Parallel use of a convolutional neural network and bagged tree ensemble for the classification of Holter ECG

Plešinger, Filip; Nejedlý, Petr; Viščor, Ivo; Halámek, Josef; Jurák, Pavel

doi:10.1088/1361-6579/aad9ee

Cited by 44 publications

(28 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since 11 features were extracted from the EEG signal, we also extracted 7 features from ECG activity as heart rate or its changes; the rest were based on EMG and airflow signals. For QRS detection, we used a method designed for 1-lead Holter-ECG processing [4].…”

Section: Used Featuresmentioning

confidence: 99%

Automated Sleep Arousal Detection Based on EEG Envelograms

Plešinger

Viščor

Nejedlý

et al. 2018

2018 Computing in Cardiology Conference (CinC)

Self Cite

View full text Add to dashboard Cite

Background: Sleep arousal is basically described as a shift in EEG activity in frequencies > 16 Hz for a duration of > 3 sec (by the American Sleep Disorders Association-ASDA). The number of these arousals during sleep is a reflection of sleep quality. In accordance with the PhysioNet/CinC Challenge 2018, we present a method for automatic detection of arousals in polysomnographic recordings. Method: Each file in the training dataset (N=994) has defined "Target Arousal Regions" (TAR, median length 33 seconds); however, arousals were usually located in the right half of these TARs. We built a method detecting EEG frequency shift to locate arousals inside ARs: envelograms (14-20, 16-25 and 20-40 Hz) were inspected in a 3-sec floating window for an increase against a 10sec background. We then extracted 133,573 blocks with such a shift from TARs (N=38,628) as well as outside TARs (N=94,945). We extracted 23 features from these blocks (how many EEG channels/frequency bands EEG frequency shift; heart rate before/during arousal; airflow and EMG changes) and trained a bagged tree ensemble model (70/30 % hold-out). Results: The method showed AUPRC 0.27 on a training set and AUPRC 0.20 on a testing set (N=989).

show abstract

Section: Used Featuresmentioning

confidence: 99%

Automated Sleep Arousal Detection Based on EEG Envelograms

Plešinger

Viščor

Nejedlý

et al. 2018

2018 Computing in Cardiology Conference (CinC)

Self Cite

View full text Add to dashboard Cite

show abstract

“…With this performance we obtained the third place out of 84 participants in the final rank of the follow-up challenge (first place: F 1n = 0.921; F 1a = 0.857; F 1o = 0.766 and F 1 = 0.848 [40]; second place: F 1n = 0.908; F 1a = 0.841; F 1o = 0.745 and F 1 = 0.832 [41] (https://groups.google.com/forum/ #!topic/physionet-challenges/qA2iUfQmRtc). A comparative summary of the algorithms [40][41][42][43], which obtained the top five scores in the follow-up phase of the Physionet Challenge 2017, is reported in Table 3. [42] 150 Multi-layer Cascaded Binary F 1 = 0.8294 Plesinger [43] 60 Neural Netwok + Bagged Tree Ensemble F 1 = 0.8278 Notably, our performance on the hidden test set of the official challenge phase was: F 1n = 0.911; F 1a = 0.784; F 1o = 0.739 and F 1 = 0.812 (obtaining the 12th place in the official ranking list).…”

Section: Physionet Challenge 2017 Databasementioning

confidence: 99%

“…A comparative summary of the algorithms [40][41][42][43], which obtained the top five scores in the follow-up phase of the Physionet Challenge 2017, is reported in Table 3. [42] 150 Multi-layer Cascaded Binary F 1 = 0.8294 Plesinger [43] 60 Neural Netwok + Bagged Tree Ensemble F 1 = 0.8278 Notably, our performance on the hidden test set of the official challenge phase was: F 1n = 0.911; F 1a = 0.784; F 1o = 0.739 and F 1 = 0.812 (obtaining the 12th place in the official ranking list). The increase in performance achieved from the official phase to the follow-up phase suggests the importance of having reliable annotations for the algorithm training and evaluation.…”

Section: Physionet Challenge 2017 Databasementioning

confidence: 99%

Automatic Detection of Atrial Fibrillation and Other Arrhythmias in ECG Recordings Acquired by a Smartphone Device

Billeci

Costi²,

Lombardi³

et al. 2018

Electronics

View full text Add to dashboard Cite

Atrial fibrillation (AF) is the most common cardiac disease and is associated with other cardiac complications. Few attempts have been made for discriminating AF from other arrhythmias and noise. The aim of this study is to present a novel approach for such a classification in short ECG recordings acquired using a smartphone device. The implemented algorithm was tested on the Physionet Computing in Cardiology Challenge 2017 Database and, for the purpose of comparison, on the MIT-BH AF database. After feature extraction, the stepwise linear discriminant analysis for feature selection was used. The Least Square Support Vector Machine classifier was trained and cross-validated on the available dataset of the Challenge 2017. The best performance was obtained with a total of 30 features. The algorithm produced the following performance: F1 Normal rhythm = 0.92; F1 AF rhythm: 0.82; F1 Other rhythm = 0.75; Global F1 = 0.83, obtaining the third best result in the follow-up phase of the Physionet Challenge. On the MIT-BH ADF database the algorithm gave the following performance: F1 Normal rhythm = 0.98; F1 AF rhythm: 0.99; Global F1 = 0.98. Since the algorithm reliably detect AF and other rhythms in smartphone ECG recordings, it could be applied for personal health monitoring systems.

show abstract

“…Convolutional Neural Networks (CNNs) have demonstrated usefulness in a wide variety of industrial and scientific fields, including image recognition 11 , speech recognition 12 , biological signal processing and reinforcement learning 13 . CNNs have proven to be superior to traditional signal processing techniques in ECG and polysomnography classification during several challenges 14,15 and have been used in variety of EEG processing tasks 16 , for example iEEG noise detection 17 , epileptic seizure detection 18 and seizure prediction 19 .…”

Section: Introductionmentioning

confidence: 99%

Exploiting Graphoelements and Convolutional Neural Networks with Long Short Term Memory for Classification of the Human Electroencephalogram

Nejedlý

Kremen

Sladky

et al. 2019

Sci Rep

Self Cite

View full text Add to dashboard Cite

The electroencephalogram (EEG) is a cornerstone of neurophysiological research and clinical neurology. Historically, the classification of EEG as showing normal physiological or abnormal pathological activity has been performed by expert visual review. The potential value of unbiased, automated EEG classification has long been recognized, and in recent years the application of machine learning methods has received significant attention. A variety of solutions using convolutional neural networks (CNN) for EEG classification have emerged with impressive results. However, interpretation of CNN results and their connection with underlying basic electrophysiology has been unclear. This paper proposes a CNN architecture, which enables interpretation of intracranial EEG (iEEG) transients driving classification of brain activity as normal, pathological or artifactual. The goal is accomplished using CNN with long short-term memory (LSTM). We show that the method allows the visualization of iEEG graphoelements with the highest contribution to the final classification result using a classification heatmap and thus enables review of the raw iEEG data and interpret the decision of the model by electrophysiology means.

show abstract

Parallel use of a convolutional neural network and bagged tree ensemble for the classification of Holter ECG

Abstract: This F1 score led to shared rank #2 in the follow-up PhysioNet/CinC Challenge 2017 ranking.

Cited by 44 publications

References 33 publications

Automated Sleep Arousal Detection Based on EEG Envelograms

Automated Sleep Arousal Detection Based on EEG Envelograms

Automatic Detection of Atrial Fibrillation and Other Arrhythmias in ECG Recordings Acquired by a Smartphone Device

Exploiting Graphoelements and Convolutional Neural Networks with Long Short Term Memory for Classification of the Human Electroencephalogram

Contact Info

Product

Resources

About