Biomedical Physics & Engineering Express最新文献

Objectives.Chronic pain affects over two billion people worldwide, significantly reducing quality of life and placing a substantial burden on healthcare systems and society. Neuropathic painoriginates from lesions of the central or peripheral nervous system. Despite pharmacological, surgical and paramedical management, many patients continue experiencing persistent pain. Epidural Spinal Cord Stimulation (SCS) has become an effective alternative treatment for neuropathic pain. That said, SCS efficacy is dependent on many parameters, including optimized spatial targeting based on paresthesia generated by chosen and tuned SCS. In this context, iterative programming designed to optimize targeting and pain relief puts a major burden on health-care professionals. The leveraging of FEM and other computational techniques would enhance understanding of SCS mechanisms, optimize parameter selection, and ultimately improve patient outcomes.Approach. In this work, we present parametrizable computational model that facilitates the study of computed paresthesia. This model used the typical workflow of two-step simulation often employed for electrical stimulation of neural structures. First, the electrical field generated within the spinal cord and its surroundings was computed using the Finite-Element Method (FEM). The effects this electric field had on axons were then assessed with Ordinary Differential Equations (ODEs). The geometry of this model was based on a section of the PAM50 template of the spinal cord and its surroundings. Somatotopy of the spinal cord is explicitly represented by the fiber's trajectories. Aβmyelinated fibers of the dorsal columns and roots are modelled using the McIntyre-Richardson-Grill (MRG) double cable model.Main results.The computational model produced paresthesia maps which generally followed expected projections in terms of lead laterality and rostro-caudal placement in paresthesia. Some interesting effects of rostro-caudal lead placement ata vertebral level were also observed and will be discussed.Significance. The computed paresthesia maps, which can be directly correlated to felt or measured paresthesia maps, represent a step towards clinical validation of in silico computational models of SCS.

Cervical implant fixation is a critical surgical intervention for stabilizing the cervical spine, often necessitated by trauma, degenerative diseases, or spinal deformities. While spinal disc disease has historically been treated with fusion-based procedures, there has been a recent surge of interest in motion-preserving disc arthroplasties. The present study provides a topical narrative review of selected and recent literature on cervical implant fixation techniques, including anterior and posterior approaches, implant materials, biomechanical considerations, and reported clinical outcomes. Traditional fusion-based procedures have long been used to treat cervical disc disease, while recent years have seen increasing interest in motion-preserving techniques such as cervical disc arthroplasty. Developments in implant design and fixation strategies have contributed to improved radiographic and functional results compared with earlier systems, although each technique presents specific benefits and limitations. Cervical implant fixation has evolved into a highly sophisticated discipline that includes anterior, posterior, and motion-preserving techniques for treating a wide range of spinal conditions. This review summarises recent advances, common complications, and emerging trends in cervical fixation, and highlights existing research gaps to support future investigation and clinical decision-making.

Human skin is continuously exposed to varying environmental conditions. While it is known that the skin's biophysical properties are influenced by seasonal changes, the impact of these conditions on its electrical characteristics-particularly skin surface susceptance-remains unclear. Therefore, this study aimed to investigate the effects of seasonal variations on skin surface susceptance using a low-frequency electrical technique. The investigation was performed on 46 (23 males and 23 females) apparently healthy volunteers. Readings of electrical skin surface susceptance were taken from the volar forearm in all four seasons between autumn 2024 and June 2025. Seasonal changes had a significant effect on skin surface susceptance, with the highest values recorded in summer and the lowest in winter. In addition, a highly significant (p< 0.05,r= 0.98) positive correlation was established between seasonal temperature and the skin surface susceptance, and a significant (p< 0.05,r= -0.97) negative correlation between seasonal temperature and the skin surface susceptance was obtained. No significant differences were observed between male and female groups in response to seasonal changes, indicating that gender is an unimportant factor in this context. Our results suggest that seasonal variations should be taken into consideration when using the skin electrical technique. In addition, this will be relevant in the applications of skin sensors and dermatology.

The increasing use of ionizing radiation in interventional cardiology raises the need for reliable estimates of operator exposure, particularly for organs such as the eyes and hands. This retrospective study analyzed personal dosimetry records from interventional cardiologists working in three Italian hospitals. A total of 1,897 valid dosimetry measurements were analyzed across three centres, including whole-body Hp(10), eye lens Hp(3), and extremity Hp(0.07) doses. We derived the following conversion factors by using the third quartile of the ratio distributions from all centres: Hp(3)/Hp(10):1.1 and Hp(0.07)/Hp(10): 2.7. However, our data show high variability across all centres, which probably reflects differences in procedure complexity and operator positioning observed in routine interventional cardiology practice. These findings support the use of whole-body dosimetry as a practical surrogate for organ dose assessment when eye lens or extremity monitoring is unavailable. The proposed conversion factors (Hp(3)/Hp(10) = 1.1 and Hp(0.07)/Hp(10) = 2.7) provide a conservative and field-applicable tool to retrospectively estimate lens and hand doses in cases of incomplete dosimetry.

The integration of artificial intelligence (AI) into breast cancer management presents transformative potential for both diagnosis and treatment planning. This study introduces a resilient AI framework designed to accomplish, from breast MRI images, two critical tasks: (1) accurate and automated segmentation of breast tumors, and (2) T-stage classification of breast cancer in accordance with the 2018 eighth edition of TNM staging system.The dataset comprises sagittal MRI scans utilized for tumor segmentation through a U-Net architecture, which yielded high precision and specificity.Segmented images were utilized as input to a ResNet-50 convolutional neural network, which demonstrated robust classification performance across all T stage categories (T1mi, T1a, T1b, T1c, T2, T3, T4a, T4b, T4c, and T4d), highlighting its high precision, specificity, and F1-scores in accurately distinguishing tumor progression.The T-stage serves as a critical determinant in selecting appropriate treatment modalities, ranging from surgery and chemotherapy to radiotherapy or palliative care, and in estimating prognosis. Our classification results underscore the clinical significance of tumor size progression in early stages (T1mi-T2), where each incremental increase in diameter is associated with poorer outcomes. For advanced categories (T3-T4a-T4d), our model consistently highlighted a uniformly poor prognosis, irrespective of tumor dimensions, reinforcing the pivotal role of anatomical invasion in staging and therapeutic decisions. This AI framework represents a significant advancement in breast cancer automation, enabling more precise staging and fostering improved clinical decision-making and patient outcomes.

Predicting the stages of Alzheimer's disease (AD) is crucial for delaying disease progression and enabling early intervention. A large amount of existing research focuses on the classification of cognitively normal (CN), mild cognitive impairment (MCI), and AD. However, the two subtypes of MCI-stable mild cognitive impairment (sMCI) and progressive mild cognitive impairment (pMCI)-should not be overlooked. Therefore, this study aims to accurately diagnose the disease stage of patients (CN, MCI, or AD) and further distinguish between sMCI and pMCI. In this work, a multi-task classification model based on multi-source feature fusion, termed MTC-MSFFNet, is proposed to accomplish two diagnostic tasks: (1) CN vs. MCI vs. AD, and (2) sMCI vs. pMCI.We select the hippocampus (HIP) and entorhinal cortex (ERC) as feature maps for the three-class task, and the hippocampus (HIP) and gray matter (GM) for the sMCI/pMCI task. The MTC-MSFFNet integrates a multi-source feature fusion module which combining brain structure maps with structural magnetic resonance imaging (sMRI) data, a task-specific weight learning module guided by brain structural information, and dedicated task heads for each classification objective. The proposed method is evaluated on a mixed dataset constructed from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and the Open Access Series of Imaging Studies (OASIS). Experimental results demonstrate that MTC-MSFFNet achieves an average accuracy of 98.09% for CN vs. MCI vs. AD classification and 95.16% for sMCI vs. pMCI classification. These results indicate that the proposed approach has significant potential to assist clinicians in developing targeted and personalized treatment plans.

Glenohumeral instability is a highly prevalent injury characterized by the early appearance of physiological fatigue, which may affect muscle activation patterns. Understanding the variability of electromyographical activity may improve the comprehension of these changes. Therefore, this study aimed to determine the variability of high-density electromyographical activity in the middle deltoid of patients with glenohumeral instability. For this purpose, this study recruited 58 adults who had suffered at least one episode of shoulder dislocation during the year preceding enrollment. These patients had to perform a protocol of maximal (100% of maximum voluntary contraction (MVC) and submaximal (10%, 30%, 50%, and 70% of the MVC) isometric contractions in lateral abduction. To assess the neural control parameters, a grid of 64 electrodes was placed in the middle deltoid, recording the signal by using high-density electromyography. The results show a great variability in the number of identified motor units, with a progressive decrease across contraction levels, with significantly lower motor units at 50%, 70%, and 100% MVC compared with 10% and 30% MVC (one-way ANOVA, F(4,277) = 18.80; p < 0.001). In addition, the firing rate, the pulse rate, and the recruitment time demonstrated a direct relation to the MVC (r = 0.974, r = 0.990, r = 0.922). Moreover, the silhouette value was highly robust (0.85-0.90). Furthermore, this study suggested potential changes in motor unit recruitment behavior and spatial variability of electromyographical activity across the muscle. This study also proposed an identification of potential error sources and practical solutions to enhance this evaluation. In conclusion, high-density electromyography enabled the characterization of neuromuscular patterns in shoulder instability. However, while the findings support its feasibility for research applications, further research is needed to formally establish its validity.

The increasing awareness of stress-related health impacts has driven demand for accurate, non-invasive stress detection methods, particularly those leveraging wearable sensors. While multimodal sensing approaches have shown promise in enhancing mental stress assessment, the critical role of feature selection in optimizing model performance remains underexplored. This study presents a comprehensive machine learning pipeline for mental stress detection that integrates data preprocessing, feature extraction, systematic feature selection, and classification. Using data collected from 17 participants, we classified stress and relaxation states based on three physiological signals: electrodermal activity (EDA), electrocardiography (ECG), and electroencephalography (EEG). Multimodal sensor fusion was compared against unimodal approaches to assess performance improvements. To identify the most informative features and improve model accuracy, we applied four feature selection methods: Analysis of Variance (ANOVA), Chi-squared (Chi2), Kruskal-Wallis (KW), and Minimum Redundancy Maximum Relevance (MRMR). External validation was conducted using the public Stress Recognition in Automobile Drivers (SRAD) dataset. Our results demonstrated a 12.9% increase in classification accuracy using multimodal data, reaching up to 95.9%, with feature selection contributing an average gain of 4.8%. Among the methods, Chi2 consistently achieved the highest mean accuracy across various feature sets. Key biomarkers included ECG-based median, mean, and root-mean-square; EEG-based beta-to-alpha ratio and relative alpha power; and EDA-based mean and sum phasic activity. These findings highlight the importance of integrating systematic feature selection with multimodal sensor data to enhance the accuracy, robustness, and interpretability of mental stress detection systems.

Alzheimer's disease (AD) classification using machine learning has increasingly relied on multimodal inputs such as Magnetic Resonance Imaging (MRI), cognitive assessments, and biological markers. This study evaluates whether integrating these sources enhances predictive performance compared to using them independently. Neural networks were trained on data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) to classify subjects into Cognitively Normal (CN), Mild Cognitive Impairment (MCI), and AD categories using unimodal, bimodal, and trimodal input configurations. Contrary to expectations, multimodal models did not consistently outperform unimodal ones. The highest test accuracy (81%) was achieved by both the cognitive-only and trimodal models, with the former also demonstrating superior class-wise performance. These findings suggest that neuropsychological features may carry greater diagnostic value than imaging or fluid biomarkers, underscoring the importance of more targeted data fusion strategies. Furthermore, the inclusion of biological markers did not significantly improve early MCI detection, likely due to their limited dimensionality and the model's constrained ability to extract meaningful patterns from such inputs.

Efficient and accurate image segmentation models play a vital role in medical image segmentation, however, high computational cost of traditional models limits clinical deployment. Based on pyramid visual transformers and convolutional neural networks, this paper proposes a lightweight Context Contrast Enhancement Network (CCE-Net) that ensures efficient inference and achieves accurate segmentation through the contextual feature synergy mechanism and feature contrast enhancement strategy. The Local Context Fusion Enhancement module is designed to obtain more specific local detail information through cross-layer context fusion and bridge the semantic gap between the encoder and decoder. The Deep Feature Multi-scale Extraction module is proposed to fully extract the comprehensive information about the deepest features in the bottleneck layer of the model and provide more accurate global contextual features for the decoder. The Detail Contrast Enhancement Decoder module is designed to effectively solve the inherent problems of missing image details and blurred edges through adaptive dual-branch feature fusion and frequency-domain contrast enhancement operations. Experiments show that CCE-Net only requires 5.40M parameters and 0.80G FLOPs, and the average Dice coefficients on the Synapse and ACDC datasets are 82.25% and 91.88%, respectively, which are 37%-62% less than the parameters of mainstream models, promoting the transformation of lightweight medical AI models from laboratory research to clinical practice.