IEEE Transactions on Biomedical Engineering最新文献

Objective: The objective of this study was to investigate the effect of electrical dose on in vivo INSPIRE treatments which administer high voltage ultrashort alternating polarity electrical pulses with active temperature control.

Methods: INSPIRE was administered to healthy swine liver in vivo via a percutaneous single applicator and grounding pad approach. Using 45 °C temperature control, 6000 V waveforms consisting of 750 ns, 1000 ns, or 2000 ns bipolar pulses were administered to examine the effect of pulses approximately shorter than, equal to, and longer than the cell membrane charging time. Treatment volumes were assessed one week post treatment via computed tomography and cardiac safety was assessed via serum troponin analysis.

Results: Pulse duration did not significantly affect treatment volumes, however, dose was found to be a critical factor affecting treatment outcomes. For 0.0025 s doses, treatment volumes of 1.3±0.6 cm³ (2.4 × 0.9 cm) were created in 0.3 minutes. This increased to 12.8±4.8 cm³ (9.7 minutes, 3.9 × 2.5 cm) for 0.04 s doses. No significant changes in troponin levels were found.

Conclusion: This study demonstrated the in vivo safety of high voltage INSPIRE treatments without cardiac synchronization. There is a strong dose dependent effect on treatment volumes. Optimal treatment efficiency was found for treatment doses between 0.01 and 0.02 s with treatment times between 2-4 minutes.

Significance: Single applicator INSPIRE treatments significantly simplify treatment planning and clinical implementation versus traditional two to six applicator approaches. This study demonstrates that INSPIRE protocols can rapidly produce large spherical treatment zones while reducing treatment times by an order of magnitude compared to existing electroporation approaches.

Objective: Disorders of consciousness (DoC) diagnosis critically depends on accurate state discrimination to guide treatment and prognosis. Current EEG-based techniques face challenges of incomplete electrode coverage and manual feature reliance, due to the complex nature of DoC conditions.

Methods: This study proposes DoC-Informer, a CNN-Transformer framework for automated DoC discrimination under adaptive EEG settings. By integrating a channel-independent architecture-enabled by electrode position encoding and spatial transformers-with channel masking training, the framework employs: 1) Shallow Temporal Feature Encoding with parallel temporal convolutions to extract channel-independent temporal features; 2) Spatiotemporal Representation Modeling using a Spatial Transformer (with 3D electrode position encoding) to infer spatial dependencies and a Temporal Transformer for long-range dynamics. A Channel Masking Training Strategy enhances robustness to incomplete data.

Results: Extensive experiments on two real-world DoC datasets (including UWS and MCS patients) demonstrate DoC-Informer's superiority over the cutting-edge deep learning counterparts and a machine learning baseline, with results showing: 1) State-of-the-art performance, 2) Robustness to channel loss, and 3) Validated module efficacy via ablation studies.

Conclusion and significance: DoC-Informer bridges brain science and clinical needs by integrating anatomical priors (electrode coordinates) with adaptive deep learning. Its resilience to variable EEG configurations offers a practical solution for real-world DoC diagnosis, particularly in settings with sparse or incomplete recordings. The source code of the framework is available at https://github.com/pilonglin/docinformer.

Objective: Estimating personalized muscle forces through musculoskeletal modeling is valuable for assessing patient status and monitoring clinical progress. However, this process involves numerous model parameters that are difficult to measure. Upper-limb applications are particularly limited due to the complexity of the system and the long computation times required for model calibration. This study proposes a rapid ($< $5 min) calibration method for upper-limb musculoskeletal models.

Methods: We calibrated maximal isometric force and optimal muscle length for 38 muscles across 10 degrees of freedom by matching muscle-generated moments with dynamically consistent joint moments. The method leverages experimental data including bony landmark trajectories from markerless motion capture, external forces, and electromyography (EMG).

Results: Joint moment estimation and calibration were completed together in less than five minutes. During hand-cycling, the calibrated model reduced EMG tracking error compared to the uncalibrated model (5.58$pm$0.92% vs. 6.30$pm$1.28%). Reliance on non-physiological residual moments was also lowered (12.68 vs. 23.61% of peak moment for calibrated vs. uncalibrated models, respectively).

Conclusion: The proposed method enables rapid calibration of upper-limb muscle parameters, improving accuracy in muscle force estimation and reducing dependence on residual moments.

Significance: This approach provides a fast and reliable framework for upper-limb musculoskeletal calibration, facilitating more accurate and clinically applicable muscle force estimation.

Objective: Endoscopic submucosal dissection (ESD) is a minimally-invasive method for removing gastrointestinal lesions. Adoption has been slow due to the procedure's technical difficulty, length of surgery, and lack of appropriate tools. Tissue traction is required to reduce the technical difficulty of ESD and ensure safe lesion removal. This paper presents the first ESD retraction device that is wireless, endoscope-independent, and provides adjustable traction force control.

Methods: The magnetic retraction device is 3.25 mm wide and 45 mm long. It is actuated wirelessly by a rotating permanent magnet held outside the abdomen. The device design is informed by a presented model of actuation, and we show characterization of the device actuation, validation of mechanical retraction, and clinical feasibility through an ex-vivo experiment.

Results: Experiments show that the device retracts 20 mm at 0.2 millimeters per second and generates a peak retraction force of 7.98 N and a four-peak average of 5.62 N at an external magnet-to-device distance of 55 mm. Ex-vivo testing with a gastric porcine model was performed with a dissection speed of 11.3 square centimeters per hour, 1.25 times faster than the reported international benchmark.

Conclusion: This enhanced tissue control would greatly improve patient clinical outcomes by expanding the use of ESD, shortening procedure times, minimizing complications, and potentially further expand its future medical applications.

Significance: This work grants surgeons improved tissue control throughout an ESD procedure, providing adequate visualization of the dissection plane, control independent of the flexible endoscope, and deployment anywhere along the gastrointestinal tract.

Objective: Magnetic Resonance Elastography (MRE) is a non-invasive imaging technique for mapping biomechanical properties of in vivo tissue, including shear wave speed (SWS), but involves intrinsically slow data acquisition and an ill-posed wave inversion. Instead of relying on handcrafted image priors, we propose a data-driven approach jointly combining image reconstruction and MRE inversion for robust SWS estimation from undersampled k-space data.

Methods: Our physics-informed reconstruction framework comprises two blocks: a model-based neural network (NN)-regularized reconstruction module and a phase-gradient inversion (k-MDEV) calculating SWS from the reconstructed images. Concatenating both blocks yields an end-to-end trainable method to estimate SWS directly from measured k-space data. We evaluated the method on retrospectively highly undersampled brain MRE data and compared it to a total variation (TV) minimization-based approach. We assessed the impact of end-to-end training (qualitative images and SWS maps as targets) versus pre-training (qualitative images as targets) and applied the method also to in vivo data.

Results: Our approach significantly reduces NRMSE by 30% compared to TV. End-to-end training improves SWS estimation over separate image reconstruction and SWS calculation.

Conclusion: Accurate SWS quantification is possible at acceleration factors up to 19. Our method significantly outperforms TV, highlighting the need for data-driven regularization in this challenging MR problem. Further, our approach successfully generalizes to in vivo data.

Significance: We present the first end-to-end trainable MRE reconstruction method for estimating SWS maps directly from k-space. NN-based reconstruction can enable rapid stiffness mapping for dynamic studies, functional imaging, and real-time clinical feedback.

Objective: Human cognitive functions are linked to hidden cognitive states, i.e., arousal and performance. The Yerkes-Dodson law suggests an inverted-U link between these two states, and they may need to be decoded concurrently. However, conventional decoders decode these states separately without including their non-linear interplay.

Methods: We develop a concurrent arousal-performance (CAP) decoder using a Bayesian state-space framework that accounts for their psychological link. The correctness of response and arousal events are binary data to be linked to performance and arousal states, respectively. The reaction time is a continuous observation jointly linked to both states via a quadratic function. We evaluate the framework on simulated data and two experimental datasets. Specifically, data acquired on subjects performing 1-back and 3-back memory tasks during which they are elicited by relaxing, exiting, and AI-generated relaxing music, as well as by smell fragrance and intake coffee are used.

Results: The CAP decoder outperforms the previously developed decoder in reflecting the inverted-quadratic arousal-performance link, suggesting the presence of the Yerkes-Dodson law. The decoded arousal state peaks during an exciting music session, while the decoded performance state is aligned with the task difficulty.

Conclusion: The developed framework reliably decodes the hidden arousal and performance and reveals their link.

Significance: This research advances the safe personalized intervention design.

Objective: The Apnea-Hypopnea Index (AHI) serves as the primary metric for Obstructive Sleep Apnea (OSA) severity but quantifies only event frequency, lacking critical information on the slope and temporal morphology of oxygen desaturation.

Methods: To quantify the desaturation and reoxygenation dynamics during apneic episodes across sleep stages, we develop a geometric modeling framework to extract three groups of event-based features from individual desaturation events: Event Falling Rate (EFR), Event Recovery Rate (ERR), and Event Baseline Shift (EBS). Using 800 polysomnography recordings from the Sleep Heart Health Study, these features are evaluated against and integrated with established oximetry indices using machine learning classifiers.

Results: Spearman correlation analysis reveals significant associations between the proposed event-based features and OSA severity (p < 0.001). The integrated model achieved accuracies of 88% (binary), 76% (three-class), and 72% (four-class). Notably, SHAP interpretability analysis identified EFR and ERR as top-ranking event-based features in OSA stratification, outperforming several traditional global metrics.

Conclusion: The proposed event-based features capture the rate and stability of hypoxic transitions, providing subtle diagnostic information complementary to AHI.

Significance: This framework offers interpretable, physiologically grounded features that enhance OSA stratification and phenotyping using accessible pulse oximetry, facilitating personalized OSA health assessment.

Human gait analysis quantifies locomotion and assesses gait performance, particularly for patients with musculoskeletal disorders. While instrumented 3D gait analysis is the gold standard, advancements in physics based musculoskeletal modeling offer deeper insights into body mechanics. However, its complexity and resource demands limit clinical use, prompting interest in machine learning (ML) as a surrogate for traditional simulations. This scoping review synthesizes ML approaches for estimating joint contact forces in the lower extremities. A systematic search was conducted according to PRISMA-ScR guidelines, covering English language publications from January 2014 to August 2024 across PubMed, IEEE Xplore, Scopus, and SpringerLink. Studies were eligible if they applied ML techniques to estimate lower extremity joint contact forces in human participants and provided sufficient methodological details. Data extraction used a standardized charting form capturing study populations, movement types, input data, ML methods, validation procedures, and performance metrics. 27 studies met the inclusion criteria. The studies showed variability in populations, movement types, input data, ML methods, validation procedures, and performance metrics. Small datasets, often underrepresenting females, limit model generalizability. Inconsistencies in validation approaches and performance metrics, along with the lack of published data and code, hinder reproducibility and comparability. Despite challenges, ML models show potential in accurately predicting joint contact loads and forces. Future research should focus on expanding and diversifying datasets, standardizing methodologies, embracing open science practices, and integrating physics-informed approaches to enhance clinical applicability.

Breast cancer is a malignant disease, and patient prognosis significantly improves when detected at an early stage. Therefore, various advanced chemiresistive sensors have been adopted to detect Volatile Organic Compounds (VOCs), which are byproducts of cellular metabolism exhaled in breath, for early breast cancer detection. In this work, gold nanoflowers (AuNFs) with a high surface area to volume ratio and a face centered cubic (FCC) crystalline structure of 203 $nm$ were synthesized, as confirmed by X-ray diffraction (XRD) and High resolution scanning electron microscopy (HRSEM). After dispersion in deionized (DI) water, the AuNFs were drop coated onto interdigital elliptical aluminum electrodes patterned on glass substrates, forming a continuous film (neighboring AuNFs closely packed) with an initial resistance of up to 2 $mathit {KOmega }$. The AuNFs films were then functionalized with phenylethyl mercaptan and 2-methyl-1-propanethiol using a simple and controllable drop coating method offering an advantage over conventional ligand ion exchange techniques. The large electrode spacing significantly reduces noise compared to traditional low spacing gold electrodes, which require costly photolithography. Furthermore, thiolation enhances both sensitivity and selectivity. The sensors exhibited very high sensitivity, attributed to the high conductivity of the AuNFs films and the sharp petal like active sites promoting strong VOC interactions. To the best of the authors' knowledge, this is the first report demonstrating high sensitivity for breast cancer related VOCs using aluminum electrodes on a glass substrate.

Objective: This paper aims to reduce human-exoskeleton knee joint misalignment arising from the nonuniform and subject-specific motion of the tibiofemoral joint, thereby improving kinematic compatibility between the user and the exoskeleton.

Methods: A bioinspired gear-based knee exoskeleton is proposed, featuring a planetary gear mechanism that approximates the physiological rolling behavior of the human knee and a three-stage gear compensatory transmission that accommodates residual sliding-induced mismatch. A virtual human-exoskeleton interaction model based on sliding misalignment metric was developed to quantify kinematic misalignment and guide gear-parameter optimization. The proposed design was evaluated through numerical simulations, prototype implementation, human-subject experiments, and bench tests to characterize kinematic consistency and drive performance.

Results: Relative to a single-axis knee joint, the proposed mechanism reduced knee joint misalignment by approximately 70%. Torque transmission tests conducted on a mechanically decoupled platform showed close agreement between commanded and measured output torques, with peak deviations of approximately 8%-15% during extension and 5%-13% during flexion. Back-drivability characterization further indicated low passive resistance, with back-driving torque accounting for less than 5% of the rated assistive torque.

Conclusion: The proposed gear-based knee joint enables anatomical motion approximation and internal misalignment accommodation while maintaining reliable torque transmission within a deterministic kinematic structure.

Significance: This work provides a generalizable mechanical and modeling framework for addressing knee joint misalignment in wearable exoskeletons and supports the development of systems with improved human-robot kinematic compatibility.