Computer methods and programs in biomedicine · 2026
Multidimensional decomposition framework for electrocardiographic interference, noise and artifacts removal from diaphragmatic electromyography.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Grońska Gabriela, Peri Elisabetta, Long Xi, Overeem Sebastiaan, van Dijk Johannes P, Mischi Massimo
Аннотация
BACKGROUND AND OBJECTIVE: Accurate extraction of respiratory diaphragmatic electromyography (dEMG) is hindered by electrocardiographic (ECG) interference, motion artifacts, and noise. Existing methods, although effective, may remove physiologically relevant information. To enable robust ECG and artifact suppression while maintaining respiratory-related amplitude variations, a higher-order singular value decomposition (HO-SVD) based approach was developed.
METHODS: The HO-SVD algorithm exploits the multidimensional structure of dEMG recordings by jointly decomposing the temporal, overlapped QRS complexes, and spatial (channel) dimensions. The method enables targeted suppression of coherent and repetitive artifacts prior to signal reconstruction. Performance was evaluated on three synthetic datasets and on clinical data from ten patients with obstructive sleep apnea, using a previously developed two-dimensional SVD algorithm (SVDRI) as a reference.
RESULTS: Across synthetic datasets, HO-SVD consistently outperformed SVDRI in all evaluation metrics. In clean conditions, HO-SVD improved the median signal-to-noise ratio (SNR) by 27 dB (131%) relative to the original signal. In datasets with mild motion artifacts and complex muscle interference, median SNR improvements of up to 19 dB (88%) were observed. The symmetric mean absolute error and mean frequency deviation were significantly lower for HO-SVD across all input SNRs (p<0.0001) compared to SVDRI. In clinical data, HO-SVD achieved a significantly higher correlation with reference esophageal pressure (p<0.02), greater ECG suppression (p<0.0001), and fewer outliers (p<0.0001) than SVDRI.
CONCLUSION: These results demonstrate that incorporating a third (spatial) dimension and adaptive thresholding across tensor modes enables robust removal of ECG interference and motion artifacts while preserving respiratory dEMG modulation. The proposed HO-SVD framework offers a flexible and effective approach for multichannel biomedical signal denoising.
METHODS: The HO-SVD algorithm exploits the multidimensional structure of dEMG recordings by jointly decomposing the temporal, overlapped QRS complexes, and spatial (channel) dimensions. The method enables targeted suppression of coherent and repetitive artifacts prior to signal reconstruction. Performance was evaluated on three synthetic datasets and on clinical data from ten patients with obstructive sleep apnea, using a previously developed two-dimensional SVD algorithm (SVDRI) as a reference.
RESULTS: Across synthetic datasets, HO-SVD consistently outperformed SVDRI in all evaluation metrics. In clean conditions, HO-SVD improved the median signal-to-noise ratio (SNR) by 27 dB (131%) relative to the original signal. In datasets with mild motion artifacts and complex muscle interference, median SNR improvements of up to 19 dB (88%) were observed. The symmetric mean absolute error and mean frequency deviation were significantly lower for HO-SVD across all input SNRs (p<0.0001) compared to SVDRI. In clinical data, HO-SVD achieved a significantly higher correlation with reference esophageal pressure (p<0.02), greater ECG suppression (p<0.0001), and fewer outliers (p<0.0001) than SVDRI.
CONCLUSION: These results demonstrate that incorporating a third (spatial) dimension and adaptive thresholding across tensor modes enables robust removal of ECG interference and motion artifacts while preserving respiratory dEMG modulation. The proposed HO-SVD framework offers a flexible and effective approach for multichannel biomedical signal denoising.