生物医学工程学杂志最新文献

Cerebral hemorrhage is a serious cerebrovascular disease with high morbidity and high mortality, for which timely diagnosis and treatment are crucial. Electrical impedance tomography (EIT) is a functional imaging technique which is able to detect abnormal changes of electrical property of the brain tissue at the early stage of the disease. However, irregular multi-layer structure and different conductivity properties of each layer affect image reconstruction of the brain EIT, resulting in low reconstruction quality. To solve this problem, an image reconstruction method based on an improved densely-connected fully convolutional neural network is proposed in this paper. On the basis of constructing a three-layer cerebral model that approximates the real structure of the human head, the nonlinear mapping between the boundary voltage and the conductivity change is determined by network training, which avoids the error caused by the traditional sensitivity matrix method used for solving inverse problem. The proposed method is also evaluated under the conditions with or without noise, as well as with brain model change. The numerical simulation and phantom experimental results show that conductivity distribution of cerebral hemorrhage can be accurately reconstructed with the proposed method, providing a reliable basis for the diagnosis and treatment of cerebral hemorrhage. Also, it promotes the application of EIT in the diagnosis of brain diseases.

During long-term electrocardiogram (ECG) monitoring, various types of noise inevitably become mixed with the signal, potentially hindering doctors' ability to accurately assess and interpret patient data. Therefore, evaluating the quality of ECG signals before conducting analysis and diagnosis is crucial. This paper addresses the limitations of existing ECG signal quality assessment methods, particularly their insufficient focus on the 12-lead multi-scale correlation. We propose a novel ECG signal quality assessment method that integrates a convolutional neural network (CNN) with a squeeze and excitation residual network (SE-ResNet). This approach not only captures both local and global features of ECG time series but also emphasizes the spatial correlation among ECG signals. Testing on a public dataset demonstrated that our method achieved an accuracy of 99.5%, sensitivity of 98.5%, and specificity of 99.6%. Compared with other methods, our technique significantly enhances the accuracy of ECG signal quality assessment by leveraging inter-lead correlation information, which is expected to advance the development of intelligent ECG monitoring and diagnostic technology.

Brain-computer interface (BCI) is a revolutionizing technology that disrupts traditional human-computer interaction by establishing direct communication and control between the brain and computer, bypassing the peripheral nervous and muscular systems. With the rapid advancement of BCI technology, growing application demands, and an increasing need for specialized BCI professionals, a new academic major-BCI major-has gradually emerged. However, few studies to date have discussed the interdisciplinary nature and training framework of this emerging major. To address this gap, this paper first introduced the application demands of BCI, including the demand for BCI technology in both medical and non-medical fields. The paper also described the interdisciplinary nature of the BCI major and the urgent need for specialized professionals in this field. Subsequently, a training program of the BCI major was presented, with careful consideration of the multidisciplinary nature of BCI research and development, along with recommendations for curriculum structure and credit distribution. Additionally, the facing challenges of the construction of the BCI major were analyzed, and suggested strategies for addressing these challenges were offered. Finally, the future of the BCI major was envisioned. It is hoped that this paper will provide valuable reference for the development and construction of the BCI major.

Neuromuscular electrical stimulation (NMES) has been proven to promote human balance, but research on its impact on motor ability mainly focuses on external physical analysis, with little analysis on the intrinsic neural regulatory mechanisms. This study, for the first time, investigated the effects of NMES on cortical activity and cortico-muscular functional coupling (CMFC) during standing balance. Twelve healthy subjects were recruited in bilateral NMES training, with each session consisting of 60 electrically induced isometric contractions. Electroencephalogram (EEG) signals, electromyogram (EMG) signals, and center of pressure (COP) signals of the foot sole were collected before stimulation, two weeks after stimulation, and four weeks after stimulation while the subjects maintained standing balance. The results showed that NMES training improved subjects' postural stability during standing balance. Additionally, based on the EMG power spectral density (PSD), the κ frequency band was defined, and EEG-EMG time-frequency maximal information coefficients (TFMIC) were calculated. It was found that NMES enhanced functional connectivity between the cortex and lower limb muscles, with varying degrees of increase in β-κ and γ-κ frequency band CMFC after stimulation. Furthermore, sample entropy (SE) of EEG signals also increased after training. The results of this study confirm that NMES training can enhance CMFC and brain activation during standing balance. This study, from the perspective of physiological electrical signals, validates the effectiveness of NMES for balance training and provides objective assessment metrics for the training effects of NMES.

Currently, the types of kidney stones before surgery are mainly identified by human beings, which directly leads to the problems of low classification accuracy and inconsistent diagnostic results due to the reliance on human knowledge. To address this issue, this paper proposes a framework for identifying types of kidney stones based on the combination of radiomics and deep learning, aiming to achieve automated preoperative classification of kidney stones with high accuracy. Firstly, radiomics methods are employed to extract radiomics features released from the shallow layers of a three-dimensional (3D) convolutional neural network, which are then fused with the deep features of the convolutional neural network. Subsequently, the fused features are subjected to regularization, least absolute shrinkage and selection operator (LASSO) processing. Finally, a light gradient boosting machine (LightGBM) is utilized for the identification of infectious and non-infectious kidney stones. The experimental results indicate that the proposed framework achieves an accuracy rate of 84.5% for preoperative identification of kidney stone types. This framework can effectively distinguish between infectious and non-infectious kidney stones, providing valuable assistance in the formulation of preoperative treatment plans and the rehabilitation of patients after surgery.

With the high incidence of neurological diseases such as stroke and mental illness, rehabilitation treatments for neurological disorders have received widespread attention. Electroencephalography (EEG) technology, despite its excellent temporal resolution, has historically been limited in application due to its insufficient spatial resolution, and is mainly confined to preoperative assessment, intraoperative monitoring, and epilepsy detection. However, traditional constraints of EEG technology are being overcome with the popularization of EEG technology with high-density over 64-lead, the application of innovative analysis techniques and the integration of multimodal techniques, which are significantly broadening its applications in clinical settings. These advancements have not only reinforced the irreplaceable role of EEG technology in neurorehabilitation assessment, but also expanded its therapeutic potential through its combined use with technologies such as transcranial magnetic stimulation, transcranial electrical stimulation and brain-computer interfaces. This article reviewed the applications, advancements, and future prospects of EEG technology in neurorehabilitation assessment and treatment. Advancements in technology and interdisciplinary collaboration are expected to drive new applications and innovations in EEG technology within the neurorehabilitation field, providing patients with more precise and personalized rehabilitation strategies.

To address the issue of a large number of network parameters and substantial floating-point operations in deep learning networks applied to image segmentation for cardiac magnetic resonance imaging (MRI), this paper proposes a lightweight dilated parallel convolution U-Net (DPU-Net) to decrease the quantity of network parameters and the number of floating-point operations. Additionally, a multi-scale adaptation vector knowledge distillation (MAVKD) training strategy is employed to extract latent knowledge from the teacher network, thereby enhancing the segmentation accuracy of DPU-Net. The proposed network adopts a distinctive way of convolutional channel variation to reduce the number of parameters and combines with residual blocks and dilated convolutions to alleviate the gradient explosion problem and spatial information loss that might be caused by the reduction of parameters. The research findings indicate that this network has achieved considerable improvements in reducing the number of parameters and enhancing the efficiency of floating-point operations. When applying this network to the public dataset of the automatic cardiac diagnosis challenge (ACDC), the dice coefficient reaches 91.26%. The research results validate the effectiveness of the proposed lightweight network and knowledge distillation strategy, providing a reliable lightweighting idea for deep learning in the field of medical image segmentation.

Diabetes and its complications pose a serious threat to human life and health. It has become a public health problem of wide concern worldwide. Currently, diabetes is mainly treated with insulin injection in clinic. However, manual insulin injection still has many shortcomings. In recent years, with the deepening of research, it has been found that an automated insulin delivery system (AID), which combines a continuous glucose monitoring device with an insulin pump, can significantly improve the effectiveness of diabetes treatment and reduce the incidence of complications in patients. This paper firstly introduces the composition of the AID system and its working principle, and then details the development history and current status of the related technologies from the aspects of continuous glucose monitoring technology, insulin pumps and the development of closed-loop control algorithms, etc. Finally, this paper looks forward to the application prospect and future development of AID system in the field of diabetes treatment, providing theoretical reference for further research.

The transmission and interaction of neural information between the hippocampus and the prefrontal cortex play an important role in learning and memory. However, the specific effects of learning memory-related tasks on the connectivity characteristics between these two brain regions remain inadequately understood. This study employed in vivo microelectrode recording to obtain local field potentials (LFPs) from the ventral hippocampus (vHPC) and medial prefrontal cortex (mPFC) in eight rats during the performance of a T-maze task, assessed both before and after task learning. Additionally, dynamic causal modeling (DCM) was utilized to analyze alterations in causal connectivity between the vHPC and the mPFC during memory task execution pre- and post-learning. Results indicated the presence of forward connections from vHPC to mPFC and backward connections from mPFC to vHPC during the T-maze task. Moreover, the forward connection between these brain regions was slightly enhanced after task learning, whereas the backward connection was diminished. These changes in connectivity corresponded with the observed trends when the rats correctly performed the T-maze task. In conclusion, this study may facilitate future investigations into the underlying mechanisms of learning and memory from the perspective of connectivity characteristics between distinct brain regions.

Magnetic resonance imaging (MRI)-based electroencephalography (EEG) forward modeling method has become prevalent in the field of EEG. However, due to the inability to obtain clear images of an infant's fontanel through MRI, the fontanelle information is often lacking in the EEG forward model, which affects accuracy of modeling in infants. To address this issue, we propose a novel method to achieve fontanel compensation for infant EEG forward modeling method. First, we employed imaging segmentation and meshing to the head MRIs, creating a fontanel-free model. Second, a projection-based surface reconstruction method was proposed, which utilized priori information on fontanel morphology and the fontanel-free head model to reconstruct the two-dimensional measured fontanel into a three-dimensional fontanel model to achieve fontanel-compensation modeling. Finally, we calculated a fontanel compensation-based EEG forward model for infants based on this model. Simulation results, based on a real head model, demonstrated that the compensation of fontanel had a potential to improve EEG forward modeling accuracy, particularly for the sources beneath the fontanel (relative difference measure larger than 0.05). Additional experimental results revealed that the uncertainty of the infant's skull conductivity had the widest impact range on the neural sources, and the absence of fontanel had the strongest impact on the neural sources below the fontanel. Overall, the proposed fontanel-compensated method showcases the potential to improve the modeling accuracy of EEG forward problem without relying on computed tomography (CT) acquisition, which is more in line with the requirements of practical application scenarios.