The muscle groups involved in human movement can be regarded as a highly coordinated complex system, and muscle fatigue may alter the coordination relationships among muscles. However, fatigue is a complex physiological process involving both intermuscle coordination and local metabolic responses, making it difficult to achieve comprehensive and stable assessment using a single signal source. In this study, multisource and multichannel biological signals were collected under different fatigue states, and surface electromyography (sEMG) complex-network features and near-infrared spectroscopy (NIRS) muscle oxygen features were extracted to characterize fatigue-related changes. For sEMG signals, muscle functional networks were constructed from the time-domain, frequency-domain, and time–frequency-domain perspectives to analyze changes in intermuscle coordination. For NIRS signals, muscle oxygen features were extracted to describe local oxygen metabolism during fatigue development. The extracted features were fused and used to construct a ThinResNet-based fatigue recognition model. The results showed that the fusion of sEMG and NIRS features improved fatigue recognition performance compared with using sEMG features alone. The proposed fusion model achieved an average accuracy of 90.03% in binary classification and 73.17% in three-class classification under the all-subject setting. These results indicate that multisource physiological feature fusion can provide a feasible approach for comprehensive muscle fatigue assessment.

To improve the efficiency of passive cooling systems for battery thermal management system (BTMS), this work innovatively utilizes the synergistic effect of flexible materials and high thermal conductivity materials. A flexible composite phase change material (FCPCM) with good adhesion, high thermal conductivity and high enthalpy was prepared using paraffin wax, expanded graphite, a polyester-based flexible matrix, and micro-copper powder. To characterize the material properties, the key thermal properties of FCPCM were verified through experiments. Battery pack charging and discharging experiments are designed, analyzed and compared to show the changes in the temperature field inside the battery pack before and after FCPCM cooling. The results show that the thermal conductivity of the prepared FCPCM is 2.72 W/(m & sdot;K) and the interface thermal resistance is 0.561 degrees C/W. At a 3C discharge rate, the maximum temperature inside the battery pack with passive FCPCM dropped from 71.5 degrees C to 39.3 degrees C, and the maximum temperature difference was 4.23 degrees C. The battery operated consistently within its optimal temperature range (20 degrees C-40 degrees C). This research provides a passive thermal management solution that meets the functional requirements for BTMS.

Lithium-ion batteries play a crucial role in electric vehicles (EVs) owing to their high energy density and long cycle life. However, maintaining their operating temperature within the optimal range requires advanced battery thermal management systems (BTMS) to meet the ever-evolving performance requirements of EVs. The purpose of this study is to effectively integrate liquid cooling with composite phase change material (CPCM) cooling and to enhance the contribution of the heat transfer performance of phase change materials to BTMS efficiency by adopting an innovative multilayer phase change material structure. To better analyze the characteristics of water flow direction, cooling plate width, inlet flow rate, CPCM doping ratio, multilayer CPCM structure, and battery pack temperature distribution in the composite thermal management system, the charging and discharging thermal behaviors of the battery were simulated and verified. The results indicate that the BTMS proposed in this study can significantly improve the temperature field distribution within the battery pack. Specifically, under an ambient temperature of 298.15 K, a water flow rate of 1 L min−1 and a 2C discharge rate, the BTMS based on multilayer CPCM coupling can reduce the maximum internal temperature of the battery pack to 303.93 K, with a maximum temperature difference of only 1.1 K. Compared with the scenario without cooling, the maximum temperature is reduced by 27.02 K, and the maximum temperature difference is reduced by 8.1 K. This study provides an efficient solution for the design and optimization of BTMS in EV lithium-ion batteries.

Brain-computer interface (BCI) is very vital in interactive rehabilitation training. Action recognition in rehabilitation training is the basis and key to realize interactive rehabilitation training. Advances in deep learning now enable the development of motion image models with superior ecoding accuracy. However, extracting single features, such as surface electromyography (sEMG) and electroencephalogram (EEG) signals, may lead to poor decoding performance in muscle fatigue states. Therefore, we present a dual-scale fusion neural network (DFNN) for accurate recognition of hand rehabilitation training actions under muscle fatigue, which realizes feature learning of EEG and EMG fusion. We evaluated the performance of the model on our own multimodal dataset collected when performing rehabilitation training actions in fatigue states, achieving an average recognition accuracy of 93.33%.

Muscle fatigue refers to the decline in strength or endurance during sustained muscle contraction, which not only affects athletic performance but may also increase the risk of muscle injury. Due to its non-invasive nature and real-time monitoring capabilities, surface electromyography (sEMG) has been widely applied in muscle fatigue research. However, muscle fatigue is a complex physiological process, and relying solely on sEMG signals makes it difficult to comprehensively assess muscle fatigue conditions in the human body. In response to the limitations of sEMG in muscle fatigue assessment, this study proposes a multimodal fatigue recognition model that integrates sEMG and sweat signals. Experimental results demonstrate that the model achieves a recognition accuracy of 87.5%, significantly outperforming the single-modality sEMG approach (77.9%), thereby validating the effectiveness of the multimodal fusion strategy. This study provides new insights for the development of robust fatigue monitoring systems, demonstrating potential applications in the fields of sports science and occupational health.

Deep learning has demonstrated remarkable performance in emotion recognition tasks based on Electroencephalogram (EEG) signals. The utilization of the spatial and temporal features of EEG signals plays a pivotal role in the task of emotion recognition. Traditional Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Transformer models have encountered challenges in effectively capturing both the temporal and spatial characteristics of EEG signals. Therefore, achieving a comprehensive integration of temporal and spatial features remains a demanding task. In this paper, we propose a Convolution-Multilayer Perceptron Network (CMLP-Net) based on the 2D representation of EEG signals. CMLP-Net comprises a temporal-stream shared convolution, a time-refinement temporal-spatial convolution, and a spatial interaction Multilayer Perceptron (MLP). The temporal-stream shared convolution is employed to uncover shared features across multiple consecutive temporal windows. The time-refinement temporal-spatial convolution is utilized to capture effective temporal-spatial features effectively. Lastly, the spatial interaction MLP facilitates spatial interactions among subregions within the EEG feature maps, thereby enhancing the global spatial dependency of the features. The proposed method was evaluated on the popular DEAP dataset, where 14 channels relevant to EEG emotions were selected. Experimental results validate that the proposed approach exhibits state-of-the-art performance in binary emotion recognition tasks, achieving average accuracies of 98.65%, 98.70%, and 98.63% in valence, arousal, and dominance dimensions, respectively. Furthermore, CMLP-Net also demonstrates excellent capabilities in handling multi-state emotion recognition tasks.

Wearable sweat sensors exhibit significant promise in the realm of smart healthcare. Efficient and non-contaminating continuous sweat collection, transport, and disposal are crucial for enhancing detection accuracy. In this study, a novel wearable sweat sensor is presented that integrates various biomimetic structures, including cactus-inspired spine, shorebird-inspired conical through-holes, and frog-foot-inspired micro-grooves. These structures facilitate sweat droplets to be automatically transported from the skin to the sensing chamber and discharged without external power. First, theoretical analysis, simulation, and experiments based on self-driven droplet mechanisms to determine the optimal transport efficiency parameters for biomimetic structures are conducted. Next, a novel sweat transport module is fabricated through ultra-precision 3D printing according to these optimal parameters. And this module has excellent sweat collection performance contrastively. Finally, the novel biomimetic sweat transport module is integrated with a high-sensitivity glucose detection electrode, resulting in a wearable sweat glucose detection system. Human experiments monitoring glucose metabolism in sweat verify the correlation between glucose concentrations in blood and sweat, which proves the feasibility of reflecting changes in the body's circulatory system by monitoring changes in sweat composition. A wearable sweat sensor that is inspired by multiple biological water collection structures has been demonstrated. The rapid collection and discharge of sweat are realized by the droplet self-driving mechanism. The glucose concentration in sweat is detected and shows a great consistency with that in blood. This work will promote the progress toward the practical application of wearable sweat sensors. image

Motor imagery brain-computer interfaces (MI-BCIs) play a crucial role in fields such as robot control and stroke rehabilitation. With the flourishing development of deep learning, there has been a continuous emergence of motor imagery deep learning models with higher decoding accuracy. Activation of specific brain regions, such as the motor regions in the frontal and parietal lobes, carries information about motor imagery. However, most studies only extract features from the entire brain region, neglecting the potential features of specific brain regions. This may lead to poorer decoding performance. Therefore, this article proposes a parallel-hierarchical neural network (PHNN), which implements a hierarchical approach to feature learning from brain region level to multi-level fusion. Learning at the brain region level mainly explores the key information of specific brain regions (region-level) and the overall features of the entire brain (global-level), to obtain region-level features (RLF) and global-level features (GLF). Furthermore, given that region-level features contain crucial information within specific brain regions, multi-level fusion is employed to capture and fully utilize the differences and connections between the RLF and GLF, resulting in more discriminative multi-level fusion features (MLFF). We evaluate the model performance on the publicly available BCI Competition IV-2a dataset and High Gamma dataset, achieving recognition accuracies of 84.67% and 94.02%, respectively.

The image recognition of cancer cells plays an important role in diagnosing and treating cancer. Deep learning is suitable for classifying histopathological images and providing auxiliary technology for cancer diagnosis. The convolutional neural network is employed in the classification of histopathological images; however, the model's accuracy may decrease along with the increase in network layers. Extracting appropriate image features is helpful for image classification. In this paper, different features of histopathological images are represented by extracting features of the gray co-occurrence matrix. These features are recombined into a 16 × 16 × 3 matrix to form an artificial image. The original image and the artificial image are fused by summing the softmax output. The histopathological images are divided into the training set, validation set, and testing set. Each training dataset consists of 1500 images, while the validation dataset and test dataset each consist of 500 images. The results indicate that the effectiveness of our fusion model is demonstrated through significant improvements in accuracy, precision, recall, and F1-score, with an average accuracy reaching 99.31%. This approach not only enhances the classification performance of tissue pathology images but also holds promise for advancing computer-aided diagnosis in cancer pathology.

Fatigue assessment is especially important in long-term physical work and training to avoid injury caused by muscle fatigue. The surface electromyography (sEMG) signal has been widely used to detect muscle fatigue states. The purpose of this article is to establish muscle functional networks and provide a comprehensive assessment of muscle fatigue using multi-channel sEMG signals from the back muscles. A muscle functional network is constructed based on the Pearson correlation coefficients between the channels of sEMG signals, and we explore how the economic properties and small-world properties of the muscle network change with cost. Then, the network threshold is determined by changes in small-world properties. We extract network parameters through complex network methods to quantify and analyze the differences in muscle functional networks under different fatigue states. The most significant results of this novel approach indicate that the economic and small-world properties of the muscle network gradually decrease with increasing fatigue, and network parameters decrease significantly after fatigue. The method presented in this article provides a new foundation for assessing muscle fatigue.

The accurate characterization of the surface microstructure of ultra-high temperature ceramics after thermal shocks is of great practical significance for evaluating their thermal resistance properties. In this paper, a fractal reconstruction method for the surface image of Ultra-high temperature ceramics after repeated thermal shocks is proposed. The nonlinearity and spatial distribution characteristics of the oxidized surfaces of ceramics were extracted. A fractal convolutional neural network model based on deep learning was established to realize automatic recognition of the classification of thermal shock cycles of ultra-high temperature ceramics, obtaining a recognition accuracy of 93.74%. It provides a novel quantitative method for evaluating the surface character of ultra-high temperature ceramics, which contributes to understanding the influence of oxidation after thermal shocks.

Ceramics are commonly used as high-temperature structural materials which are easy to fracture because of the propagation of thermal shock cracks. Characterizing and controlling crack propagation are significant for the improvement of the thermal shock resistance of ceramics. However, observing crack morphology, based on macro and SEM images, costs much time and potentially includes subjective factors. In addition, complex cracks cannot be counted and will be simplified or omitted. Fractals are suitable to describe complex and inhomogeneous structures, and the multifractal spectrum describes this complexity and heterogeneity in more detail. This paper proposes a crack characterization method based on the multifractal spectrum. After thermal shocks, the multifractal spectrum of alumina ceramics was obtained, and the crack fractal features were extracted. Then, a deep learning method was employed to extract features and automatically classify ceramic crack materials with different strengths, with a recognition accuracy of 87.5%.

This work utilized a combination of experimental evidence and fractal geometric method to assess the effect of crack extension concerning the thermal shock on residual strength of ceramics. Sintered alumina (Al2O3) ceramic slabs were bundled and quenched in water under different thermal shock temperatures. The fractal dimension of thermal shock crack patterns on the interior surface and the cooled surface was calculated by the Box-counting method. Fracture energy of a fractal pattern of microcracks in quasi-brittle solids was employed to explain the relationship between crack length and fractal dimensions. The results show that if the crack propagation has the same crack length but a larger fractal dimension, it will absorb more fracture energy. The thermal shock crack patterns of Al2O3 ceramics with different grain sizes were analyzed, and the smaller grain size ceramic had a higher fractal dimension of crack patterns than the larger one.

To develop prosthesis controlled by spontaneous brain signal, we studied the motion intention underlying EEG activity with the imagination of right hand and left hand. During hand motor imaginary, EEG activities are collected by 64 electrodes from healthy adults. The temporal and frequency characteristics associated with the imagination of movement are explored and combined as the feature vector for the linear discriminant analysis. Our findings show a classification accuracy of 77% for our datasets and 83% for the datasets of BCI Competition 2003.

Strength training using patients' desired force level is helpful to improve training effect and promote rehabilitation. Generally, force levels are recognized by applying EMG or biomechanical information, these methods were not suitable for patients who lost important muscle groups or have weakened muscle functions. This paper proposed a method for identifying force level based on cerebral hemoglobin information, rather than the information depending on limbs. Ten subjects performed pedaling movement in three force levels. Features were extracted in both the time-domain and frequency-domain, with deoxygenated hemoglobin (deoxy) and the difference between oxygenated hemoglobin (oxy) and deoxy as parameters. Important frequency bands (0.01-0.03Hz, 0.03-0.06Hz, 0.06-0.09Hz, 0.09- 0.12Hz) were confirmed by performing power spectrum density analysis. And significant measure channels were selected by performing one-way analyses of variance on three time periods around the start of movement. Force level was recognized by applying extreme learning machine (ELM). The corresponding precision rate was up to 78.7%. The proposed identification method was not restricted to the existence of limbs or the strength of limb information. It was realized based on brain information recorded in a real movement environment; it is helpful to realize the desired force level of subjects and to provide a control command for rehabilitation training equipment.

Brain-computer interface (BCI) instead of depending on the brain's normal output pathways, can use electroencephalogram (EEG) from the scalp as the representation of brain activity to control external devices. EEG during motor imagery (MI) provides a non-muscular communication way to control external devices and has advantage of non-invasiveness and high time resolution. However the application is still limited by time-consuming training and poor classification rate with multiple categories etc. We recorded 64-channel scalp EEG from eight healthy subjects during imagery tasks of left, right hand movements and stop. EEG was analyzed in time-frequency distribution and spatial topographies were explored too. A one versus one common spatial pattern was applied to construct feature vector and then linear discriminant analysis was used for the classification. For the purpose of real time control in the future, small training size was used and we got discrimination among three types of motor imagery at the accuracy rate about 90%.

Intelligent prosthetics aim to provide a communication channel between disabled people and external world. Among them, walking-assistive device to help people with lower-limb disability attracts attention but still limited by unnatural gait or lack accurate real-time control. Here we studied the coupling between the upper limbs and the lower limbs since the whole body actually involved during walking and there exists coordination among limbs. We collected motion information of different joints from healthy people during bipedal walking by attitudes sensors. After gait cycle identified and normalized, the relationship between different limbs was built. Based on the polynomial approximation, the knee joint angle can be estimated by the shoulder. Thus in the future it can be potentially applied to control walking-assistive prosthetics by the upper limbs movement instead of the motion information of lower limbs.

Lots of patients suffer from lack of motion ability or even lost of limbs. Rehabilitation robot aims to assist them to regain some motion ability. We study the recognition of ten upper limb movements based on surface electromyography (sEMG) signals for future potential robotic arm control. After sEMG are collected and denoised, wavelet transform is used to construct the feature vector. Recognition rate among ten movements reaches 96.75% revealed by eight channel sEMG signals and 96.25% by two channels sEMG. This provides the basis for further real-time control of robotic arm.

For the intelligent control of lower limb prostheses, kinematic reference trajectories are required. Due to individual difference and various preferred reference walking-speed, a general reference knee angle trajectory is difficult to adapt different people and their different walking speed. To solve this problem, gait kinematic information of thirty healthy subjects was collected at their preferred speeds and the relationship between kinematic models of knee joint and speeds was analyzed. Each knee angle trajectory was modeled based on Fourier functions, and the parameters of models were proved to be related to walking speed. Finally, kinematic models for knee angle trajectories were reconstructed based on regression equations. The reconstructed trajectories not only can adapt to the various walking-speeds, but also adapt to different people. The result of analysis showed that the reconstructed trajectories matched the measured motion trajectories well.

BACKGROUND: In recent years, MR images have been increasingly used in therapeutic applications such as image-guided radiotherapy (IGRT).However, images with low contrast values and noises present challenges for image segmentation.OBJECTIVE: The objective of this study is to develop a robust method based on fuzzy C-means (FCM) method which can segment MR images polluted with Gaussian noise.METHODS: A modified FCM algorithm accommodating non-local pixel information via Hausdorff distance was developed for segmenting MR images.The membership and objective functions were modified accordingly.Segmentations with different weights of the Hausdorff distance were compared.RESULTS: Segmentation tests using synthetic and MR images showed that the proposed algorithm was better at resolving boundaries and more robust to Gaussian noise.By segmenting a sample MR image of a tumor, we further showed the capability of the method in capturing the centroid of the target region.CONCLUSIONS: The modified FCM algorithm with neighboring information can be used to segment blurry images with potential applications in segmenting motion MR images in image-guided radiotherapy (IGRT).