A Cross-Stage Partial Network and a Cross-Attention-Based Transformer for an Electrocardiogram-Based Cardiovascular Disease Decision System

Cardiovascular disease (CVD) is one of the leading causes of death globally. Currently, clinical diagnosis of CVD primarily relies on electrocardiograms (ECG), which are relatively easier to identify compared to other diagnostic methods. However, ensuring the accuracy of ECG readings requires specialized training for healthcare professionals. Therefore, developing a CVD diagnostic system based on ECGs can provide preliminary diagnostic results, effectively reducing the workload of healthcare staff and enhancing the accuracy of CVD diagnosis. In this study, a deep neural network with a cross-stage partial network and a cross-attention-based transformer is used to develop an ECG-based CVD decision system. To accurately represent the characteristics of ECG, the cross-stage partial network is employed to extract embedding features. This network can effectively capture and leverage partial information from different stages, enhancing the feature extraction process. To effectively distill the embedding features, a cross-attention-based transformer model, known for its robust scalability that enables it to process data sequences with different lengths and complexities, is employed to extract meaningful embedding features, resulting in more accurate outcomes. The experimental results showed that the challenge scoring metric of the proposed approach is 0.6112, which outperforms others. Therefore, the proposed ECG-based CVD decision system is useful for clinical diagnosis.

Evaluating a Novel Infant Heart Rate Detector for Neonatal Resuscitation Efforts: Protocol for a Proof-of-Concept Study

BACKGROUND

Over 10 million newborns worldwide undergo resuscitation at birth each year. Pediatricians may use electrocardiogram (ECG), pulse oximetry (PO), and stethoscope in determining heart rate (HR), as HR guides the need for and steps of resuscitation. HR must be obtained quickly and accurately. Unfortunately, the current diagnostic modalities are either too slow, obtaining HR in more than a minute, or inaccurate. With time constraints, a reliable robust heart rate detector (HRD) modality is required. This paper discusses a protocol for conducting a methods-based comparison study to determine the HR accuracy of a novel real-time HRD based on 3D-printed dry-electrode single-lead ECG signals for cost-effective and quick HR determination. The HRD's HR results are compared to either clinical-grade ECG or PO monitors to ensure robustness and accuracy.

OBJECTIVE

The purpose of this study is to design and examine the feasibility of a proof-of-concept HRD that quickly obtains HR using biocompatible 3D-printed dry electrodes for single-lead neonatal ECG acquisition. This study uses a novel HRD and compares it to the gold-standard 3-lead clinical ECG or PO in a hospital setting.

METHODS

A cross-sectional study is planned to be conducted in the neonatal intensive care unit or postpartum unit of a large community teaching hospital in Toronto, Canada, from June 2023 to June 2024. In total, 50 newborns will be recruited for this study. The HRD and an ECG or PO monitor will be video recorded using a digital camera concurrently for 3 minutes for each newborn. Hardware-based signal processing and patent-pending embedded algorithm-based HR estimation techniques are applied directly to the raw collected single-lead ECG and displayed on the HRD in real time during video recordings. These data will be annotated and compared to the ECG or PO readings at the same points in time. Accuracy, F1-score, and other statistical metrics will be produced to determine the HRD's feasibility in providing reliable HR.

RESULTS

The study is ongoing. The projected end date for data collection is around July 2024.

CONCLUSIONS

The study will compare the novel patent-pending 3D-printed dry electrode-based HRD's real-time HR estimation techniques with the state-of-the-art clinical-grade ECG or PO monitors for HR accuracy and examines how fast the HRD provides reliable HR. The study will further provide recommendations and important improvements that can be made to implement the HRD for clinical applications, especially in neonatal resuscitation efforts. This work can be seen as a stepping stone in the development of robust dry-electrode single-lead ECG devices for HR estimations in the pediatric population.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

DERR1-10.2196/45512.

Norwegian Endurance Athlete ECG Database

Athletes often have training-induced remodeling of the heart, and this can sometimes be seen as abnormal but non-pathological changes in the electrocardiogram. However, these changes can be confused with severe cardiovascular diseases that, in some cases, can cause cardiovascular death. Electrocardiogram data from athletes is therefore important to learn more about the difference between normal athletic remodeling and pathological remodeling of the heart. This work provides a dataset of electrocardiograms from 28 Norwegian elite endurance athletes. The electrocardiograms are standard 12-lead resting ECGs, recorded for 10 seconds while the athlete's lay supine on a bench. The electrocardiograms were then interpreted by an interpretation algorithm and by a trained cardiologist. The electrocardiogram waveform data and the interpretations were stored in Python Waveform Database format and made publicly available through PhysioNet.

A Meta-Transfer Learning Approach to ECG Arrhythmia Detection

Automatic classification of cardiac abnormalities is becoming increasingly popular with the prevalence of ECG recordings. Many signal processing and machine learning algorithms have shown the potential to identify cardiac ab-normalities accurately. However, most of these methods heavily rely on a large amount of relatively homogeneous datasets. In real life, chances are that there is not enough data for a specific category, and regular deep learning may fail in this scenario. A straightforward intuition is to use the knowledge learned from previous data to solve the problem. This idea leads to learning-to-learn: extrapolating the knowledge accumulated from the old dataset and using it in a different but somewhat related dataset. In this way, we do not need to have considerable data to learn the new task because the underlying features of the old and new datasets resemble one another. In this paper, a novel machine learning method is introduced to solve the ECG arrhythmia detection problem with a limited amount of data. The proposed method combines the popular techniques of meta-learning and transfer learning. It is shown that our method achieves much higher accuracy in ECG arrhythmia classification with a few data and learns the new task faster than regular deep learning.

From 12 to 1 ECG lead: multiple cardiac condition detection mixing a hybrid machine learning approach with a one-vs-rest classification strategy

OBJECTIVE

Detecting different cardiac diseases using a single or reduced number of leads is still challenging. This work aims to provide and validate an automated method able to classify ECG recordings. Performance using complete 12-lead systems, reduced lead sets, and single-lead ECGs is evaluated and compared.

APPROACH

Seven different databases with 12-lead ECGs were provided during the PhysioNet/Computing in Cardiology Challenge 2021, where 88,253 annotated samples associated with none, one, or several cardiac conditions among 26 different classes were released for training, whereas 42,896 hidden samples were used for testing. After signal preprocessing, 81 features per ECG-lead were extracted, mainly based on heart rate variability, QRST patterns and spectral domain. Next, a One-vs-Rest classification approach made of independent binary classifiers for each cardiac condition was trained. This strategy allowed each ECG to be classified as belonging to none, one or several classes. For each class, a classification model among two binary Supervised Classifiers and one Hybrid Unsupervised-Supervised classification system was selected. Finally, we performed a 3-fold cross-validation to assess the system's performance.

MAIN RESULTS

Our classifiers received scores of 0.39, 0.38, 0.39, 0.38, and 0.37 for the 12, 6, 4, 3 and 2-lead versions of the hidden test set with the Challenge evaluation metric (CM). Also, we obtained a mean G-score of 0.80, 0.78, 0.79, 0.79, 0.77 and 0.74 for the 12, 6, 4, 3, 2 and 1-lead subsets with the public training set during our 3-fold cross-validation.

SIGNIFICANCE

We proposed and tested a machine learning approach focused on flexibility for identifying multiple cardiac conditions using one or more ECG leads. Our minimal-lead approach may be beneficial for novel portable or wearable ECG devices used as screening tools, as it can also detect multiple and concurrent cardiac conditions.

Opening the black box: interpretability of machine learning algorithms in electrocardiography

Recent studies have suggested that cardiac abnormalities can be detected from the electrocardiogram (ECG) using deep machine learning (DL) models. However, most DL algorithms lack interpretability, since they do not provide any justification for their decisions. In this study, we designed two new frameworks to interpret the classification results of DL algorithms trained for 12-lead ECG classification. The frameworks allow us to highlight not only the ECG samples that contributed most to the classification, but also which between the P-wave, QRS complex and T-wave, hereafter simply called 'waves', were the most relevant for the diagnosis. The frameworks were designed to be compatible with any DL model, including the ones already trained. The frameworks were tested on a selected Deep Neural Network, trained on a publicly available dataset, to automatically classify 24 cardiac abnormalities from 12-lead ECG signals. Experimental results showed that the frameworks were able to detect the most relevant ECG waves contributing to the classification. Often the network relied on portions of the ECG which are also considered by cardiologists to detect the same cardiac abnormalities, but this was not always the case. In conclusion, the proposed frameworks may unveil whether the network relies on features which are clinically significant for the detection of cardiac abnormalities from 12-lead ECG signals, thus increasing the trust in the DL models. This article is part of the theme issue 'Advanced computation in cardiovascular physiology: new challenges and opportunities'.

Identification of 27 abnormalities from multi-lead ECG signals: an ensembled SE_ResNet framework with Sign Loss function

Objective. Cardiovascular disease is a major threat to health and one of the primary causes of death globally. The 12-lead ECG is a cheap and commonly accessible tool to identify cardiac abnormalities. Early and accurate diagnosis will allow early treatment and intervention to prevent severe complications of cardiovascular disease. Our objective is to develop an algorithm that automatically identifies 27 ECG abnormalities from 12-lead ECG databases.Approach. Firstly, a series of pre-processing methods were proposed and applied on various data sources in order to mitigate the problem of data divergence. Secondly, we ensembled two SE_ResNet models and one rule-based model to enhance the performance of various ECG abnormalities' classification. Thirdly, we introduce a Sign Loss to tackle the problem of class imbalance, and thus improve the model's generalizability.Main results. In the PhysioNet/Computing in Cardiology Challenge (2020), our proposed approach achieved a challenge validation score of 0.682, and a full test score of 0.514, placed us 3rd out of 40 in the official ranking.Significance. We proposed an accurate and robust predictive framework that combines deep neural networks and clinical knowledge to automatically classify multiple ECG abnormalities. Our framework is able to identify 27 ECG abnormalities from multi-lead ECG signals regardless of discrepancies in data sources and the imbalance of data labeling. We trained our framework on five datasets and validated it on six datasets from various countries. The outstanding performance demonstrate the effectiveness of our proposed framework.