Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference最新文献

Walking on compliant surfaces, such as carpets, grass, and soil, presents a unique challenge, particularly for those relying on prosthetic interventions. Ensuring the safety, stability, and fluidity of movement on these surfaces is paramount to prevent falls and related balance issues in this population. This study presents the first attempt to classify and predict surface compliance in individuals with transtibial lower-limb amputations. By integrating electromyographic (EMG), kinematic, and kinetic data, our system effectively distinguishes user intent across varying surface stiffnesses representing diverse real-world terrains. As we demonstrate the algorithm's success within a clinical population, we achieve up to 83% prediction accuracy, attaining comparable results as in previously tested healthy populations. The suggested framework is a critical component for high-level controllers for advanced prostheses and it holds potential for real-time integration, enabling adaptive adjustments to the prosthetic device in response to both user intent and environmental stimuli.

Ultrasound image segmentation is often limited by the scarcity of annotated datasets, especially in resource-constrained clinical settings. To address this issue, we employ BT-UNet, a self-supervised learning framework that combines Barlow Twins (BT) with the UNet architecture, and aim to enhance segmentation performance in low-data conditions. Unlike previous work that trains BT-UNet exclusively on clinical datasets, our approach explores the benefits of pre-training BT-UNet on musculoskeletal phantom ultrasound images, before fine-tuning it on a small set of annotated clinical images. Our results demonstrate that this strategy significantly improves segmentation performance under limited annotated data. Specifically, with only 5% of the labeled clinical dataset, BT-UNet achieves a Dice score of 0.9311, slightly outperforming the standard UNet's 0.9250. However, at an extreme data scarcity level of 1%, BT-UNet maintains a Dice score of 0.7114, whereas UNet drops to 0.2253. These results highlight the potential of self-supervised pre-training on phantom datasets to address data scarcity challenges in medical imaging. By utilizing unlabeled phantom data for representation learning, BT-UNet enhances segmentation accuracy with minimal clinical annotations, offering a promising solution for real-world medical applications where annotated data is limited.Clinical relevance: This study shows that pre-training a self-supervised learning model on musculoskeletal phantom ultrasound images and fine-tuning it with limited clinical data can significantly improve segmentation accuracy, offering a promising solution to reduce reliance on large annotated datasets.

Time-Up-and-Go (TUG) is a commonly used clinical test to evaluate an individual's gait and frailty state. By combining TUG data with other knowledge, e.g., nutrition and daily habits, informed decisions can be made to delay the progression of or alleviate chronic diseases, such as Parkinson's. Scheduling TUG tests in clinics requires assisted transportation and appointment. With the increasingly overloaded healthcare system, recent advances in e-Health provide an alternative solution. Research studies suggest that it is feasible to perform tests at home and automate gait analysis using intelligent software to classify frailty levels in a remote setting. This allows more frequent monitoring, and clinical appointments are made only to patients at higher risk or those in need. However, conducting the TUG test at home comes with challenges. In this paper, we discuss these challenges, e.g., cluttered environment, and propose solutions. In addition, we investigate whether Body Mass Index (BMI) and gender can affect gait measurement. Our experimental results demonstrate that some machine learning models perform better and the choice of input parameters plays an important role in the classification accuracy. Our experimental results demonstrate that high BMI can be reflected in an individual's TUG, if a robust machine learning model is deployed, while men and women in general show distinct gait measurements. Based on this finding, different thresholds should be defined when making the frail, pre-frail and healthy assessment.

Earables, due to their unobtrusive and lightweight nature, are increasingly being recognized for their potential in estimating digital biomarkers, yet their application in gait analysis (GA) remains limited because comprehensive analytic tools are missing. Existing ear-worn systems have primarily addressed isolated aspects such as gait classification, stride time, or step length estimation, lacking a full end-to-end pipeline. Such pipelines are essential for efficient and automated workflows and real-world applications. This work presents a complete end-to-end GA pipeline for ear-worn accelerometers incorporating multiple algorithms to process raw sensor signals into spatio-temporal parameters. This multi-step approach includes gait sequence detection, event identification, and parameter estimation. We introduce a novel gait sequence detector (GSD) that automatically detects regions of interest in continuous recordings. The integrated spatio-temporal algorithms have already been validated in an isolated setting as part of a previous evaluation study. Using a dataset with three walking speeds and foot-worn IMUs as references, the GSD effectively detects 91 % of gait sequences. The pipeline achieves stride time and SL errors of around 4 % and a gait velocity error of 5.7 %, consistent with prior evaluation for the individual isolated steps. To our knowledge, this is the first end-to-end GA pipeline for earables. Furthermore, the pipeline was released as open-source toolbox (https://github.com/mad-lab-fau/eargait), to facilitate research access and reusability. Our work lays the foundation for automated, continuous, and long-term mobility assessment in home environments using lightweight, unobtrusive earables.

Image-text multimodal disease diagnostic models have the potential to provide more precise diagnosis results compared to conventional image-only diagnostic models. Existing image-text multimodal diagnostic methods struggle to address the significant image-text distribution differences and realize comprehensive information interaction between the two modalities. To tackle these issues, we propose a novel foundation model-driven multimodal fusion model, MedFMIT, for image-text disease diagnosis. MedFMIT utilizes the DCA encoders to extract informative and well-aligned visual and textual feature representations, effectively reducing the distributional gap between images and texts while ensuring robust feature extraction. The DMII module is introduced to facilitate comprehensive information interaction between image and text features at both coarse-grained and fine-grained levels. For performance evaluation, we conducted experiments on two multimodal medical classification datasets (image + text) containing computed tomography images and endoscopic optical images. MedFMIT outperforms other state-of-the-art multimodal algorithms, achieving the AUC scores of 92.7% and 85.4% on the two datasets respectively, demonstrating its strong potential for precise medical diagnosis.

The simultaneous activation of agonist and antagonist muscles during limb movement is known as muscle coactivation. The characteristics of muscle coactivation are closely related to the athletic performance of sports athletes, and they have significant implications for guiding athletes in scientific training, reducing muscle fatigue, and preventing sports injuries. The purpose of this study is to investigate the factors that influence muscle coactivation characteristics during elbow joint movements of the upper limbs. We recruited seven professional rock climbers as subjects and used an isokinetic dynamometer along with surface electromyography (sEMG) technology to record the surface EMG of the biceps brachii and triceps brachii during different levels of isokinetic and isotonic movements of the elbow joint. We then analyzed and processed the sEMG signals to calculate the muscle coactivation index for elbow extension and flexion movements. Additionally, we employed a multifactorial repeated measures analysis of variance (ANOVA) to explore the effects of joint movement type, muscle contraction level, and limb laterality. The muscle coactivation index in flexion (40.8% ± 6.3%) was significantly higher than in extension (35.9% ± 12.6%). The results indicated that only joint movement type had a significant main effect on the coactivation index, while the other factors did not demonstrate a significant effect. As rock climbing is a sport that requires balance between the left and right limbs, the coactivation characteristics of the dominant and non-dominant upper limbs show no laterality. This finding suggests that different types of joint movements substantially influence coactivation levels between the biceps brachii and triceps brachii during isokinetic contraction, thereby modulating the synergistic actions of these muscles. This study preliminarily explores the influencing factors of muscle coactivation characteristics in athletes, providing valuable guidance for scientific training.

Bioheat transfer is the study of heat transport applied in anatomy and physiology, and it is a critical tool when analyzing thermal exposures and various treatments and diagnostic methods. Blood flow significantly impacts heat transfer throughout the body, facilitating the diffusion of heat. Several models have been developed to quantify bioheat transfer and the effect of blood flow through tissue for many biological functions and medical procedures, one of which is radiofrequency ablation for cardiac arrhythmia. While some previous studies suggested that the effect of tissue perfusion may be critical only for highly vascularized organs, such as the liver, other studies concluded that the convective effect at the endocardium is a more significant factor than inner tissue perfusion. Nevertheless, significant improvements to models and assumptions are still required, as success rates for this procedure remain low for various arrhythmia types. In the effort to quantitatively assess the impact of considering (or not) the effect of tissue perfusion when modeling thermal ablation, this work focuses on studying the effects of perfusion using a tissue-mimicking phantom both experimentally and numerically. We conducted a parametric study of the flow rate through piping system embedded inside the tissue-mimicking phantom and analyzed the transient thermal profile at different locations and depths in the phantom. This study used a physical experimental setup and its homologous computational fluid dynamics model, with material properties and conditions for the numerical simulations from previous research. The numerical results were compared with the computational results. The findings of this study supported that perfusion impacts the transient thermal profile and that further research is needed to expand this foundation into clinically relevant experimentation.Clinical Relevance- This paper investigates the effect of tissue perfusion in thermal ablation modeling of cardiac tissue.

Medical image segmentation is pivotal for diagnosing and analyzing brain tumors, particularly lower-grade gliomas (LGG). Accurate tumor delineation is critical for clinical decision-making and treatment planning, yet this task remains challenging due to the complex structure of brain tissues and the heterogeneity of tumor characteristics. In this paper, we propose Genomic Attention U-Net (GenAU-net), an enhanced segmentation framework that integrates genomic clustering data into the widely used Attention U-Net architecture. By incorporating patient-specific genomic information, GenAU-net achieves a more personalized approach to LGG MRI segmentation, demonstrating a DICE score of 0.827 on a public LGG dataset. Leveraging genomic data not only improves segmentation performance but also opens avenues for an individualized diagnosis and treatment strategy.Clinical relevance-This research underscores the potential of incorporating genomic information for more accurate LGG segmentation in brain MRI. By providing richer context in the segmentation process, GenAU-net could help clinicians better identify tumor boundaries, optimize surgical resection or radiation therapy plans, and ultimately guide tailored patient care, improving outcomes and survival rates.

Automatic coronary artery segmentation is crucial for computer-aided diagnosis and treatment planning of coronary artery disease (CAD). It helps clinicians identify potential stenotic lesions and formulate treatment plans, thereby improving the efficiency and effectiveness of diagnosis and treatment. However, the complex tree-like tubular structure of the coronary artery makes it challenging to accurately identify small branches, leading to incomplete topology. This paper proposes a topology-guided progressive refinement network (TPRNet) that progresses from global to focal perspective, leveraging the anatomical topology of the coronary artery to accurately identify small branches and reconstruct vascular structure. Specifically, the globalnet branch performs global segmentation to capture the spatial location information of the coronary artery in the image; the localnet branch segments local vessel regions based on location information and extracts vascular topology; the focalnet branch performs fine-grained segmentation along the centerline to capture vascular details; and finally, the refinement branch reconstructs and optimizes the topology. Experiments show that TPRNet outperforms existing methods on the public coronary artery segmentation dataset ARCADE. The code is available at https://github.com/IPMINWU/TPRNet.

Brain-Computer interfaces can assist motor rehabilitation for people with severe paralysis by directly decoding their brain signals into movement intention and executing with external devices without passing the impaired neural pathways. It is crucial to restore natural and smooth daily movements, and continuous force control is one of the most important kinaesthetic functions. However, the complex continuous force decoding and limited relevant public datasets greatly challenge this field. How the brain coordinates the motor command or sensory feedback during the force control behaviour also remains to be discussed. This work investigated these questions through a novel experimental setup by isolating the motor intention and sensory feedback and combining both components flexibly for hand grip. We applied functional electrical stimulation to induce passive gripping and collected grip force with multi-modal brain signals. Significant neural pattern differences were found in EEG time-frequency representation by comparing the brain responses under different task conditions, including voluntary movement, motor imagery, and passive perception status. Additionally, we present a multi-modal graph fusion model fusing both EEG and fNIRS for continuous bimanual grip force decoding. These contributions are beneficial to developing neural interfaces for rehabilitation and assistive devices that involve force manipulation or operate in isometric schemes.