生物医学工程学杂志最新文献

Mental fatigue, a detrimental psychophysiological state induced by high-intensity cognitive tasks, impairs athletes' attention, reaction, and decision-making, increasing the risk of errors and injuries. Traditional questionnaire-based assessments of mental fatigue suffer from subjectivity and response bias, whereas objective examining and analyzing methods such as electroencephalography (EEG) are often costly and time-consuming, highlighting the need for efficient and convenient objective approaches. This study proposes a hybrid convolutional neural network (CNN)-Transformer model that combines CNN-based feature extraction with Transformer-based global dependency modeling for accurate and efficient mental fatigue recognition. The model was evaluated on the Shanghai Jiao Tong University Emotion EEG Dataset (SEED) and the Sustained-Attention Driving Task (SADT) dataset. The model proposed in this study achieved accuracies of 78.07% and 85.42%, respectively, outperforming conventional methods and demonstrating good cross-subject generalization. Furthermore, channel analysis highlighted the occipital regions' signal as key contributors to fatigue detection, providing theoretical basis for the development of portable and lightweight device for mental fatigue monitoring. Overall, this work provides a feasible solution for efficient and objective mental fatigue detection, and has potential applications in athletic training monitoring and performance optimization.

Flexible electrode as a revolutionary brain computer interface (BCI) technology in the field of neural engineering, has achieved high-fidelity acquisition and long-term stable transmission of electroencephalographic signals through their exceptional bio-compatibility. This review systematically elucidates the design paradigms and material innovation systems of flexible electrodes, focusing on their transitional medical value from aspects such as electrode materials, signal acquisition and processing. It identifies the current technical bottlenecks that urgently need to be broken through and outlines the future development directions, hoping to provide a systematic technical road-map and evaluation framework for the technical development of next-generation BCI.

In vitro hemolysis testing for blood pumps currently faces several challenges, including randomness in control group selection, and numerous sources of uncertainty in the testing methods. These issues lead to high uncertainty, insufficient result credibility, and limited comparability, which hinders the effective evaluation of blood damage induced by blood pumps. This study aims to address these limitations by developing a magnetically-levitated blood pump benchmark model and optimizing the hemolysis testing protocol to reduce result uncertainty. A magnetic bearing was utilized to minimize blood damage, and the injection molding was employed to enhance the machining precision of the pump. The experimental procedures, including blood collection, test loop setup, and the testing process, were optimized to effectively control experimental uncertainty. The results showed that the performance curve of the benchmark pump model was flat, and the coefficient of variation for the hydraulic experimental results was less than 5%. The secondary flow path exhibited good blood washout with no thrombus formation. Under low-flow condition, the average normalized index of hemolysis (NIH) was 0.014 g/100L, with a coefficient of variation of 19.50%. Under high-flow condition, the average NIH was 0.045 g/100L, with a coefficient of variation of 16.45%. The hemolysis values under both conditions were lower than the Abbott CentriMag. In conclusion, we designed a benchmark blood pump model with excellent hemocompatibility and optimized hemolysis testing protocol, which led to low uncertainty in experimental results. The benchmark and optimized hemolysis protocol help to improve the credibility and comparability of in vitro hemolysis testing data, providing a reliable solution for both the industry and regulatory agencies to assess hemocompatibility.

Dining robots significantly enhance the quality of life for individuals with upper limb motor impairments by enabling autonomous feeding. This paper systematically reviewed the technological evolution and representative products in this field, with a focused analysis of key technologies including kinematic modeling, trajectory planning, and intelligent control. Future development trends were also discussed, highlighting the need for innovative structural designs, optimized human-robot interaction, and deeper multi-source sensory fusion to advance the field toward more precise and human-like robotic feeding systems.

To enhance the accuracy of depression (DP) recognition, this paper proposes a DP recognition method based on improved variational mode decomposition (VMD). Firstly, the adaptive particle swarm optimization (APSO) algorithm is adopted to improve VMD, aiming to find the optimal combination of the number of modes K and the penalty factor α, and thereby achieve the decomposition of electroencephalogram (EEG) signals. Then EEG signals are reconstructed based on the fitness between signal components and the original signal, noise is removed to obtain pure EEG signals, and their frequency-space features are extract. Next, a self-attention (SA) mechanism is introduced into the parallel architecture of two-dimensional convolutional neural network (2D-CNN) and bidirectional long short-term memory network (BiLSTM), to form the 2D-CNN-BiLSTM-SA detection model. Finally, the frequency-spatial features of the EEG signal are input into 2D-CNN-BILSTM-SA for DP recognition. Through comparative experiments on public datasets, the research results of this paper show that the improved VMD not only outperforms VMD but also achieves DP recognition accuracy rate of up to 94.47%. In conclusion, the method proposed in this paper provides a potential computer-aided tool for DP recognition.

To accurately capture and address the multi-dimensional feature variations in cross-subject motor imagery electroencephalogram (MI-EEG), this paper proposes a time-frequency transform and Riemannian manifold based domain adaptation network (TFRMDANet) in a high-dimensional brain source space. Source imaging technology was employed to reconstruct neural electrical activity and compute regional cortical dipoles, and the multi-subband time-frequency feature data were constructed via wavelet transform. The two-stage multi-branch time-frequency-spatial feature extractor with squeeze-and-excitation (SE) modules was designed to extract local features and cross-scale global features from each subband, and the channel attention and multi-scale feature fusion were introduced simultaneously for feature enhancement. A Riemannian manifold embedding-based structural feature extractor was used to capture high-order discriminative features, while adversarial training promoted domain-invariant feature learning. Experiments on public BCI Competition IV dataset 2a and High-Gamma dataset showed that TFRMDANet achieved classification accuracies of 77.82% and 90.47%, with Kappa values of 0.704 and 0.826, and F1-scores of 0.780 and 0.905, respectively. The results demonstrate that cortical dipoles provide accurate time-frequency representations of MI features, and the unique multi-branch architecture along with its strong time-frequency-spatial-structural feature extraction capability enables effective domain adaptation enhancement in brain source space.

Precise segmentation of gastric cancer computed tomography (CT) images is a critical step for clinical precision diagnosis and treatment. However, it currently faces two major challenges: the low contrast between tumors and surrounding normal tissues makes boundary delineation difficult, and the high variability in tumor shape, size, and location leads to inaccurate localization. To address these issues, a cross-modal prior knowledge-guided gastric cancer CT image automatic segmentation method (CGP-Net) was proposed. In this method, visual priors were extracted from diagnostic reports using a large language model (LLM), and lesion localization was assisted by a semantic anchoring and parsing module. A mixed context-aware Mamba module was constructed to synergistically optimize feature modeling for adapting to tumor morphological variations. Furthermore, a boundary-aware gated convolution module was designed to improve the delineation accuracy of fuzzy boundaries. Experiments on a large-scale dataset of 349 gastric cancer patients demonstrated that the Dice coefficient and 95th percentile of Hausdorff distance (HD95) of the proposed method reached 78.10% and 16.44 mm, respectively. It outperformed state-of-the-art methods such as U-Mamba and nnUNet in terms of segmentation accuracy and boundary prediction. This method effectively integrates textual priors to significantly enhance segmentation accuracy, offering significant value for clinical applications.

Brain-computer interfaces (BCIs) are communication and control systems centered on neural signals that incorporate both the user and the brain into a closed-loop interaction framework, and are widely regarded as a transformative paradigm in human-computer interaction. However, despite the existence of broadly accepted definitions within the research community, the rapid acceleration of BCI translation and commercialization has led to increasing ambiguity in scientific definitions, expansion of conceptual scope, and overstatement of technical capabilities. To address these issues, this paper proposed a scientifically grounded definition of BCIs and systematically analyzed their essential system components and fundamental characteristics. On this basis, the major and specific factors that constrain the capability boundaries of current and foreseeable BCI systems were examined. Furthermore, the scope of BCI was explicitly delineated by distinguishing BCIs from adjacent neurotechnologies based on their functional roles and system characteristics. This work aims to promote a more rigorous and coherent understanding of BCI definitions, scope, and capability limits within the academic community, and to provide essential theoretical foundations for responsible translation and long-term development. By clarifying conceptual boundaries and realistic expectations, it seeks to mitigate risks associated with conceptual generalization and distorted projections in both research and industrial practice, thereby fostering a more rational, robust, and sustainable ecosystem for the BCI field.

Alzheimer's disease (AD) is a common elderly illness, and the hand movement abilities of patients differ from those of normal individuals. Focusing on the utilization of RGB, optical flow, and hand skeleton as tri-modal image information for early AD recognition, a method for early AD recognition via multi-modal hand motion quality assessment (EADR) is proposed. First, a hybrid modality feature encoder incorporating global contextual information was designed to integrate the global contextual information of features from three specific modality branches. Subsequently, a fusion modality feature decoder network incorporating specific modality features was proposed to decode the overlooked information in the fusion modality branch from specific modality features, thereby enhancing feature fusion. Experiments demonstrated that EADR effectively could capture high-quality hand motion features and excelled in hand motion quality assessment tasks, outperforming existing models. Based on this, the action quality scoring regression model trained using the k-nearest neighbors algorithm demonstrated the best recognition performance for AD patients, with Spearman's rank correlation coefficient and Kendall's rank correlation coefficient reaching 90.98% and 83.44%, respectively. This indicates that the assessment of hand motor ability may serve as a potential auxiliary tool for early AD identification.

The adhesion of cancer cells to the vascular endothelium during hematogenous metastasis is a crucial first step, involving the interaction of multiple adhesion molecules between cancer cells and endothelial cells. Here, a parallel-plate flow chamber combined with fluorescence microscopy was used to observe the adhesion behavior and subsequent calcium response of MDA-MB-231 cells on different functionalized substrates under flows, revealing the underlying force-regulation mechanism by analyzing and extracting relevant characteristic parameters. Our results demonstrated that fluid shear stress positively regulated the rolling velocity of cells by affecting the dissociation rate constant of CD44/E-selectin, and rapidly activated integrin α5β1 at the sub-second level, slowing down the rolling velocity of cells, but not enough to firm adhesion. Force triggered the calcium response of MDA-MB-231 cells on E-selectin. Furthermore, the activated integrin α5β1 binding with fibronectin enhanced and quickened cellular calcium response with higher activation ratio and peak intensity, and shorter delay time. This study can deepen the understanding of the hematogenous metastasis process of breast cancer cells, and provide reference for relevant clinical treatment strategies and drug development.