Noisy Neonatal Chest Sound Separation for High-Quality Heart and Lung Sounds

Grooby, Ethan; Sitaula, Chiranjibi; Fattahi, Davood; Sameni, Reza; Tan, Kenneth; Zhou, Lindsay; King, Alicia; Ramanathan, Ashwin; Malhotra, Atul; Dumont, G.A.; Marzbanrad, Faezeh

doi:10.1109/jbhi.2022.3215995

Cited by 11 publications

(20 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(3) Calculate Pj=|DFT(zj mix )| for j ranging from 1 to M. (11) Decode Z hear and Z lung to obtain the initial separated heart sound and lung sound. (12) Apply VMD processing to the obtained heart sound and lung sound to obtain pure heart sound and pure lung sound.…”

Section: Figure 4 Dae-nmf-vmd Algorithm Flow Chartmentioning

confidence: 99%

Research on Heart and Lung Sound Separation Method Based on DAE-NMF-VMD

sun,

CHEN,

ZHANG

et al. 2023

Preprint

View full text Add to dashboard Cite

Auscultation is the most effective method for diagnosing cardiovascular and respiratory diseases. However, stethoscopes typically capture mixed signals of heart and lung sounds, which can affect the auscultation effect of doctors.Therefore, the efficient separation of mixed heart and lung sound signals plays a crucial role in improving the diagnosis of cardiovascular and respiratory diseases. In this paper, we propose a blind source separation method for heart and lung sounds based on Deep Autoencoder (DAE), Non-Negative Matrix Factorization (NMF), and Variational Mode Decomposition (VMD).Firstly, DAE is employed to extract highly informative features from the heart and lung sound signals. Subsequently, NMF clustering is applied to group the heart and lung sounds based on their distinct periodicities, achieving the separation of the mixed heart and lung sounds. Finally, Variational Mode Decomposition is used for denoising the separated signals. Experimental results demonstrate that the proposed method effectively separates heart and lung sound signals and exhibits significant advantages in terms of standardized evaluation metrics when compared to Non-Negative Matrix Factorization methods and DAE-NMF algorithms without denoising.

show abstract

Section: Figure 4 Dae-nmf-vmd Algorithm Flow Chartmentioning

confidence: 99%

Research on Heart and Lung Sound Separation Method Based on DAE-NMF-VMD

sun,

CHEN,

ZHANG

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

Section: Preprocessingmentioning

confidence: 99%

“…To remove noise such as lung sounds, and the environment, we used the sound separation method developed in our previous work [5]. This method implements non-negative matrix factorisation (NMF) with a reference database of clean heart and noise sounds, to separate the PCG recording into these respective components [5]. The PCG recording was represented in the time-frequency domain using Short-time Fourier transform with a window size of 512 samples, hop size of 256 samples and a Fast Fourier transform size of 1024 points.…”

Section: Preprocessingmentioning

confidence: 99%

“…The PCG recording was represented in the time-frequency domain using Short-time Fourier transform with a window size of 512 samples, hop size of 256 samples and a Fast Fourier transform size of 1024 points. Sound separation using NMF, Kullback-Leibler divergence with a sparsity of 0.1 was used in the cost function and 20 basis components for each heart, and noise sound, respectively [5].…”

Section: Preprocessingmentioning

confidence: 99%

See 1 more Smart Citation

A Fusion of Handcrafted Features and Deep Learning Classifiers for Heart Murmur Detection

Imran¹,

Grooby²,

Sitaula³

et al. 2022

Computing in Cardiology Conference (CinC)

View full text Add to dashboard Cite

As part of George B. Moody Physionet Challenge 2022, our team Melbourne Kangas, proposed an algorithm for identifying abnormal heart sounds from paediatric phonocardiograms (PCGs). We developed a Deep Learning (DL) approach and a handcrafted feature-based approach. The DL classifier was based on bidirectional long-short-termmemory and Mel-frequency cepstrum coefficients from raw PCG signals. The feature-based approach used nonnegative matrix factorisation to denoise PCG signals and then extracted the features based on the whole and segmented recordings, followed by feature selection. A random under-sampling boosting classifier for murmur classification and robust boosting classifier for outcome classification were given the subset of features. The feature-based performed better than the DL classifiers on the validation set. The feature-based classifier received a weighted accuracy of 0.632 (29th out of 41 teams) and a challenge cost of 11,735 (3rd out of 39 teams) on the test set. Decision fusion of the two approaches decreased 10-fold crossvalidation results.

show abstract

“…[4][5][6][7][8] Furthermore, in neonates, even infant-sized acoustic stethoscopes can be unreliable and produce poor-quality sounds due to neonatal factors such as their small size, fast heart and respiratory rates, irregular breathing patterns, and noise interference (e.g., crying and respiratory support noise). [9][10][11][12][13] This can hinder accurate clinician analysis and may lead to delayed diagnosis and management of neonatal conditions (Fig. 1).…”

Section: Introductionmentioning

confidence: 99%

Digital Stethoscope Use in Neonates: A Systematic Review

Roff,

Slifirski,

Grooby

et al. 2023

Newborn

Self Cite

View full text Add to dashboard Cite

Aim: To assess the evidence for the use of digital stethoscopes in neonates and evaluate whether they are effective, appropriate, and advantageous for neonatal auscultation. Methods: A systematic review and narrative synthesis of studies published between January 1, 1990 and May 29, 2023 was conducted following searches of MEDLINE, Embase, PubMed, Scopus, and Google Scholar databases, as well as trial registries. Results: Of 3,852 records identified, a total of 41 papers were eligible and included in the narrative synthesis. Thirteen records were non-full-text articles, either in the form of journal letters or conference abstracts, and these were included separately for completion purposes but may be unreliable. Twenty eight papers were full-text articles and were included in a full qualitative analysis. Digital stethoscopes have been studied in neonatology across various clinical areas, including artificial intelligence for sound quality assessment and chest sound separation (n = 5), cardiovascular sounds (n = 11), respiratory sounds (n = 4), bowel sounds (n = 4), swallowing sounds (n = 2), and telemedicine (n = 2). This paper discusses the potential utility of digital stethoscope technology for the interpretation of neonatal sounds for both humans and artificial intelligence. The limitations of current devices are also assessed. Conclusions:The utilization of digital stethoscopes in neonatology is an emerging field with a wide range of potential applications, which has the capacity to advance neonatal auscultation. Artificial intelligence and digital stethoscope technology offer novel objective avenues for automatic pathological sound detection. Further, digital stethoscopes may improve our scientific understanding of normal neonatal physiology and can be employed in telemedicine to facilitate remote medical access. Digital stethoscopes can also provide phonocardiograms, enabling enhanced interpretation of neonatal cardiac sounds. However, current digital stethoscopes necessitate refinement as they consistently produce low-quality sounds when used on neonates.

show abstract

Noisy Neonatal Chest Sound Separation for High-Quality Heart and Lung Sounds

Cited by 11 publications

References 26 publications

Research on Heart and Lung Sound Separation Method Based on DAE-NMF-VMD

Research on Heart and Lung Sound Separation Method Based on DAE-NMF-VMD

A Fusion of Handcrafted Features and Deep Learning Classifiers for Heart Murmur Detection

Digital Stethoscope Use in Neonates: A Systematic Review

Contact Info

Product

Resources

About