Classification of Short Segment Pediatric Heart Sounds Based on a Transformer-Based Convolutional Neural Network

Md Hassanuzzaman,Nurul Akhtar Hasan,Mohammad Abdullah Al Mamun,Khawza I Ahmed,Ahsan H Khandoker,Raqibul Mostafa

2024-03-31

Abstract:Congenital anomalies arising as a result of a defect in the structure of the heart and great vessels are known as congenital heart diseases or CHDs. A PCG can provide essential details about the mechanical conduction system of the heart and point out specific patterns linked to different kinds of CHD. This study aims to investigate the minimum signal duration required for the automatic classification of heart sounds. This study also investigated the optimum signal quality assessment indicator (Root Mean Square of Successive Differences) RMSSD and (Zero Crossings Rate) ZCR value. Mel-frequency cepstral coefficients (MFCCs) based feature is used as an input to build a Transformer-Based residual one-dimensional convolutional neural network, which is then used for classifying the heart sound. The study showed that 0.4 is the ideal threshold for getting suitable signals for the RMSSD and ZCR indicators. Moreover, a minimum signal length of 5s is required for effective heart sound classification. It also shows that a shorter signal (3 s heart sound) does not have enough information to categorize heart sounds accurately, and the longer signal (15 s heart sound) may contain more noise. The best accuracy, 93.69%, is obtained for the 5s signal to distinguish the heart sound.

Sound,Machine Learning,Audio and Speech Processing

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **Determine the shortest signal duration for automatic classification of pediatric heart sounds, and optimize the signal quality evaluation indicators to improve classification accuracy**.

Specifically, the paper focuses on the following aspects:

1. **Diagnosis of congenital heart disease (CHD)**:
- Congenital heart disease is a class of diseases caused by structural defects in the heart and large blood vessels. Phonocardiogram (PCG) can be used to obtain important information about the mechanical conduction system of the heart and identify specific patterns associated with different types of CHD.
- Traditional heart sound analysis methods rely on experienced medical professionals and are easily affected by background noise, environmental noise, and motion artifacts, resulting in difficulties in diagnosis.

2. **The need for automatic classification of heart sounds**:
- In order to achieve automated heart sound classification, it is necessary to determine the minimum signal duration to ensure that the classification model can extract sufficient feature information from limited heart sound data.
- At the same time, the study also explored the optimal signal quality evaluation indicators (such as RMSSD and ZCR) to ensure the data quality of the input model.

3. **Experimental design and methods**:
- The study used a Transformer - based convolutional neural network (CNN) model, combined with Mel - frequency cepstral coefficients (MFCCs) as input features, to classify heart sounds of different durations.
- The experimental results show that a signal length of 5 seconds is the most suitable for effective classification, and the best signal quality can be obtained when RMSSD and ZCR are 0.4 respectively.

In summary, this paper aims to optimize the signal duration and quality evaluation indicators for heart sound classification through deep - learning techniques, especially the Transformer - based 1D CNN model, thereby improving the accuracy of automatic diagnosis of congenital heart disease.