Extraction of Fetal Electrocardiogram by Combining Deep Learning and SVD-ICA-NMF Methods

Ziani, Said; Farhaoui, Yousef; Moutaib, Mohammed

doi:10.26599/bdma.2022.9020035

Cited by 15 publications

(5 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These results surpass the performance reported in the referenced studies, indicating a significant improvement in the signal separation process. The results of this paper can be optimized by incorporating the learning methods and wavelet theory as reported in the references [10,11,12,13]. It is also noted that regularization methods can be effectively applied in the fields of control and energy, as indicated in the references [14,15,16,17,18].…”

Section: Ecg Recordingmentioning

confidence: 87%

Integration of Time-Frequency Analysis and Regularization Technique for Improved Identification of Fetal Electrocardiogram

Ziani,

El Ghmary,

Achmad Rizal

2023

Int. J. Onl. Eng.

View full text Add to dashboard Cite

This research article presents a novel methodology for effectively extracting the fetal electrocardiogram (FECG) from a single-channel signal acquired on the maternal abdomen. The signal comprises a mixture of the FECG, maternal electrocardiogram (MECG), and ambient noise. The central concept involves projecting the signal into higher-dimensional spaces and leveraging the assumption of statistical independence among the constituent components to achieve their separation from the mixture. To accomplish this, singular value decomposition (SVD) is initially applied to the spectrogram, followed by an iterative application of independent component analysis (ICA) on the principal components. The SVD technique contributes to the enhanced separability of each individual component, while ICA facilitates the promotion of statistical independence between the fetal and maternal ECGs. Furthermore, we refine and customize the aforementioned approach specifically for ECG signals by incorporating knowledge of the frequency distribution of the MECG and other inherent ECG characteristics. The effectiveness of the proposed methodology is validated through comprehensive experimental studies, demonstrating its superior accuracy and performance compared to existing techniques.

show abstract

Section: Ecg Recordingmentioning

confidence: 87%

Integration of Time-Frequency Analysis and Regularization Technique for Improved Identification of Fetal Electrocardiogram

Ziani,

El Ghmary,

Achmad Rizal

2023

Int. J. Onl. Eng.

View full text Add to dashboard Cite

show abstract

“…Numerous scholars have utilized CNNs to extract fetal ECG signals from pregnant women [51][52][53][54][55][56][57][58]. However, these algorithms exhibit certain limitations.…”

Section: Multi-feature Fusion Neural Networkmentioning

confidence: 99%

A Non-Invasive Fetal QRS Complex Detection Method Based on a Multi-Feature Fusion Neural Network

Huang,

Yu,

Shan

et al. 2024

Preprint

View full text Add to dashboard Cite

Fetal heart monitoring, as a crucial part of fetal monitoring, can timely and accurately reflect the fetus's health status. To address the issues of high computational cost, inability to observe fetal heart morphology, and insufficient accuracy associated with the traditional method of calculating fetal heart rate using a four-channel maternal electrocardiogram (ECG), a method for extracting fetal QRS complexes from a single-channel non-invasive fetal ECG based on a multi-feature fusion neural network is proposed. Firstly, a signal entropy data quality detection algorithm based on the blind source separation method is designed to select maternal ECG signals that meet the quality requirements from all channel ECG data, followed by data preprocessing operations such as denoising and normalization on the signals. After being segmented by the sliding window method, the maternal ECG signals are calculated as data in four modes: time domain, frequency domain, time-frequency domain, and data eigenvalues. Finally, the deep neural network using three multi-feature fusion strategies—feature-level fusion, decision-level fusion, and model-level fusion—achieves the effect of quickly identifying fetal QRS complexes. Among the proposed networks, the one with the best performance has an accuracy of 95.85%, sensitivity of 97%, specificity of 95%, and PPV (Positive Predictive Value) of 95%. This method, employing the sliding window technique and lightweight deep neural networks, can quickly and accurately identify fetal QRS complexes from single-channel maternal ECG signals, laying the foundation for home-based fetal QRS shape recognition and fetal risk prediction.

show abstract

“…Indeed, one of the applications which has greatly seen the use of transformers is that of medical image analysis [82]. While numerous works have previously aimed at applying a variety of artificial intelligence algorithms towards solving key issues within the realm of medicine, such as COVID-19 detection [83] and the extraction and detection of a fetal electrocardiogram [84,85], with the introduction of transformers for vision, a large number of techniques such as image synthesis/reconstruction, registration, segmentation, detection, and diagnosis have been unlocked. Indeed, as Li et al [86] discuss, the ability of transformers to capture long-range dependencies as well as the scalability of self-attention enables their diverse usage within the medical field.…”

Section: Recent Directionsmentioning

confidence: 99%

A Historical Survey of Advances in Transformer Architectures

Sajun,

Zualkernan,

Sankalpa

2024

Applied Sciences

View full text Add to dashboard Cite

In recent times, transformer-based deep learning models have risen in prominence in the field of machine learning for a variety of tasks such as computer vision and text generation. Given this increased interest, a historical outlook at the development and rapid progression of transformer-based models becomes imperative in order to gain an understanding of the rise of this key architecture. This paper presents a survey of key works related to the early development and implementation of transformer models in various domains such as generative deep learning and as backbones of large language models. Previous works are classified based on their historical approaches, followed by key works in the domain of text-based applications, image-based applications, and miscellaneous applications. A quantitative and qualitative analysis of the various approaches is presented. Additionally, recent directions of transformer-related research such as those in the biomedical and timeseries domains are discussed. Finally, future research opportunities, especially regarding the multi-modality and optimization of the transformer training process, are identified.

show abstract

Extraction of Fetal Electrocardiogram by Combining Deep Learning and SVD-ICA-NMF Methods

Cited by 15 publications

References 41 publications

Integration of Time-Frequency Analysis and Regularization Technique for Improved Identification of Fetal Electrocardiogram

Integration of Time-Frequency Analysis and Regularization Technique for Improved Identification of Fetal Electrocardiogram

A Non-Invasive Fetal QRS Complex Detection Method Based on a Multi-Feature Fusion Neural Network

A Historical Survey of Advances in Transformer Architectures

Contact Info

Product

Resources

About