Endocardial acceleration (sonR) vs. ultrasound-derived time intervals in recipients of cardiac resynchronization therapy systems

Optimized CNN-based denoising strategy for enhancing longitudinal monitoring of heart failure

Cardiac vibration signal analysis emerges as a remarkable tool for the diagnosis of heart conditions. Our recent study shows the feasibility of the longitudinal monitoring of chronic heart diseases, particularly heart failure, using a gastric fundus implant. However, cardiac vibration data, captured from the implant, positioned at the gastric fundus, can be highly affected by different noises and artefacts. This study introduces a novel methodology for addressing denoising challenges in the longitudinal monitoring of chronic heart diseases, using gastric fundus implants. More precisely, a novel method is designed, by repurposing pre-trained convolutional neural network models, originally designed for classification tasks, with adequately chosen convolution filters. The proposed approach efficiently tackles noise and artefacts reduction in the acquired accelerometer signals. Moreover, the integration of additional Hilbert and Homomorphic envelopes enhances the implant's ability to better segment heart sounds, namely S1 and S2. The quality assessment of this denoising strategy is performed, in the lack of ground truth, by rather evaluating its impact on a classification stage that is introduced to the proposed pipeline. Compared to standard denoising matrix factorization and tensor decomposition-based methods, results on a real 3D accelerometer dataset acquired from a set of pigs, with and without heart failure, demonstrate the efficacy of such a proposed optimized CNN-based approach with the best balance between enhancing the segmentation accuracy and preserving a maximum usable record.

Machine learning-powered, device-embedded heart sound measurement can optimize AV delay in patients with CRT

BACKGROUND

Continuous optimization of atrioventricular (AV) delay for cardiac resynchronization therapy (CRT) is mainly performed by electrical means.

OBJECTIVE

The purpose of this study was to develop an estimation model of cardiac function that uses a piezoelectric microphone embedded in a pulse generator to guide CRT optimization.

METHODS

Electrocardiogram, left ventricular pressure (LVP), and heart sounds were simultaneously collected during CRT device implantation procedures. A piezoelectric alarm transducer embedded in a modified CRT device facilitated recording of heart sounds in patients undergoing a pacing protocol with different AV delays. Machine learning (ML) was used to produce a decision-tree ensemble model capable of estimating absolute maximal LVP (LVPmax) and maximal rise of LVP (LVdP/dtmax) using 3 heart sound-based features. To gauge the applicability of ML in AV delay optimization, polynomial curves were fitted to measured and estimated values.

RESULTS

In the data set of ∼30,000 heartbeats, ML indicated S1 amplitude, S2 amplitude, and S1 integral (S1 energy for LVdP/dtmax) as most prominent features for AV delay optimization. ML resulted in single-beat estimation precision for absolute values of LVPmax and LVdP/dtmax of 67% and 64%, respectively. For 20-30 beat averages, cross-correlation between measured and estimated LVPmax and LVdP/dtmax was 0.999 for both. The estimated optimal AV delays were not significantly different from those measured using invasive LVP (difference -5.6 ± 17.1 ms for LVPmax and +5.1 ± 6.7 ms for LVdP/dtmax). The difference in function at estimated and measured optimal AV delays was not statiscally significant (1 ± 3 mm Hg for LVPmax and 9 ± 57 mm Hg/s for LVdP/dtmax).

CONCLUSION

Heart sound sensors embedded in a CRT device, powered by a ML algorithm, provide a reliable assessment of optimal AV delays and absolute LVPmax and LVdP/dtmax.

Left Ventricular Pressure Estimation Using Machine Learning-Based Heart Sound Classification

Objective

A method to estimate absolute left ventricular (LV) pressure and its maximum rate of rise (LV dP/dtmax) from epicardial accelerometer data and machine learning is proposed.

Methods

Five acute experiments were performed on pigs. Custom-made accelerometers were sutured epicardially onto the right ventricle, LV, and right atrium. Different pacing configurations and contractility modulations, using isoflurane and dobutamine infusions, were performed to create a wide variety of hemodynamic conditions. Automated beat-by-beat analysis was performed on the acceleration signals to evaluate amplitude, time, and energy-based features. For each sensing location, bootstrap aggregated classification tree ensembles were trained to estimate absolute maximum LV pressure (LVPmax) and LV dP/dtmax using amplitude, time, and energy-based features. After extraction of acceleration and pressure-based features, location specific, bootstrap aggregated classification ensembles were trained to estimate absolute values of LVPmax and its maximum rate of rise (LV dP/dtmax) from acceleration data.

Results

With a dataset of over 6,000 beats, the algorithm narrowed the selection of 17 predefined features to the most suitable 3 for each sensor location. Validation tests showed the minimal estimation accuracies to be 93% and 86% for LVPmax at estimation intervals of 20 and 10 mmHg, respectively. Models estimating LV dP/dtmax achieved an accuracy of minimal 93 and 87% at estimation intervals of 100 and 200 mmHg/s, respectively. Accuracies were similar for all sensor locations used.

Conclusion

Under pre-clinical conditions, the developed estimation method, employing epicardial accelerometers in conjunction with machine learning, can reliably estimate absolute LV pressure and its first derivative.

Analysis of Cardiac Vibration Signals Acquired From a Novel Implant Placed on the Gastric Fundus

The analysis of cardiac vibration signals has been shown as an interesting tool for the follow-up of chronic pathologies involving the cardiovascular system, such as heart failure (HF). However, methods to obtain high-quality, real-world and longitudinal data, that do not require the involvement of the patient to correctly and regularly acquire these signals, remain to be developed. Implantable systems may be a solution to this observability challenge. In this paper, we evaluate the feasibility of acquiring useful electrocardiographic (ECG) and accelerometry (ACC) data from an innovative implant located in the gastric fundus. In a first phase, we compare data acquired from the gastric fundus with gold standard data acquired from surface sensors on 2 pigs. A second phase investigates the feasibility of deriving useful hemodynamic markers from these gastric signals using data from 4 healthy pigs and 3 pigs with induced HF with longitudinal recordings. The following data processing chain was applied to the recordings: (1) ECG and ACC data denoising, (2) noise-robust real-time QRS detection from ECG signals and cardiac cycle segmentation, (3) Correlation analysis of the cardiac cycles and computation of coherent mean from aligned ECG and ACC, (4) cardiac vibration components segmentation (S1 and S2) from the coherent mean ACC data, and (5) estimation of signal context and a signal-to-noise ratio (SNR) on both signals. Results show a high correlation between the markers acquired from the gastric and thoracic sites, as well as pre-clinical evidence on the feasibility of chronic cardiovascular monitoring from an implantable cardiac device located at the gastric fundus, the main challenge remains on the optimization of the signal-to-noise ratio, in particular for the handling of some sources of noise that are specific to the gastric acquisition site.

Haemodynamic monitoring of cardiac status using heart sounds from an implanted cardiac device

AIM

The aim of this study was to evaluate the haemodynamic correlates of heart sound (HS) parameters such as third HS (S3), first HS (S1), and HS-based systolic time intervals (HSTIs) from an implantable cardiac device.

METHODS AND RESULTS

Two unique animal models (10 swine with myocardial ischaemia and 11 canines with pulmonary oedema) were used to evaluate haemodynamic correlates of S1, S3, and HSTIs, namely, HS-based pre-ejection period (HSPEP), HS-based ejection time (HSET), and the ratio HSPEP/HSET during acute haemodynamic perturbations. The HS was measured using implanted cardiac resynchronization therapy defibrillator devices simultaneously with haemodynamic references such as left atrial (LA) pressure and left ventricular (LV) pressure. In the ischaemia model, S1 amplitude (r = 0.76 ± 0.038; P = 0.002), HSPEP (r = -0.56 ± 0.07; P = 0.002), and HSPEP/HSET (r = -0.42 ± 0.1; P = 0.002) were significantly correlated with LV dP/dt_max . In contrast, HSET was poorly correlated with LV dP/dt_max (r = 0.14 ± 0.14; P = 0.23). In the oedema model, a physiological delayed response was observed in S3 amplitude after acute haemodynamic perturbations. After adjusting for the delay, S3 amplitude significantly correlated with LA pressure in individual animals (r = 0.71 ± 0.07; max: 0.92; min: 0.17) as well as in aggregate (r = 0.62; P < 0.001). The S3 amplitude was able to detect elevated LA pressure, defined as >25 mmHg, with a sensitivity = 58% and specificity = 90%.

CONCLUSIONS

The HS parameters such as S1, S3, and HSTIs measured using implantable devices significantly correlated with haemodynamic changes in acute animal models, suggesting potential utility for remote heart failure patient monitoring.

Moving toward automatic and standalone delineation of seismocardiogram signal

The purpose of this research is to propose an algorithm that could accomplish automatic delineation of the seismocardiogram (SCG) signal without using a reference electrocardiogram R-wave. As a result, the SCG signal could be used, as a stand-alone solution for many cardiovascular medical applications such as hemorrhage detection, cardiac computed tomographic gating, cardiac resynchronization therapy, hemodynamics estimations and diastolic timed vibration. Multiple envelopes were derived from the seismocardiogram signal by using filtering and triple integration. The first envelope is referred as the heart rate envelope, which has the characteristics of having a period of exactly one cardiac cycle and its purpose is to replace the ECG R-wave as a reference point. Our dataset is based on the lower body negative pressure (LBNP) test that was conducted on 18 individuals, containing 21610 cardiac cycles. For 94% of the LBNP dataset, the aforementioned envelope estimated heart rate within 3 beats per minute. Three different peaks of the SCG signal are of our interest: isovolumic contraction (IM), aortic valve opening (AO) and aortic valve closure (AC). For each of these desired peaks of the SCG signal, a different envelope was designed in a manner that its peak is very close to IM, AO and AC, respectively. For the same lower body negative pressure data set, a mean difference of (9, 9, 6) and standard deviation of (8, 9, 9) millisecond between the peak of envelopes and IM, AO and AC is accomplished. This could be used as a good initial estimation of the annotation points.

First clinical evaluation of an atrial haemodynamic sensor lead for automatic optimization of cardiac resynchronization therapy

AIMS

One option to improve cardiac resynchronization therapy (CRT) responder rates lies in the optimization of pacing intervals. A haemodynamic sensor embedded in the SonRtip atrial lead measures cardiac contractility and provides a systematic automatic atrioventricular and interventricular delays optimization. This multi-centre study evaluated the safety and performance of the lead, up to 1 year.

METHODS AND RESULTS

A total of 99 patients were implanted with the system composed of the lead and a CRT-Defibrillator device. Patients were followed at 1, 3, 6, and 12 months post-implant. The primary safety objective was to demonstrate that the atrial lead complication free rate was superior to 90% at 3-months follow-up visit. A lead handling questionnaire was filled by implanting investigators. Lead electrical performances and the performance of the system to compute AV and VV delays were evaluated at each study visit over 1 year. The complication free rate at 3 months post-implant was 99.0% [95%CI 94.5-100.0%], P < 0.001. Electrical performances of the lead were adequate whatever the atrial lead position and remained stable over the study period. The optimization algorithm was able to compute AV and VV delays in 97% of patients, during >75% of the weeks.

CONCLUSION

The atrial lead is safe to implant and shows stable electrical performance over time. It therefore offers a promising tool for automatic CRT optimization to further improve responder rates to CRT.

Analysis of endocardial acceleration during intraoperative optimization of cardiac resynchronization therapy

Cardiac resynchronization therapy (CRT) is the therapy of choice for selected patients suffering from drug-refractory congestive heart failure and presenting an interventricular desynchronization. CRT is delivered by an implantable biventricular pacemaker, which stimulates the right atrium and both ventricles at specific timings. The optimization and personalization of this therapy requires to quantify both the electrical and the mechanical cardiac functions during the intraoperative and postoperative phases. The objective of this paper is to evaluate the feasibility of the calculation of features extracted from endocardial acceleration (EA) signals and the potential utility of these features for the intraoperative optimization of CRT. Endocardial intraoperative data from one patient are analyzed for 33 different pacing configurations, including changes in the atrio-ventricular and inter-ventricular delays and different ventricular stimulation sites. The main EA features are extracted for each pacing configuration and analyzed so as to estimate the intra-configuration and inter-configuration variability. Results show the feasibility of the proposed approach and suggest the potential utility of EA for intraoperative monitoring of the cardiac function and defining optimal, adaptive pacing configurations.

Myocardial contractility: a seismocardiography approach

Features are extracted from seismocardiogram data to correlate with two indexes of myocardial contractility: dP/dt(max) (maximum first derivative of left ventricular pressure) and stroke volume. In the first study on three pigs, it is shown that the time period between the R peak of the ECG and the first peak of the SCG (R-AO period or pre-ejection period, PEP) correlated (r= -0.86) with dP/dt(max). In the second study, stroke volume is gradually reduced in five human subjects using lower body negative pressure. The same feature as the pigs (R-AO) is correlated the most with stroke volume (r= -0.90).