As the most direct means to diagnose cardiovascular diseases, heart sound classification is attracting the attention of researchers all over the world. Auscultation as a traditional method, its effectiveness largely depends on the physician's clinical experience. Therefore, the development of heart sound classification and recognition in the direction of intelligence has become the mainstream trend. At present, most of the researches mainly focus on the hierarchical feature extraction of signals, which will cause the problem of insufficient feature extraction and affect the accuracy and stability of heart sound signal classification. In order to extract more comprehensive features and improve the recognition accuracy of heart sound signals, this paper constructs a multi-feature fusion recognition network. Firstly, the heart sound signal is preprocessed and converted into GADF image data. Then, the preprocessed one-dimensional signal and corresponding GADF image are input into CNN-ViT1 and CNN-ViT2 channels respectively for feature extraction. The combination of SimAM module and LeakyReLU activation function avoids the problem of "dead neurons" while enhancing the learning ability of nonlinear features and improving the ability to identify key features. Finally, the features extracted from the two channels are spliced in the channel dimension and input into the fully connected layer for classification recognition. The experimental results show that the recognition accuracy of this method is 98.81%, the sensitivity is 98.41%, the specificity is 97.86%, the accuracy is 98.84%, and the F1 score is 97.92%, which provides reliable technical support for the classification and recognition of heat sound signals. The dataset can be accessed at “https://github.com/yaseen21khan/Classification-of-Heart-Sound-Signal-Using-Multiple-Features-/blob/master/README.md”.

Xiong Weihua, Jiang Yufei, Cao Lixian. Multi-feature Fusion Recognition Method of Heart Sound Signal Based on GADF and CNN-ViT. Academic Journal of Computing & Information Science (2025), Vol. 8, Issue 4: 1-10. https://doi.org/10.25236/AJCIS.2025.080401.

[1] Griffel B, Zia M K, Fridman V, et al. Path length entropy analysis of diastolic heart sounds[J]. Computers in biology and medicine, 2013, 43(9): 1154-1166.

[2] Wang Y, Li W, Zhou J, et al. Identification of the normal and abnormal heart sounds using wavelet-time entropy features based on OMS-WPD[J]. Future Generation Computer Systems, 2014, 37: 488-495.

[3] Zheng Y, Guo X, Ding X. A novel hybrid energy fraction and entropy-based approach for systolic heart murmurs identification[J]. Expert Systems with Applications, 2015, 42(5): 2710-2721.

[4] Deng S W, Han J Q. Towards heart sound classification without segmentation via autocorrelation feature and diffusion maps[J]. Future Generation Computer Systems, 2016, 60: 13-21.

[5] Li P, Ge J, Li H. Lysine acetyltransferases and lysine deacetylases as targets for cardiovascular disease[J]. Nature Reviews Cardiology, 2020, 17(2): 96-115.

[6] Roth G A, Mensah G A, Johnson C O, et al. Global burden of cardiovascular diseases and risk factors, 1990–2019: update from the GBD 2019 study[J]. Journal of the American college of cardiology, 2020, 76(25): 2982-3021.

[7] Tsao C W, Aday A W, Almarzooq Z I, et al. Heart disease and stroke statistics—2022 update: a report from the American Heart Association[J]. Circulation, 2022, 145(8): e153-e639.

[8] Milani M G M, Abas P E, De Silva L C, et al. Abnormal heart sound classification using phonocardiography signals[J]. Smart Health, 2021, 21: 100194.

[9] Mei N, Wang H, Zhang Y, et al. Classification of heart sounds based on quality assessment and wavelet scattering transform[J]. Computers in biology and medicine, 2021, 137: 104814.

[10] Chen J, Guo Z, Xu X, et al. A robust deep learning framework based on spectrograms for heart sound classification[J]. IEEE/ACM Transactions on Computational Biology and Bioinformatics, 2023.

[11] Wang Y, Yang X, Qian X, et al. Assistive diagnostic technology for congenital heart disease based on fusion features and deep learning[J]. Frontiers in Physiology, 2023, 14: 1310434.

[12] Zhang H, Zhang P, Wang Z, et al. Multi-feature decision fusion network for heart sound abnormality detection and classification[J]. IEEE Journal of Biomedical and Health Informatics, 2023, 28(3): 1386-1397.

[13] Zhang X, Liu X, Liu G. A Heart Sound Signal Classification Method Based on the Mixed Characteristics of Mel Cepstrum Coefficient and Second-Order Spectrum[J]. Circuits, Systems, and Signal Processing, 2024, 43(6): 3533-3552.

[14] Wang Z , Oates T .Imaging Time-Series to Improve Classification and Imputation //International Conference on Artificial Intelligence. AAAI Press, 2015: 3308-4132.

[15] Wang Z, Oates T. Imaging time-series to improve classification and imputation[J]. arXiv preprint arXiv:1506.00327, 2015.

[16] Dosovitskiy A, Beyer L, Kolesnikov A, et al. An image is worth 16x16 words: Transformers for image recognition at scale[J]. arXiv preprint arXiv:2010.11929, 2020.

[17] Yang L, Zhang R Y, Li L, et al. Simam: A simple, parameter-free attention module for convolutional neural networks[C]//International conference on machine learning. PMLR, 2021: 11863-11874.

[18] Maniatopoulos A, Mitianoudis N. Learnable leaky relu (LeLeLU): An alternative accuracy-optimized activation function[J]. Information, 2021, 12(12): 513.

[19] Li S, Li F, Tang S, et al. A review of computer‐aided heart sound detection techniques[J]. BioMed research international, 2020, 2020(1): 5846191.

[20] Chen Y, Wei S, Zhang Y. Classification of heart sounds based on the combination of the modified frequency wavelet transform and convolutional neural network[J]. Medical & Biological Engineering & Computing, 2020, 58: 2039-2047.

[21] Chen P, Zhang Q. Classification of heart sounds using discrete time-frequency energy feature based on S transform and the wavelet threshold denoising[J]. Biomedical Signal Processing and Control, 2020, 57: 101684.

[22] Varghees V N, Ramachandran K I. Effective heart sound segmentation and murmur classification using empirical wavelet transform and instantaneous phase for electronic stethoscope[J]. IEEE Sensors Journal, 2017, 17(12): 3861-3872.

[23] Yingwei L, Sundararajan N, Saratchandran P. Performance evaluation of a sequential minimal radial basis function (RBF) neural network learning algorithm[J]. IEEE Transactions on neural networks, 1998, 9(2): 308-318.

[24] Augusteijn M F, Clemens L E, Shaw K A. Performance evaluation of texture measures for ground cover identification in satellite images by means of a neural network classifier[J]. IEEE Transactions on Geoscience and Remote Sensing, 1995, 33(3): 616-626.

[25] Zhou X, Guo X, Zheng Y, et al. Detection of coronary heart disease based on MFCC characteristics of heart sound[J]. Applied Acoustics, 2023, 212: 109583.

[26] Deng E, Jia Y, Zhu G, et al. Heart sound signals classification with image conversion employed[J]. Electronics, 2024, 13(7): 1179.

[27] Wang J J, Xiong F L. Research on the recognition method of heart sound signal based on improved MFCC and CNN. Computer Measurement and Control, 2024, 32(12): 201-207+215.

[28] Ismail Fawaz H, Lucas B, Forestier G, et al. Inceptiontime: Finding alexnet for time series classification[J]. Data Mining and Knowledge Discovery, 2020, 34(6): 1936-1962.