A Spectral Centroid Based Analysis of Heart sounds for Disease Detection Using Machine Learning

Ferdoushi, Munia; Paul, Mala; Fattah, Shaikh Anowarul

doi:10.1109/wiecon-ece48653.2019.9019975

Cited by 4 publications

(2 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In 2019, researchers (Ferdoushi et al, 2019) making the final prediction using a voting process. The authors implemented two late fusion techniques in their research, namely, majority voting (MV), and margin sampling voting (MSV).…”

Section: Spectrogrammentioning

confidence: 99%

“…In 2019, researchers (Ferdoushi et al, 2019) used PASCAL Classifying Heart Sounds Challenge 2011 database, clipped all PCG signals to 5 s from the data set, and computed the STFT after sampling the signal at the rate of 22,050 Hz. The time‐domain features such as zero crossings were extracted, while the rest were frequency domain features such as spectral roll‐off, spectral bandwidth, MFCC, and spectral centroids obtained from STFT.…”

Section: Literature Reviewmentioning

confidence: 99%

See 1 more Smart Citation

Classifying the heart sound signals using textural‐based features for an efficient decision support system

2023

View full text Add to dashboard Cite

Cardiovascular diseases have surpassed cancer as the leading cause of death on the planet today. Numerous decision‐making systems with computer‐assisted support have been developed to assist cardiologists to detect heart disease, and thus, lowering the mortality rate. The purpose of this research is to classify audio signals received from the heart as normal or abnormal. The PhysioNet Computing in Cardiology (CinC) 2016 benchmark dataset, popularly known as PhysioNet 2016, has been used to validate the proposed methodology presented here. PhysioNet 2016 contains a total of 3200 phonocardiogram (PCG) recordings divided into sub‐datasets A‐F. The state‐of‐the‐art studies conducted till date have not considered the harmonic details of the beat that can be extracted from its equivalent chromagram image. In this work, textural features such as linear binary pattern (LBP), adaptive‐LBP, and ring‐LBP have been extracted from the existing spectrogram and combined with the features extracted from the chromagram. It has been observed that the combination of features extracted from both the image variants has resulted in a greater accuracy as compared to the scenario where researchers were using only the spectrogram. The experiment yielded the mean accuracy, precision, and F1‐score as 94.87, 93.11, and 95.273, respectively. The heart sound classification models employ spectrogram, scalogram, and mel‐spectrogram images to view and analyse the acoustic properties of a PCG signal. Although these visual tools provide useful information about the signal, yet they are unable to distinguish between pitch and resonance in heart sound generation. However, this paper proposes an alternative approach of heart sound signal representation that allows for a more precise measure of pitch‐related changes in the heart sound. Its results highlight the significance of extracting textural features from time‐chroma representation (i.e., chromagram) of PCG signals that have not been explored yet in the domain related to classification of heart sound signals as normal and abnormal.

show abstract

Section: Spectrogrammentioning

confidence: 99%

Section: Literature Reviewmentioning

confidence: 99%