal interactions present in the data. Also, it displays robustness to varying noise levels and data sizes, and to the presence of multiple interaction delays in the same connected network. Obtained results for the motor imagery task show that our approach codes discriminant spatiotemporal patterns for the left and right-hand motor imagination tasks, with classification performances that compare favorably to the state-of-the-art.

RESUMEN

We introduce Enhanced Kernel-based Relevance Analysis (EKRA) that aims to support the automatic identification of brain activity patterns using electroencephalographic recordings. EKRA is a data-driven strategy that incorporates two kernel functions to take advantage of the available joint information, associating neural responses to a given stimulus condition. Regarding this, a Centered Kernel Alignment functional is adjusted to learning the linear projection that best discriminates the input feature set, optimizing the required free parameters automatically. Our approach is carried out in two scenarios: (i) feature selection by computing a relevance vector from extracted neural features to facilitating the physiological interpretation of a given brain activity task, and (ii) enhanced feature selection to perform an additional transformation of relevant features aiming to improve the overall identification accuracy. Accordingly, we provide an alternative feature relevance analysis strategy that allows improving the system performance while favoring the data interpretability. For the validation purpose, EKRA is tested in two well-known tasks of brain activity: motor imagery discrimination and epileptic seizure detection. The obtained results show that the EKRA approach estimates a relevant representation space extracted from the provided supervised information, emphasizing the salient input features. As a result, our proposal outperforms the state-of-the-art methods regarding brain activity discrimination accuracy with the benefit of enhanced physiological interpretation about the task at hand.

RESUMEN

Human emotion recognition (HER) allows the assessment of an affective state of a subject. Until recently, such emotional states were described in terms of discrete emotions, like happiness or contempt. In order to cover a high range of emotions, researchers in the field have introduced different dimensional spaces for emotion description that allow the characterization of affective states in terms of several variables or dimensions that measure distinct aspects of the emotion. One of the most common of such dimensional spaces is the bidimensional Arousal/Valence space. To the best of our knowledge, all HER systems so far have modelled independently, the dimensions in these dimensional spaces. In this paper, we study the effect of modelling the output dimensions simultaneously and show experimentally the advantages in modeling them in this way. We consider a multimodal approach by including features from the Electroencephalogram and a few physiological signals. For modelling the multiple outputs, we employ a multiple output regressor based on support vector machines. We also include an stage of feature selection that is developed within an embedded approach known as Recursive Feature Elimination (RFE), proposed initially for SVM. The results show that several features can be eliminated using the multiple output support vector regressor with RFE without affecting the performance of the regressor. From the analysis of the features selected in smaller subsets via RFE, it can be observed that the signals that are more informative into the arousal and valence space discrimination are the EEG, Electrooculogram/Electromiogram (EOG/EMG) and the Galvanic Skin Response (GSR).

Asunto(s)

RESUMEN

A novel method for hand movement pattern recognition from electromyography (EMG) biological signals is proposed. These signals are recorded by a three-channel data acquisition system using surface electrodes placed over the forearm, and then processed to recognize five hand movements: opening, closing, supination, flexion, and extension. Such method combines the Hilbert-Huang analysis with a fuzzy clustering classifier. A set of metrics, calculated from the time contour of the Hilbert Spectrum, is used to compute a discriminating three-dimensional feature space. The classification task in this feature-space is accomplished by a two-stage procedure where training cases are initially clustered with a fuzzy algorithm, and test cases are then classified applying a nearest-prototype rule. Empirical analysis of the proposed method reveals an average accuracy rate of 96% in the recognition of surface EMG signals.

Asunto(s)

RESUMEN

In recent years Microelectrode recording (MER) analysis has proved to be a powerful localization tool of basal ganglia for Parkinson disease's treatment, especially the Subthalamic Nucleus (STN). In this paper, a signal-dependent method is presented for identification of the STN and other brain zones in Parkinsonian patients. The proposed method, refereed as optimal wavelet feature extraction method (OWFE), is constructed by lifting schemes (LS), which are a flexible and fast implementation of the wavelet transform (WT). The operators in the LS are optimized by means of Genetic Algorithms and Lagrange multipliers considering information contained in MER signals. Then a basic Bayesian classifier (LDC) is used to identify STN and other types of basal ganglia nuclei. The proposed method introduced several advantages from similar works reported in literature. First, the method is signal-dependent and non a priori information is required to decompose the MER signal. Second, the classification accuracy is mostly depended on the feature selection stage because it is not enhanced by elaborated classifiers such as support vector machines or hidden Markov models. Finally, the generalization property of the OWFE has been validated with two databases and different types of classifiers such as k-NN classifier and quadratic Bayesian classifier (QDC). Results have shown that proposed method is able to identify the STN with average accuracy superior than 97%.

Asunto(s)

RESUMEN

Stereotactic neurosurgery for Parkinson's disease (PD) is one of the most used treatments for relief symptoms of this degenerative disorder. Current methods include ablation and deep brain stimulation (DBS) that can be applied to the various nuclei in the basal ganglia (BG), for instance to the Subthalamic nucleus (STN) or the Ventral medial nucleus (Vim). Identification of thus regions must be rigorous and within a minimum position error. Usually, skilled specialist identifies the brain area by comparing and listening to the rhythm created by the temporal and spatial aggregation of action potentials presented in microelectrode recordings (MER). We present a novel system for automatic identification of the various nuclei in the BG which addresses the limitations of the subjectivity and the non-stationary nature of MER signals. This system incorporates the time-frequency analysis using the Hilbert-Huang Transform (HHT), which is a recent tool for processing nonlinear and non-stationary data, with a dynamic classifier based on Hidden Markov Models (HMM). Classification accuracy in two different databases is compared to validate the performance of the proposed method. Results show that system can recognize selected nuclei with a mean accuracy of 90%.

Asunto(s)