Cardiovascular Engineering and Technology最新文献

Premature ventricular contraction (PVC) is a common cardiac arrhythmia, and its timely and automated detection is crucial for preventing life-threatening cardiovascular events and reducing clinical workload in long-term monitoring. This paper reviews state-of-the-art PVC detection methods based on electrocardiogram (ECG) and photoplethysmogram (PPG) signals, employing signal processing techniques, traditional machine learning, and deep learning approaches. The existing methods are categorized into three groups: threshold-based or heuristic-based techniques, conventional machine learning models, and deep learning frameworks. Additionally, the paper provides an overview of ECG and PPG signal databases and the benchmark metrics used for performance evaluation. Given that most methods utilize R-peak and systolic peak detection during preprocessing, we also review various preprocessing techniques for detecting R-peaks in ECG signals and systolic peaks in PPG signals. Based on the performance and key contributions of existing PVC detection methods, we highlight major challenges and future directions, considering the presence of various noise sources in ECG and PPG signals-particularly under ambulatory and exercise conditions-and the resource constraints of wearable or portable long-term ECG and/or PPG monitoring devices.

Purpose: Cardiopulmonary bypass (CPB) is a well-established procedure that uses cannulae during cardiac surgery to drain and return blood. In challenging cases (e.g. aortic dissection, reoperation), peripheral cannulation in vessels such as the axillary artery are used. However, standard cannulae at these sites may inhibit blood flow to distal extremities or require grafts that increase surgical time. This study evaluates a novel, flexible-tip arterial cannula designed for bi-directional flow.

Methods: A 22Fr body cannula was developed to achieve a pressure loss (ΔP) < 100 mmHg and 80/20 bi-directional flow distribution between tip outlets. A full factorial design of experiments (2-levels, 4 factors: tip A-width, B-height, C-depth, D-outlet shape) was completed to evaluate sixteen cannula tip designs using benchtop hydraulic and bi-directional flow models. Prototypes were fabricated using 3D printing via stereolithography (Form 3 + , Formlabs, Somerville, MA) with a flexible (50A) cured resin.

Results: The most efficient cannula design (A - , B + , C + , D + ) achieved 4 L/min of total bi-directional flow at a ΔP of 74 mmHg. The average ΔP at 4 L/min for all candidate design was 87 ± 9 mmHg (range 72-100 mmHg). Primary outlet flow distribution averaged 81% ± 6% (range 70-95%) at 1-5 L/min flow rates. Tip width had the greatest influence on ΔP, followed by outlet shape, depth, and their interactions, respectively. Cylindrical and prolate spheroid shaped tips outperformed spherical designs.

Conclusion: The novel cannula demonstrated feasibility and proof-of-concept as evidenced by bi-directional flow with ΔP comparable to commercial cannulae. Future work will involve CFD modeling and pre-clinical validation (e.g. hemolysis, cadaver fit) to support development of a low-cost, clinical grade bi-directional flow cannula for peripheral CPB to reduce surgical complexity and lower risk of adverse events.

Purpose: Inefficient energy delivery to blood remains a primary challenge in cardiac radiofrequency ablation (RFA), limiting procedural efficacy. This paper introduces and computationally validates a novel multi-surface microelectrode catheter (MSMC) designed to enhance targeted energy delivery and improve overall procedural efficiency.

Methods: A 3D multiphysics computational model coupling electrical, thermal, fluid-dynamic, and mechanical fields was developed to simulate RFA. The performance of the MSMC was systematically compared against a traditional catheter by analyzing its energy distribution and thermal lesion characteristics under both standard (10-18 W vs. 30 W) and high-power short-duration (25 W vs. 60 W) protocols, assessing the impact of varying catheter angles (vertical, 45°, and parallel), and exploring its potential for real-time lesion monitoring via impedance analysis.

Results: The MSMC directed over 75% of its energy to the myocardium, a threefold improvement over the traditional catheter (~22%), allowing the creation of comparable lesions with 40% less power. The design demonstrated high stability across different orientations. Furthermore, analysis of its impedance characteristics via Cole-Cole plots revealed a greater sensitivity for real-time lesion monitoring compared to the traditional catheter.

Conclusions: The MSMC's design, which synergizes a multi-surface electrode structure with a contact-based discharge strategy, enables more efficient and predictable lesion formation. This computational proof-of-concept study confirms its potential to improve the safety, efficacy, and real-time control of RFA procedures, offering a promising pathway for the development of next-generation therapeutic devices.

Background: Medical image segmentation using artificial intelligence (AI) is a prominent area of research with diverse applications across various fields. During the last years, a multitude of datasets representing different body structures have been developed and made publicly available. However, the volume of data-particularly the ground truth data, which often relies on manual annotation-remains limited. Supervised learning remains the state-of-the-art approach for deep learning methods; however, its performance is often reported as dependent on the expertise of the operator for the ground truth generation. This dependency becomes more critical when dealing with challenging medical imaging modalities, such as ultrasound, often characterized by low image quality and various artifacts.

Methods: This study aims to investigate the influence of user expertise on the accuracy of ground truth annotations and their impact on the final performance of the segmentation method. Specifically, we focus on the task of segmenting the left atrial appendage (LAA) in ultrasound images. Two datasets were initially created: one annotated by an Expert and the other by a novice observer. Additionally, synthetic variations of these manually annotated datasets were generated by introducing both systematic and non-systematic errors to examine their effects on segmentation outcomes.

Results: Using the nnU-Net framework as the computational basis, the network was trained on each dataset, and the results were evaluated against the Expert's test labels. Training with Expert and Naive contours achieved Dice values in the test set of 0.81 ± 0.09 and 0.77 ± 0.12, respectively, with no statistically significant differences between them. Similarly, training with synthetic variations obtained showed no statistically significant differences for non-systematic errors, whereas systematic errors result in statistically significant differences against manual contours.

Conclusions: These findings demonstrate that the AI network remains highly effective across most tested scenarios, even when synthetic errors are introduced, showcasing its ability to handle non-systematic errors efficiently, which synthetically mimic the variability between observers. However, the network encounters greater challenges with systematic errors, failing to accurately delineate the LAA boundaries.

Purpose: Tricuspid valve chordae tendineae rupture is a valvular lesion that is often overlooked, though is postulated to be more prevalent than currently known. We examined the hemodynamics and biomechanical response of the tricuspid valve leaflets following chordae rupture to understand how acute changes in the post-rupture mechanical environment may contribute to long-term remodeling responses.

Methods: Porcine valve leaflet deformation was studied in an intact heart in an ex vivo setup using sonomicrometry techniques before and after chordae rupture, which was induced by severing a chordae bundle connected to the septal leaflet.

Results: Following chordae rupture, pulmonary artery pressure dropped approximately 5 mmHg ( $p=0.048$ ), indicating that valvular regurgitation occurred immediately after rupture. Mean maximum principal stretch of the septal leaflet increased 12% after rupture ( $p=0.006$ ).

Conclusion: The immediate changes in post-rupture septal leaflet stretches show that chordae tendineae rupture acutely alters the biomechanical environment of the tricuspid valve, which may result in chronic tissue remodeling responses. The tricuspid valve is one of the four valves in the heart. Rupture of supporting structures of the tricuspid valve leaflet, known as chordae tendineae, may be more common than previously thought. In this study, we used excised pig hearts to examine how chordae rupture affects valvular function. With our experimental beating heart system, we pumped fluid through the hearts under realistic conditions and measured changes in pressure and leaflet motion before and after chordae rupture. After rupture, we observed a change in pressures and leaflet motion, causing the valve to leak and become less efficient. These changes may influence how the valve functions over time.

Objective: Electrocardiogram (ECG) signals are well-known non-stationary heart signals of lower strength. Due to their small amplitude, it attracts other biomedical artefacts from the surrounding. This research mainly focuses on removing artefacts from the electrocardiogram signals.

Method: The work presented uses the recent metaheuristic techniques to design a nonlinear adaptive Hammerstein filter-based structure efficiently. Many powerful metaheuristic optimisation algorithms, such as particle swarm optimisation algorithm with constriction factor, flower pollination algorithm, marine predators' algorithm and growth optimiser, have been applied for the optimal design of adaptive Hammerstein filter-based structures. The proposed structure has been analysed for electrocardiogram with various noise signals such as muscle artefact, white Gaussian noise etc. RESULTS: Among the adopted-metaheuristic algorithms applied to adaptive Hammerstein filter-based structures, the growth optimiser-optimised adaptive Hammerstein filter-based structures performed better with improved signal-to-noise ratio and minimal mean squared error values. A digital signal processor kit is used to authenticate the simulation outcomes.

Conclusion: The results (mean squared error: 3.698E-08 and signal-to-noise ratio improvement: 12 dB) obtained through the proposed technique ensure its supremacy compared to other state-of-the-art techniques.

Significance: Hence, the proposed method can be utilised for electrocardiogram signal enhancement.

Purpose: Calcific Aortic Valve Disease (CAVD) is a common cause for aortic stenosis characterized by the progressive calcification and stiffening of the aortic valve, often leading to significant hemodynamic changes and impaired cardiac function. Several Computational Fluid Dynamics (CFD) simulations have been conducted in the literature to provide more detailed analysis of CAVD, but are mainly reliant on uniform calcification. Also, outcomes from current clinical diagnostic techniques do not account for the effect of non-uniform calcification. The purpose of this study is to extend previous CFD simulations to non-uniformly calcified aortic valves and to evaluate the accuracy of clinical diagnostic methods under these conditions.

Methods: High-fidelity simulations of non-uniformly calcified aortic valves are performed by coupling fluid and solid solvers using a partitioned fluid-structure interaction (FSI) method for a patient-specific valve model extracted from computed tomography (CT) images. Non-uniform calcification is modelled by varying the elasticity along the leaflet, with several levels of calcification ranging from mild to severe.

Results: Non-uniform calcification alters flow physics, leading in up to 35-50% increase in maximum jet velocity and up to 150-170% rise in TPG compared to the normal valve, significant vortex shedding, extended flow separation regions, and intensified wall shear stress (WSS) fluctuations, especially on the ventricular side of the leaflet. The results indicate that the severity of calcification cannot be accurately predicted by several clinical diagnostic methods, with only effective orifice area (EOA) and maximum opening ratio (MOR) emerging as the reliable predictors.

Conclusions: The progression of non-uniform calcification on the aortic valve leaflets significantly impacts both hemodynamics and leaflet mechanics, with clear alterations in flow patterns and biomechanical responses as calcification severity increases. These findings highlight the need for more accurate diagnostic techniques and may drive the development of improved clinical strategies for the management and treatment of CAVD.

Purpose: Computational fluid dynamics (CFD) has been widely used to understand various cardiovascular diseases such as acute ischemic stroke (AIS), which occurs when a blood clot lodges in the cerebrovasculature and obstructs blood flow that may lead to brain damage or death. Compared with medical imaging, CFD can predict hemodynamics and clot migration, which are crucial in better understanding the biomechanics of AIS. To rely on computational modeling, however, the simulations need to be validated by comparing with experiments METHODS: In this study, we develop an in vitro experimental model of pulsatile flow in the aorta and cerebrovasculature. The model was filled with a blood analog fluid and pulsatile flow was driven by a piston pump to generate realistic physiological flow conditions. Experimental measurements of the time-varying pressure and flow rate were acquired and are used to validate corresponding CFD simulations RESULTS: CFD predictions of the time-averaged pressure at the outlets are shown to be within 8% of the experimental measurements, while the time-averaged flow rate is within 1%.

Conclusions: This work demonstrates a promising capability for modeling embolus migration and lodging in the brain. Future work will validate simulations of clot migration that may be used to better understand AIS biomechanics and treatment options.

Purpose: This study aimed to evaluate the blood-sampling performance of an automatic fingertip blood-sampling system with a fingertip vessel-puncture function (FBS-FV) and to examine the relationship between sampled blood volume and fingertip blood-vessel image features.

Methods: To obtain a consistent blood volume for testing, the FBS-FV selects and punctures near a large blood based on fingertip blood-vessel imaging and promotes bleeding by alternately pressing and releasing the fingertip. A blood-sampling experiment was conducted with 18 participants (men and women in their 20 to 60 s). Puncture accuracy, blood volume, and image features (relative brightness at the puncture position V and brightness change due to compression C) were analyzed. Multiple regression was applied to assess the predictive value of V and C for blood volume.

Results: (1) The deviation between the target and actual puncture positions was less than 1 mm, indicating high accuracy. (2) The proportion of blood samples obtained using the FBS-FV that exceeded the target volume (650 μL) was 42%, which was lower than in a previous experiment where the puncture position selected by the FBS-FV was manually punctured and blood was sampled. (3) Multiple regression analysis using image features V and C yielded coefficients of determination of 0.64 and 0.41 for high- and low-volume groups, respectively, suggesting that the possibility of predicting blood volume using these variables.

Conclusion: The FBS-FV demonstrated precise puncture performance and potential for predicting blood volume using image features. Further optimization of the FBS-FV's compression control might improve the consistency of blood sampling.