International Journal of Neural Systems最新文献

Neural Architecture Search (NAS) has recently shown a powerful ability to engineer networks automatically on various tasks. Most current approaches navigate the search direction with the validation performance-based architecture evaluation methodology, which estimates an architecture's quality by training and validating on a specific large dataset. However, for small-scale datasets, the model's performance on the validation set cannot precisely estimate that on the test set. The imprecise architecture evaluation can mislead the search to sub-optima. To address the above problem, we propose an efficient multi-objective evolutionary zero-shot NAS framework by evaluating architectures with zero-cost metrics, which can be calculated with randomly initialized models in a training-free manner. Specifically, a general zero-cost metric design principle is proposed to unify the current metrics and help develop several new metrics. Then, we offer an efficient computational method for multi-zero-cost metrics by calculating them in one forward and backward pass. Finally, comprehensive experiments have been conducted on NAS-Bench-201 and MedMNIST. The results have shown that the proposed method can achieve sufficiently accurate, high-throughput performance on MedMNIST and 20[Formula: see text]faster than the previous best method.

Recent work presented a framework for space-time-resolved neurophysiological process imaging that augments existing electromagnetic source imaging techniques. In particular, a nonlinear Analytic Kalman filter (AKF) has been developed to efficiently infer the states and parameters of neural mass models believed to underlie the generation of electromagnetic source currents. Unfortunately, as the initialization determines the performance of the Kalman filter, and the ground truth is typically unavailable for initialization, this framework might produce suboptimal results unless significant effort is spent on tuning the initialization. Notably, the relation between the initialization and overall filter performance is only given implicitly and is expensive to evaluate; implying that conventional optimization techniques, e.g. gradient or sampling based, are inapplicable. To address this problem, a novel efficient framework based on blackbox optimization has been developed to find the optimal initialization by reducing the signal prediction error. Multiple state-of-the-art optimization methods were compared and distinctively, Gaussian process optimization decreased the objective function by 82.1% and parameter estimation error by 62.5% on average with the simulation data compared to no optimization applied. The framework took only 1.6[Formula: see text]h and reduced the objective function by an average of 13.2% on 3.75[Formula: see text]min 4714-source channel magnetoencephalography data. This yields an improved method of neurophysiological process imaging that can be used to uncover complex underpinnings of brain dynamics.

This paper describes a longitudinal study to analyze the effects of acoustic stimulation with Binaural Beats (BBs) at 14[Formula: see text]Hz (beta band) in patients with Parkinson's Disease (PD). Participants ([Formula: see text], age [Formula: see text], stage [Formula: see text] Hoehn and Yahr scale) listened to binaural stimulation for 10[Formula: see text]min a day, 3 days a week, during six months and were assessed 3 times during this period using electroencephalography (EEG), cognitive (PD-CRS), quality of life (PDQ-39) and wearing-off (WOQ-19) tests. During each assessment (basal, and after 3 and 6 months), the relative power in theta band was analyzed before, during and after the stimulation. Focusing the analysis on the motor cortex, the results obtained have confirmed the initial hypothesis for the first session, but they have shown a habituation effect which decreases its efficiency with time. Also, different reactions have been detected among individuals, with some reacting as expected from the beginning, while others would react in an opposite way at the beginning but they have shown afterwards a tendency towards the expected outcome. Anyhow, the relative power of the theta band was reduced between the first and the last session for more than half of the participants, although with very different values. Subtle changes have also been observed in some items of the PD-CRS, PDQ-39 and WOQ-19 tests.

Parkinson's disease (PD) is a neurodegenerative condition with constantly increasing prevalence rates, affecting strongly life quality in terms of neuromotor and cognitive performance. PD symptoms include voice and speech alterations, known as hypokinetic dysarthria (HD). Unstable phonation is one of the manifestations of HD. Repetitive transcranial magnetic stimulation (rTMS) is a rehabilitative treatment thathas been shown to improve some motor and non-motor symptoms of persons with PD (PwP). This study analyzed the phonation functional behavior of 18 participants (13 males, 5 females) with PD diagnosis before (one pre-stimulus) and after (four post-stimulus) evaluation sessions of rTMS treatment, to assess the extent of changes in their phonation stability. Participants were randomized 1:1 to receive either rTMS or sham stimulation. Voice recordings of a sustained vowel [a:] taken immediately before and after the treatment, and at follow-up evaluation sessions (immediately after, at six, ten, and fourteen weeks after the baseline assessment) were processed by inverse filtering to estimate a biomechanical correlate of vocal fold tension. This estimate was further band-pass filtered into EEG-related frequency bands. Log-likelihood ratios (LLRs) between pre- and post-stimulus amplitude distributions of each frequency band showed significant differences in five cases actively stimulated. Seven cases submitted to the sham protocol did not show relevant improvements in phonation instability. Conversely, four active cases did not show phonation improvements, whereas two sham cases did. The study provides early preliminary insights into the capability of phonation quality assessment by monitoring neuromechanical activity from acoustic signals in frequency bands aligned with EEG ones.

The prevalence of dementia is currently increasing worldwide. This syndrome produces a deterioration in cognitive function that cannot be reverted. However, an early diagnosis can be crucial for slowing its progress. The Clock Drawing Test (CDT) is a widely used paper-and-pencil test for cognitive assessment in which an individual has to manually draw a clock on a paper. There are a lot of scoring systems for this test and most of them depend on the subjective assessment of the expert. This study proposes a computer-aided diagnosis (CAD) system based on artificial intelligence (AI) methods to analyze the CDT and obtain an automatic diagnosis of cognitive impairment (CI). This system employs a preprocessing pipeline in which the clock is detected, centered and binarized to decrease the computational burden. Then, the resulting image is fed into a Convolutional Neural Network (CNN) to identify the informative patterns within the CDT drawings that are relevant for the assessment of the patient's cognitive status. Performance is evaluated in a real context where patients with CI and controls have been classified by clinical experts in a balanced sample size of [Formula: see text] drawings. The proposed method provides an accuracy of [Formula: see text] in the binary case-control classification task, with an AUC of [Formula: see text]. These results are indeed relevant considering the use of the classic version of the CDT. The large size of the sample suggests that the method proposed has a high reliability to be used in clinical contexts and demonstrates the suitability of CAD systems in the CDT assessment process. Explainable artificial intelligence (XAI) methods are applied to identify the most relevant regions during classification. Finding these patterns is extremely helpful to understand the brain damage caused by CI. A validation method using resubstitution with upper bound correction in a machine learning approach is also discussed.

Electrical stimulation of the peripheral nervous system is a promising therapeutic option for several conditions; however, its effects on tissue and the safety of the stimulation remain poorly understood. In order to devise stimulation protocols that enhance therapeutic efficacy without the risk of causing tissue damage, we constructed computational models of peripheral nerve and stimulation cuffs based on extremely high-resolution cross-sectional images of the nerves using the most recent advances in computing power and machine learning techniques. We developed nerve models using nonstimulated (healthy) and over-stimulated (damaged) rat sciatic nerves to explore how nerve damage affects the induced current density distribution. Using our in-house computational, quasi-static, platform, and the Admittance Method (AM), we estimated the induced current distribution within the nerves and compared it for healthy and damaged nerves. We also estimated the extent of localized cell damage in both healthy and damaged nerve samples. When the nerve is damaged, as demonstrated principally by the decreased nerve fiber packing, the current penetrates deeper into the over-stimulated nerve than in the healthy sample. As safety limits for electrical stimulation of peripheral nerves still refer to the Shannon criterion to distinguish between safe and unsafe stimulation, the capability this work demonstrated is an important step toward the development of safety criteria that are specific to peripheral nerve and make use of the latest advances in computational bioelectromagnetics and machine learning, such as Python-based AM and CNN-based nerve image segmentation.

Developmental dyslexia is characterized by a deficit of phonological awareness whose origin is related to atypical neural processing of speech streams. This can lead to differences in the neural networks that encode audio information for dyslexics. In this work, we investigate whether such differences exist using functional near-infrared spectroscopy (fNIRS) and complex network analysis. We have explored functional brain networks derived from low-level auditory processing of nonspeech stimuli related to speech units such as stress, syllables or phonemes of skilled and dyslexic seven-year-old readers. A complex network analysis was performed to examine the properties of functional brain networks and their temporal evolution. We characterized aspects of brain connectivity such as functional segregation, functional integration or small-worldness. These properties are used as features to extract differential patterns in controls and dyslexic subjects. The results corroborate the presence of discrepancies in the topological organizations of functional brain networks and their dynamics that differentiate between control and dyslexic subjects, reaching an Area Under ROC Curve (AUC) up to 0.89 in classification experiments.

Alzheimer's disease (AD) is the most prevalent form of dementia. Although there is no current cure, medical treatment can help to control its progression. Hence, early-stage diagnosis is crucial to maximize the living standards of the patients. Biochemical markers and medical imaging in combination with neuropsychological tests represent the most extended diagnosis procedure. However, these techniques require specialized personnel and long processing time. Furthermore, the access to some of these techniques is often limited in crowded healthcare systems and rural areas. In this context, electroencephalography (EEG), a non-invasive technique to obtain endogenous brain information, has been proposed for the diagnosis of early-stage AD. Despite the valuable information provided by clinical EEG and high density montages, these approaches are impractical in conditions such as those described above. Consequently, in this study, we evaluated the feasibly of using a reduced EEG montage with only four channels to detect early-stage AD. For this purpose, we involved eight clinically diagnosed AD patients and eight healthy controls. The results we obtained reveal similar accuracies ([Formula: see text]-value[Formula: see text]0.66) for the reduced montage (0.86) and a 16-channel montage (0.87). This suggests that a four-channel wearable EEG system could be an effective tool for supporting early-stage AD detection.

While the brain connectivity network can inform the understanding and diagnosis of developmental dyslexia, its cause-effect relationships have not yet enough been examined. Employing electroencephalography signals and band-limited white noise stimulus at 4.8 Hz (prosodic-syllabic frequency), we measure the phase Granger causalities among channels to identify differences between dyslexic learners and controls, thereby proposing a method to calculate directional connectivity. As causal relationships run in both directions, we explore three scenarios, namely channels' activity as sources, as sinks, and in total. Our proposed method can be used for both classification and exploratory analysis. In all scenarios, we find confirmation of the established right-lateralized Theta sampling network anomaly, in line with the assumption of the temporal sampling framework of oscillatory differences in the Theta and Gamma bands. Further, we show that this anomaly primarily occurs in the causal relationships of channels acting as sinks, where it is significantly more pronounced than when only total activity is observed. In the sink scenario, our classifier obtains 0.84 and 0.88 accuracy and 0.87 and 0.93 AUC for the Theta and Gamma bands, respectively.