Abstract
Early diagnosis of heart diseases bears a major role in saving lives. Presence of spurious extra-frequency components, termed as murmurs within the phonocardiography record may be indicative of valvular disorders like stenosis, lesions or regurgitation. It is difficult to identify the subtitle spectral components of murmurs through subjective audition. In this paper, a technique is proposed to detect the presence of murmur from the heart signal by analyzing their non-stationarity behavior by using autocorrelation based features namely, Standard Error (SE) of Auto-Correlation Function (ACF) and absolute deviation of SE from the reciprocal of the square root of number of samples (β). The selected features corresponding to normal and murmur differ with a ‘P’ value of 1.80 x10-14 (dataset 1) and 2.20 x10-76 (dataset 2) for SE and β, respectively. It is found that SE and β could effectively distinguish normal and murmur with 100% accuracy, sensitivity, and specificity.
Highlights
Diagnosis of heart diseases bears a major role in saving lives
The wave shape of preprocessed heart signal corresponding to normal (fig.4 (a-b)) is less random and their average amplitude is comparatively lesser than the wave shape of murmur
Compared to the methods presented in table 5, we proposed a technique to detect heart sound and to differentiate them as normal or murmur by using time domain features, exclusively autocorrelation features such as Standard Error (SE) and â by utilizing the PCG records collected from two publicly available databases namely, Pascal HSDB and Physionet HSDB.The two features are successfully tested on the two datasets, statistically evaluated and separability among the features is examined
Summary
Diagnosis of heart diseases bears a major role in saving lives. Presence of spurious extra-frequency components, termed as murmurs within the phonocardiography record may be indicative of valvular disorders like stenosis, lesions or regurgitation. Identification of Still’s murmur in children with a sensitivity of 84-94% and a specificity of 9199% They used time domain features (average Shannon energy and envelope detection) and frequency domain features (spectral width and peak frequency) of S1 and S2 heart sounds. Han [3] proposed a method based on the autocorrelation features like Sub-band autocorrelation function These features were extracted from the sub-band envelopes derived from the sub-band coefficients of PCG signals obtained using Discrete Wavelet Transform (DWT). An algorithm was proposed to extract features such as maximum amplitude values from consecutive S1 and S2 and vice versa These features were normalized and input into Artificial Neural Network (ANN) classifier (average success rate of 94 %.). The system was able to differentiate normal and murmur with a success rate of 98.04%
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.