IEEE Transactions on Biomedical Engineering最新文献

Objective: To construct a personalizable spatio-temporal disease progression model of patients with a late dry form of Age-Related Macular Degeneration (AMD), known as Geographic Atrophy (GA).

Methods: From a series of retinal optical coherence tomography (OCT) scans, we infer the coefficients for the parametrized partial differential equation (PDE), such that the parametrized PDE best describes the observed imaging data. Acting as a soft constraint, the recovered PDE helps to extrapolate an implicit neural representation (INR) of the GA segmentation map progression. To enable efficient training, we propose an iterative method - PINN-EM, designed to recover coefficients of non-linear PDEs. At each iteration, the method decouples the problem into PDE coefficients fitting and data fitting steps, resembling Expectation Maximization algorithm.

Results: We extensively tested the proposed method in large-scale experiments using the open-source PDEBench benchmark to validate its performance. Furthermore, we applied the method to the challenging problem of GA progression modeling, where patients exhibit a high variance in GA growth patterns and speed. The proposed spatio-temporal disease progression model outperformed the baselines, even outperforming posterior knowledge models in Dice score for newly affected growth areas.

Conclusion: We demonstrated that the proposed spatio-temporal disease progression model fitted with introduced PINN-EM outperforms existing baselines in synthetic and real clinical applications, highlighting the extrapolation capabilities of the INR models.

Significance: The proposed spatio-temporal disease progression model and PINN-EM fitting procedure can be applied across diverse domains facing the challenge of fitting parametrized PDE to the empirical datasets.

Objective: Total joint arthroplasty (TJA) effectively treats end-stage hip and knee joint diseases, improving patients' quality of life. However, 1-2% of TJA patients develop prosthetic joint infections (PJI), which are challenging to diagnose and treat. This study investigates non-invasive active vibration and passive acoustic emission analyses for PJI monitoring.

Methods: In this ex vivo study, periprosthetic joint effusions were simulated in seven cadaveric specimens with knee replacements by injecting saline and bacterial solutions into the joint space. Active sensing involved non-invasively stimulating the tibia with a miniature shaker, while passive sensing used manual stress to induce vibrations. Wideband, low-noise accelerometers captured the resulting vibrations, with spectral and temporal features extracted from the active and passive recordings, respectively. A qualitative analytical beam model of the knee-tibia system was developed to represent the fluid as structural changes at the boundary of the system.

Results: Both methods proved to be sensitive to the fluid in the joint space. Linear regression models were built using the most informative features, estimating fluid volume with Pearson's r of 0.79 and mean absolute errors of 11.1 mL (active) and 11.9 mL (passive). Trends in frequency and time signals were consistent between the experimental results and the analytical model.

Conclusion: The results of this study demonstrated the utility of novel vibration-based techniques to monitor periprosthetic joint effusions.

Significance: These non-invasive techniques can lead to wearable devices for joint health monitoring, enabling PJI detection and personalized treatment plans, potentially improving patient outcomes and reducing PJI-related healthcare costs.

Multi-task decoding from electroencephalogram (EEG) signals is valuable for brain-computer interface (BCI) applications in naturalistic settings. Most existing studies focus on decoding distinctly different tasks, leaving the diversity of cognitive responses elicited by a single stimulus underexplored. We introduced a novel experimental paradigm where a common visual stimulus elicits five distinct cognitive processes: single reach, interception reach, sequence reach, attention reach, and inhibition reach. EEG signatures were analyzed using temporal and spectral methods. A regularized linear discriminant analysis (RLDA) classifier was employed for decoding, utilizing both temporal and event-related spectral perturbation (ERSP) features. Significant neural activation differences (p < 0.05) were observed across tasks and brain regions. The RLDA classifier achieved high decoding accuracy: 91.72% ± 6.10% for classifying the five cognitive states using ERSP features. Furthermore, for the sequence reach task, temporal features enabled classification of normal versus catch trials with 77.96% ± 7.03% accuracy. All these results demonstrate the potential for EEG-based BCI applications to distinguish diverse cognitive states elicited by identical stimuli, offering new insights for improving the naturalness and intelligence of BCI systems. Future work will focus on enhancing decoding performance and extending this research to online applications.

Approximately 30% of stage II and 70% of stage III colorectal cancer (CRC) patients suffer postoperative recurrence owing to delayed diagnosis. However, Existing diagnostic ap proaches, particularly electrochemical biosensors, face challenges including poor diagnostic accuracy in single-biomarker analysis, cross-reactivity in multiplex detection, and the absence of long-term home monitoring device. This study introduced a portable point-of-care testing (POCT) biosensing system integrating a 256-channel microelectrode array (MEA) sensor for at-home detection of CRC recurrence biomarkers-carcinoembryonic antigen (CEA), microRNA-21 (miR-21), and interleukin-6 (IL-6). These sensors exhibited ultra-high sensitivity of 21.47 nA/lg(ng/ml), 26.50 nA/lg(pM), 15.72 nA/lg(ng/ml) for CEA, miR-21, and IL-6. Clinical validation showed strong concordance with conventional assays (ELISA and RT-qPCR), with correlation coefficients of 0.9809, 0.9998, and 0.9950.

Objective: This study proposes a novel sterilization strategy that integrates ultraviolet (UV) light, blue light (400 nm), and riboflavin-5'-phosphate (FMN) to achieve effective microbial inactivation while reducing UV energy usage and minimizing material degradation caused by prolonged UV exposure. Mechanistic analysis revealed that UV irradiation induces cyclobutane pyrimidine dimer (CPD) formation, resulting in DNA damage, whereas blue light in combination with FMN generates reactive oxygen species (ROS), which disrupt microbial cellular structures. However, blue light also activates CPD photolyase, facilitating CPD repair and thereby potentially diminishing sterilization efficacy. Importantly, the irradiation sequence was found to be critical: applying blue light prior to UV exposure (B→U) led to greater DNA destabilization, promoting higher CPD accumulation and enhanced microbial inactivation. This synergistic effect enabled a significant reduction in the required UV energy, which in turn delayed the aging of sterilized materials. The findings offer valuable insights into the design of advanced sterilization solutions for medical instruments, medical devices, and household products that incorporate materials typically susceptible to UV-induced aging and degradation, providing a balanced approach to sterilization efficacy and material preservation.

Objective: Identifying the first (S1) and second (S2) heart sounds from phonocardiogram (PCG) signals is an essential step in automating the diagnosis of cardiac conditions such as irregular heartbeat, valve misfunctions, and heart failure. Recent research inspired by image segmentation has shown promise in utilising deep neural networks for point-wise PCG segmentation with the support of synchronised electrocardiograms (ECG). This paper shifts the focus from point-wise segmentation to identifying the onset/offset of S1 and S2 in the PCG signal.

Methods: We incorporate the ECG signal and its keypoints to improve the detection of the heart sounds. Our proposed method employs a joint classifier-regressor architecture for predicting the probability and the location of onset/offset in the PCG.

Results: When evaluated on the largest publicly available PhysioNet/CinC 2016 dataset, the proposed approach outperforms existing state-of-the-art methods, achieving a sensitivity of 0.97 and a positive predictive value of 0.98 in identifying midpoints of S1 and S2 segments. It also identifies the onset/offset locations with an 11.11 ms error.

Conclusion: It is evident that identifying the transitions simplifies, leading to better training and inference.

Significance: In addition to achieving state-of-the-art results, this proposed approach could also be adapted for locating regions of interest in other physiological signals, such as respiration, blood pressure, or muscle activity.

We present a fully open-source, air-driven bidirectional, rotational MRI phantom. It enables an accurate and reproducible evaluation of displacement artefacts for any MRI sequence and velocity field and acceleration sensitivity for phase-contrast MRI (PC-MRI) sequences. Its unique feature of analytically defined motion is expected to narrow the gap between simulations and experiments for in-silico and in-vitro experiments using the very same sequence code.

Methods: A rotational phantom was bidirectionally driven (clockwise (CW) / counterclockwise (CCW)) by an actively controlled airflow. The rotating cylinder filled with a Polyvinylpyrrolidone-water mixture was monitored via an external laser-based tachometer system. Vendor-supplied and custom open-source PC-MRI sequences were evaluated on a 3T MRI system and used as input for Bloch simulations. Resulting magnitude and velocity images were evaluated against the phantom's ground truth data.

Results: For physiological angular velocities, displacement errors resulted in a 10% radial stretch while apparent acceleration sensitivity is 5% of v_enc. The time difference between velocity and spatial encoding time points of 1.9ms determining the severity of the artefacts could be quantified without prior knowledge of details about the MR sequence. Simulation and experiment yielded excellent agreement.

Conclusion: The phantom enables an easy, precise and repeatable evaluation of motion sensitivity of MR sequences and may offer a future reference measurement. Additional timing parameters of MR sequences may be reported in future literature to improve comparability. The seamless MRI sequence definition for in-silico and in-vitro experiments narrows a significant gap in MR research.

Significance: This work establishes a reproducible, standardized validation framework for PC-MRI techniques that can be readily implemented across institutions, facilitating quality assurance procedures and supporting the development of more accurate flow quantification methods in clinical applications.

Objective: To develop a real-time method for designing gradient waveforms for arbitrary k-space trajectories that are time-optimal and hardware-compliant.

Methods: The gradient waveform is solved recursively under both the slew-rate and the trajectory constraints, which form a quadratic equation. The gradient constraint is enforced by thresholding the L2-norm of the gradient vectors. To ensure the existence of the solution, gradient magnitude is thresholded by the escape velocity. A Discrete-Time Forward and Backward Sweep strategy is then applied to further constrain the slew-rate. Trajectory and gradient reparameterization strategies are adopted to enhance the generality and preserve the sampling accuracy. The proposed method is compared with the conventional optimal control method across seven commonly adopted non-Cartesian trajectories. Imaging feasibility of the designed time-optimal gradient waveform was demonstrated by phantom and in vivo imaging experiments.

Results: The proposed method achieves a >89% reduction in computation time and a >98% reduction in slew-rate error simultaneously. The computation time of the proposed method is shorter than the gradient duration for all tested cases, validating the real-time capability of the proposed method.

Conclusions: The proposed method enables real-time and hardware-compliant gradient waveform design, achieving significant reductions in computation time and slew-rate overshoot compared to the previous method.

Significance: This is the first method achieving real-time gradient waveform design for arbitrary k-space trajectories.

Objective: The study seeks to determine whether a powered, cable-driven exosuit has the potential to lower the lumbar muscle activity and overall metabolic expenditure of symmetric and asymmetric lifting tasks.

Methods: A lightweight, cable-driven back exosuit, using a three-state impedance controller, was developed to provide variable assistance based on user posture. Experimental electromyography (EMG), metabolic cost, and user preference data were recorded for ten participants evaluated wearing the powered back exosuit versus the backX, a commercially available passive back support exoskeleton, and a no exo baseline.

Results: Both exoskeletons significantly reduced (p$< $0.05) muscle activation of certain lumbar flexor and extensor muscles when compared to a no exo condition across all conditions tested, though neither significantly reduced the metabolic cost associated with lifting. Users tended to prefer lifting with the powered device as opposed to the passive or no exo condition.

Conclusion: Despite the increased mass of powered back support exoskeletons, these devices can reduce lumbar muscle activity to a similar degree as passive exoskeletons, and are favored by users over their passive counterparts.

Significance: While current powered back support devices tend to incur the cost of being heavy, rigid, and inconvenient for certain lifting postures, these results show that cable-driven powered devices may minimize these factors to the point that they are favored over the currently popular passive devices on the market.

Brain-computer interface (BCI) technology has significant applications in neuro rehabilitation and motor function restoration, especially for patients with stroke or spinal cord injury. Motor imagery electroencephalog-raphy (MI-EEG) is widely used in BCIs, but its nonlinear dynamics and inter-subject variability limit decoding accuracy. In this paper, a multiscale hybrid attention network (MSHANet) for MI-EEG decoding, which consists of spatiotemporal feature extraction (STFE), talking head self-attention (THSA), dynamic squeeze-and-excitation attention (DSEA), and a temporal convolutional network (TCN), is proposed. MSHANet was evaluated via within-subject experiments using BCI Competition IV Datasets 2a and 2b, as well as EEGMMID, achieving decoding accuracies of 83.56%, 89.75%, and 75.66%, respectively. In cross-subject experiments on the three datasets, the mode lattained accuracies of 69.93% on BCI-2a, 81.85% on BCI-2b, and 79.67% on EEGMMID. In addition, we propose an electrode spatial structure-aware encoder. This technique encodes the spatial positions of electrodes in the original data, enabling the model to obtain richer spatial electrode information at the input stage. In within-subject and cross-subject tasks on BCI-2a, this encoding improved the decoding performance by 2.83% and 2.91%, respectively. Visualization was also employed to elucidate feature distributions and the effec tiveness of its attention mechanisms. Experimental results demonstrate that MSHANet performs exceptionally well in MI-EEG decoding tasks and has high potential for clinical applications, particularly in neurorehabilitation and motor function reconstruction.