Deep learning-based methods have demonstrated high classification performance in the detection of cardiovascular diseases from electrocardiograms (ECGs). However, their blackbox character and the associated lack of interpretability limit their clinical applicability. To overcome existing limitations, we present a novel deep learning architecture for interpretable ECG analysis (xECGArch). For the first time, short- and long-term features are analyzed by two independent convolutional neural networks (CNNs) and combined into an ensemble, which is extended by methods of explainable artificial intelligence (xAI) to whiten the blackbox. To demonstrate the trustworthiness of xECGArch, perturbation analysis was used to compare 13 different xAI methods. We parameterized xECGArch for atrial fibrillation (AF) detection using four public ECG databases ( $$n = 9854$$n = 9854 ECGs) and achieved an F1 score of 95.43% in AF versus non-AF classification on an unseen ECG test dataset. A systematic comparison of xAI methods showed that deep Taylor decomposition provided the most trustworthy explanations ( $$+24\%$$+ 24 % compared to the second-best approach). xECGArch can account for short- and long-term features corresponding to clinical features of morphology and rhythm, respectively. Further research will focus on the relationship between xECGArch features and clinical features, which may help in medical applications for diagnosis and therapy.

We propose a novel method for evaluating the similarity between two 1d patterns. Our method, referred to as two-dimensional signal warping (2DSW), extends the basic ideas of known warping techniques such as dynamic time warping and correlation optimized warping. By employing two-dimensional piecewise stretching 2DSW is able to take into account inhomogeneous variations of shapes. We apply 2DSW to ECG recordings to extract beat-to-beat variability in QT intervals (QTV) that is indicative of ventricular repolarization lability and typically characterised by a low signal-to-noise ratio. Simulation studies show high robustness of our approach in presence of typical ECG artefacts. Comparison of short-term ECG recorded in normal subjects versus patients with myocardial infarction (MI) shows significantly increased QTV in patients (normal subject 2.36 ms ± 1.05 ms vs. MI patients 5.94 ms ± 5.23 ms (mean ± std), ). Evaluation of a standard QT database shows that 2DSW allows highly accurate tracking of QRS-onset and T-end. In conclusion, the two-dimensional warping approach introduced here is able to detect subtle changes in noisy quasi-periodic biomedical signals such as ECG and may have diagnostic potential for measuring repolarization lability in MI patients. In more general terms, the proposed method provides a novel means for morphological characterization of 1d signals.

Camera-based photoplethysmography (cbPPG) is an optical measurement technique that reveals pulsatile blood flow in cutaneous microcirculation from a distance. cbPPG has been shown to reflect pivotal haemodynamic events like cardiac ejection in healthy subjects. In addition, it provides valuable insight into intrinsic microcirculatory regulation as it yields dynamic, two-dimensional perfusion maps. In this study, we evaluate the feasibility of a clinical cbPPG application in critical care patients. A mobile camera set-up to record faces of patients at the bed site was constructed. Videos were made during the immediate recovery after cardiac surgery under standard critical care conditions and were processed offline. Major motion artefacts were detected using an optical flow technique and suitable facial regions were manually annotated. cbPPG signals were highpass filtered and Fourier spectra out of consecutive 10s signal segments calculated for heart rate detection. Signal-to-noise ratios (SNR) of the Fourier spectra were derived as a quality measure. Reference data of vital parameters were synchronously acquired from the bed site monitoring system. Seventy patient videos of an average time of 28.6±2.8 min were analysed. Heart rate (HR) was detected within a±5 bpm range compared to reference in 83% of total recording time. Low SNR and HR detection failure were mostly, but not exclusively, attributed to non-physiological events like patient motion, interventions or sudden changes of illumination. SNR was reduced by low arterial blood pressure, whereas no impact of other perioperative or disease-related parameters was identified. Cardiac ejection is detectable by cbPPG under pathophysiologic conditions of cardiovascular disease and perioperative medicine. cbPPG measurements can be seamlessly integrated into the clinical work flow of critical care patients.

Numerous vital signs are reported in association with stress response assessment, but their application varies widely. This work provides an overview over methods for stress induction and strain assessment, and presents a multimodal experimental study to identify the most important vital signs for effective assessment of the response to acute mental stress. We induced acute mental stress in 65 healthy participants with the Mannheim Multicomponent Stress Test and acquired self-assessment measures (Likert scale, Self-Assessment Manikin), salivary α-amylase and cortisol concentrations as well as 60 vital signs from biosignals, such as heart rate variability parameters, QT variability parameters, skin conductance level, and breath rate. By means of statistical testing and a self-optimizing logistic regression, we identified the most important biosignal vital signs. Fifteen biosignal vital signs related to ventricular repolarization variability, blood pressure, skin conductance, and respiration showed significant results. The logistic regression converged with QT variability index, left ventricular work index, earlobe pulse arrival time, skin conductance level, rise time and number of skin conductance responses, breath rate, and breath rate variability (F1 = 0.82). Self-assessment measures indicated successful stress induction. α-amylase and cortisol showed effect sizes of -0.78 and 0.55, respectively. In summary, the hypothalamic-pituitary-adrenocortical axis and sympathetic nervous system were successfully activated. Our findings facilitate a coherent and integrative understanding of the assessment of the stress response and help to align applications and future research concerning acute mental stress.

Camera-based photoplethysmography (cbPPG) is a measurement technique which enables remote vital sign monitoring by using cameras. To obtain valid plethysmograms, proper regions of interest (ROIs) have to be selected in the video data. Most automated selection methods rely on specific spatial or temporal features limiting a broader application. In this work, we present a new method which overcomes those drawbacks and, therefore, allows cbPPG to be applied in an intraoperative environment. We recorded 41 patients during surgery using an RGB and a near-infrared (NIR) camera. A Bayesian skin classifier was employed to detect suitable regions, and a level set segmentation approach to define and track ROIs based on spatial homogeneity. The results show stable and homogeneously illuminated ROIs. We further evaluated their quality with regards to extracted cbPPG signals. The green channel provided the best results where heart rates could be correctly estimated in 95.6% of cases. The NIR channel yielded the highest contribution in compensating false estimations. The proposed method proved that cbPPG is applicable in intraoperative environments. It can be easily transferred to other settings regardless of which body site is considered.