Physiological measurement最新文献

Objective: Arrhythmia classification from electrocardiograms (ECGs) suffers from high false positive rates and limited cross-dataset generalization, particularly for atrial fibrillation detection where specificity ranges from 0.72 to 0.98 using conventional 30-second analysis windows. While conventional deep learning approaches analyze isolated 30-second ECG segments, many arrhythmias, particularly atrial fibrillation (AF) and atrial flutter, exhibit diagnostic features that emerge over extended time scales.

Approach: We introduce S4ECG, a deep learning architecture based on structured state-space models (S4), designed to capture long-range temporal dependencies by jointly analyzing multiple consecutive ECG windows spanning up to 20 minutes. We evaluated S4ECG on four publicly available databases for multi-class arrhythmia classification, with systematic cross-dataset evaluations to assess out-of-distribution robustness.

Main results: Multi-window analysis consistently outperformed single-window approaches across all datasets, improving the macro-averaged area under the receiver operating characteristic curve (AUROC) by 1.0-11.6 percentage points. For AF detection specifically, specificity increased from 0.718-0.979 (single-window) to 0.967-0.998 (multi-window) at a fixed sensitivity threshold, representing a 3-10 fold reduction in false positive rates.

Significance: Comparative analysis against convolutional neural network baselines demonstrated superior performance of the S4 architecture. Cross-dataset evaluation revealed that multi-window approaches substantially improved generalization performance, with smaller performance degradation when models were tested on held-out datasets from different institutions and acquisition protocols. A systematic investigation revealed optimal diagnostic windows of 10-20 minutes, beyond which performance plateaus or degrades. These findings demonstrate that structured incorporation of extended temporal context enhances both arrhythmia classification accuracy and cross-dataset robustness. The identified optimal temporal windows provide practical guidance for ECG monitoring system design and may reflect underlying physiological timescales of arrhythmogenic dynamics.

Self-supervised learning (SSL) has enabled Vision Transformers (ViTs) to learn robust representations from large-scale natural image datasets, enhancing their generalization across domains. In retinal imaging, foundation models pretrained on either natural or ophthalmic data have shown promise, but the benefits of in-domain pretraining remain uncertain. To investigate this, we benchmark six SSL-pretrained ViTs on seven digital fundus image (DFI) datasets totaling 70,000 expert-annotated images for the task of moderate-to-late age-related macular degeneration (AMD) identification. Our results show that iBOT pretrained on natural images, achieves the highest out-of-distribution generalization, with AUROCs of 0.80-0.97, outperforming domain-specific models, which achieved AUROCs of 0.78-0.96 and a baseline ViT-L with no pretraining, which achieved AUROC of 0.68-0.91. These findings highlight the value of foundation models in improving AMD identification, and challenge the assumption that in-domain pretraining is necessary. Furthermore, we release BRAMD, an open-access dataset (n=587) of DFIs with AMD labels from Brazil.

Objective. Fetal and maternal health during pregnancy can be monitored with sensors such as Doppler or scalp fetal ECG. This study focuses on single-channel dry electrode maternal abdominal ECG (aECG) to extract fetal heart rate (fHR) using a low-complexity algorithm suitable for low-power wearables.Approach. A hybrid model combining machine learning, QRS masking, and data fusion was trained on two PhysioNet databases and synthetically generatedaECG. Model selection employed the Akaike criterion with data balancing and random sampling.Main results. The algorithm was tested on 80 recordings from the Computer in Cardiology Challenge 2013 (CCC) and the abdominal and direct fetal database (ADFD), augmented with 100 syntheticaECG. Performance for fetal QRS detection reachedPrecision=97.2(82.2)%,Specificity=99.8(93.8)%, andSensitivity=97.4(93.9)% on ADFD and CCC, respectively. Clinical validation used the Polar Electro Oy H10 dry-electrode device at the Maternity Hospital of Southwest Finland. Four subjects (gestational age39.8±1.3 weeks) were analyzed, with seven discarded. ForfHR, the mean absolute percentage error was1.9±1.0%, Availability79.6±3.9%, and coverage probabilityCP5=76.2%,CP10=87.5%.Significance. These results demonstrate the feasibility offHRmonitoring from dry-electrodeaECGtailored for low-power wearables. Signal quality in clinical subjects matched the lowest PhysioNet cases, confirming robustness under low signal-to-noise conditions.

Objective.the fetal electrocardiogram (FECG) is critical for monitoring fetal health, however, its extraction remains technically challenging due to strong interference from the maternal electrocardiogram (MECG) in abdominal electrocardiogram (AECG). Therefore, an attention-based generative adversarial network (AGAN) is proposed for source separation of FECG from single-lead AECG signals.Approach.the AGAN architecture uniquely combines two powerful techniques: GAN-style adversarial training for high-quality data generation and attention-based focus mechanisms for intelligent feature selection, leading to superior target signal extraction from complex mixtures. The innovation of the proposed method lies in addressing the amplitude bias issue in multi-objective learning tasks. This work innovatively employs the Hadamard product as the learning objective for the model, preventing the model from favoring high-amplitude components (e.g. MECG) while neglecting low-amplitude yet critical features (e.g. FECG).Main results.experimental results demonstrate that the proposed method can effectively and simultaneously separate both MECG and FECG components from single-lead AECG signals. When evaluated on the ADFECGDB, B2_LABOUR, and PCDB datasets, the proposed method demonstrated consistent performance, achieving the following SE, PPV, andF1 scores: 96.67%, 97.13%, and 96.90% on ADFECGDB; 95.90%, 96.56%, and 96.22% on B2_LABOUR; and 94.96%, 95.18%, and 95.06% on PCDB.Significance.this study presents a robust method for FECG extraction while simultaneously introducing an innovative data-driven framework for blind source separation problems.

Infrared thermography (IRT) is projected as an innovative and very promising tool for the observation of muscle response to fatigue. The aim of this study was to observe the skin temperature (Tsk) by thermography about acute muscle fatigue and delayed soreness (DOMS) following a exercise protocol until exhaustion in the triceps suralis. An open longitudinal descriptive observational study of the posterior leg region was performed in 73 healthy subjects. Data on age, sex, body mass index (BMI) and triceps suralis thermography pre, post and 24 h after maximum muscle fatigue physical exercise, as well as pressure-pain threshold (PPT) and pain sensation by Analogic visual scale (VAS scale) were recruited. Results showed significant difference in skin temperature over time (p= <0.001; η2p= 0.272), side (p= 0.021; η2p= 0.072) and a time x side interaction (p= 0.011; η2p= 0.061), as well as in PPT over time (p= <0.001; η 2 p= 0.328) and in the interaction between time and sex (p= 0.050; η 2 p= 0.041). and Vas scale over time (p= <0.001; η 2 p= 0.831). According to the results obtained, this technique could be a reliable method to evaluate DOMS. Exploring the integration of thermography with other modalities.

Objective: As cardiovascular diseases continue to rise, the accurate and convenient calculation of stroke volume (SV) and cardiac output (CO) has become an important topic. Studies have shown that electrical impedance tomography (EIT) can provide continuous non-invasive SV measurements. Despite its potential, a review of the various calculation methods for EIT-based SV and CO, along with their clinical utility, is lacking. Approach: A literature search was conducted on PubMed and Web of Science Core Collection. Full-text research articles in English were reviewed and discussed. Main results: In recent years, advancements in technology, clinical research, and intelligent algorithms have revealed EIT's substantial potential in SV monitoring. Significance: This article offers a review of the evolution of EIT technology in measuring SV, introducing various calculation methods, their advantages, challenges, and clinical applications. .

Objective: Autonomous Sensory Meridian Response (ASMR) is a tingling sensation induced while attending to specific sounds, including whispering, tapping, scratching, or other soft, repetitive noises. While previous studies focused on low arousal-positive emotions such as relaxation and calmness, this study explores a broader range of emotions elicited by ASMR auditory stimuli, including happiness, sadness, and disgust.

Approach: The proposed study collects the multi-modal physiological data from Electroencephalography (EEG), Photoplethysmography (PPG), and Electrodermal Activity (EDA) via wearable bio-sensors from 23 ASMR-experiencing participants while exposed to different ASMR-inducing auditory stimuli. It employs the rmANOVA test on the collected physiological responses and self-reported ratings for quantitative analysis and results in a significant difference between the emotions induced from the four audio stimuli, i.e., Happy from A1, Sad from A2, Calm from A3, Disgust from A4, and the neutral state. The proposed study also applies deep learning classifiers, Artificial Neural Network (ANN), and Convolution Neural Network (CNN) to the collected multi-modal physiological data to classify the four induced emotions from the ASMR auditory stimuli using the dimensions of arousal, valence, and dominance.

Main results: The classification accuracy results from ANN, and CNN prove an excellent success rate of 96.12% and 74.25% with multi-modal Valence-Arousal-Dominance (VAD) for ANN and CNN, respectively, in classifying the four emotions induced by ASMR stimuli. And the statistical rmANOVA test results indicated distinctions among the four emotions, as the p-values exceeded the significance threshold of 0.05.

Significance: The results highlight the effectiveness of multi-modal physiological signals and deep learning in reliably classifying ASMR-induced emotions, contributing to advancements in emotion recognition for mental health and therapeutic applications.

Objective: Wearable devices with embedded photoplethysmography (PPG) enable continuous non-invasive monitoring of cardiac activity, offering a promising strategy to reduce the global burden of cardiovascular diseases. However, monitoring during daily life introduces motion artifacts that can compromise the signals. Traditional signal decomposition techniques often fail with severe artifacts. Deep learning denoisers are more effective but have poorer interpretability, which is critical for clinical acceptance. This study proposes a framework that combines the advantages of both signal decomposition and deep learning approaches.

Approach: We leverage algorithm unfolding to integrate prior knowledge about the PPG structure into a deep neural network, improving its interpretability. A learned convolutional sparse coding model encodes the signal into a sparse representation using a learned dictionary of kernels that capture recurrent morphological patterns. The network is trained for denoising using the PulseDB dataset and a synthetic motion artifact model from the literature. Performance is benchmarked with PPG during daily activities using the PPG-DaLiA dataset and compared with two reference deep learning methods.

Main results: On the synthetic dataset, the proposed method, on average, improved the signal-to-noise ratio (SNR) from -7.06 dB to 11.23 dB and reduced the heart rate mean absolute error (MAE) by 55%. On the PPG-DaLiA dataset, the MAE decreased by 23%. The proposed method obtained higher SNR and comparable MAE to the reference methods.

Significance: Our method effectively enhances the quality of PPG signals from wearable devices and enables the extraction of meaningful waveform features, which may inspire innovative tools for monitoring cardiovascular diseases.

Objective.Cardiovascular disease (CVD) causes severe global health threat, and electrocardiogram (ECG) is crucial for early CVD diagnosis. Recently, two popular deep learning methods, that is, convolutional neural network (CNN) and long short-term memory (LSTM) network are studied for ECG modeling and CVD diagnosis, but CNN adopts fixed kernels, thereby reducing efficiency and introducing noise, and LSTM struggles with local feature correlations.Approach.This study proposes an adaptive CNN-LSTM (aCNN-LSTM) fusion network for ECG diagnosis. An adaptive convolutional kernel is newly designed, which can dynamically adjust size based on local signal variance. Smaller kernels optimize efficiency in stationary segments, while larger kernels extract diverse features in non-stationary regions. The adaptive features from aCNN are further fed into LSTM to capture temporal relationships. Finally, a spatial-temporal fusion mechanism is used and a multi-class classification is achieved via the output layer.Main results.Experiments on the PTB-XL dataset show that the proposed aCNN-LSTM net outperforms CNN, LSTM, and CNN-LSTM in diagnosis performance: its overall accuracy reaches 89.89%, macro-averageF1-score is 0.9640, and weighted-averageF1-score is 0.9698.Significance.This method enhances the efficiency and accuracy of automatic ECG diagnosis, and provides reliable technical support for early CVD screening in clinical and primary medical settings.

Objective.Prompt identification of haematomas is crucial for effective clinical treatment. Magnetic induction phase shift technology (MIPS), known for its portability, non-contact nature, and affordability, is limited by the weak signal induced by cerebral hemorrhage leading to poor sensitivity, which is urgent to be improved.Approach. Tracer of magnetic nanoparticles is introduced to produce robust induced magnetic field. A symmetrical gradiometer coil is used as the receiving coil to nullify the effect of primary magnetic field generated by the excitation coil, which is designed as a Helmholtz coil.Main results.In vitroexperiments showcase the remarkably improved sensitivity and stability of the detection system, with magnetic nanoparticles notably boosting the MIPS signal for hemorrhage. Moreover,in vivoexperiments employing a rabbit autologous blood cerebral hemorrhage model reveal that with a hemorrhage volume of 2 ml, the experimental group with employed magnetic nanoparticles increased the MIPS signal change by 23-fold compared to the control group without magnetic nanoparticles.Significance. The sensitivity of MIPS for hemorrhage detection is significantly improved compared to traditional method. The magnetic nanoparticle-enhanced MIPS detection technique holds promise as an optimal solution for real-time, non-invasive bedside monitoring for cerebral hemorrhage.