Classification of normal/abnormal PCG recordings using a time–frequency approach

Hanie Hazeri, Pega Zarjam, Ghasem Azemi*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

7 Citations (Scopus)


The early and accurate diagnosis of cardiovascular diseases (CVDs) are of great importance as they allow early and proper medical treatment and therefore result in reducing the chance of the CVDs being developed to an acute level. In medical procedures, the first step in examining the cardiovascular function is the auscultation of the heart. However, the correct medical diagnosis based on the heart sounds through a stethoscope requires a lot of expertise and, in some cases, needs referral of the patient to a cardiologist. This is not only time-consuming but also imposes a financial burden on the medical system. Thus, automated detection and analysis of the recorded heart sound auscultation has received a lot of attentions in recent years. This study presents a new time–frequency (T–F) based approach for classifying phonocardiogram (PCG) signals into normal and abnormal. In the proposed methodology, each PCG recording is first segmented into the 4 fundamental heart cycles, i.e. S1, systole, S2, and diastole. From each state, a set of T–F features are extracted with the aim of identifying their characteristics in the T–F domain. The features are then applied to a support vector machine to classify the PCG signal into normal or abnormal. The performance of the proposed method is evaluated using the 2016 PhysioNet challenge database and compared with that of the best performing existing methods. The experimental results using tenfold cross-validation show that the proposed method outperforms the existing methods in terms of sensitivity, specificity, and accuracy.

Original languageEnglish
Pages (from-to)459-465
Number of pages7
JournalAnalog Integrated Circuits and Signal Processing
Issue number2
Publication statusPublished - Nov 2021
Externally publishedYes


  • Cardiovascular disease
  • PCG classification
  • PCG signals
  • Time–frequency feature extraction


Dive into the research topics of 'Classification of normal/abnormal PCG recordings using a time–frequency approach'. Together they form a unique fingerprint.

Cite this