Design of ear-contactless stethoscope and improvement in the performance of deep learning based on CNN to classify
Tanmay Sinha Roy1, Joyanta Kumar Roy2, Nirupama Mandal3
1Department of Electrical Engineering, Haldia Institute of Technology, Haldia, West Bengal, India. tanmoysinha.roy@gmail.com.
Insights
A novel contactless stethoscope and deep learning models were developed for accurate heart sound classification. The CNN-based Inception Net achieved 99.65% accuracy, outperforming other models for diagnosing valvular heart conditions.
Area of Science:
- Biomedical Engineering
- Artificial Intelligence in Healthcare
- Cardiology
Background:
- Cardiac-related disorders are a growing global health concern, necessitating accurate diagnostic tools.
- The COVID-19 pandemic highlighted challenges in traditional auscultation due to infection control measures, driving the need for contactless methods.
- Accurate classification of cardiovascular diseases is crucial for timely and effective patient management.
Purpose of the Study:
- To design a low-cost, contactless stethoscope for remote auscultation of heart sounds.
- To enhance the performance of deep learning models, specifically Recurrent Neural Networks (RNN) and Convolutional Neural Networks (CNN), for classifying valvular heart diseases.
- To optimize deep learning model performance through hyperparameter tuning for real-time analysis.
Main Methods:
- Development of a contactless stethoscope utilizing a Bluetooth-enabled micro-speaker.
- Implementation and comparison of deep learning models (CNN, RNN, hybrid CNN-RNN, LSTM) using acoustic, time, and frequency domain features.
- Hyperparameter tuning (learning rate, dropout rate, hidden layers) to optimize model performance and learning curves.
- Validation using heart sound recordings from normal and diseased patients from a standard data repository.
Main Results:
- The proposed CNN-based Inception Network model achieved a high accuracy of 99.65% ± 0.06%, with 98.8% ± 0.05% sensitivity and 98.2% ± 0.19% specificity.
- A hybrid CNN-RNN architecture attained 91.17% ± 0.03% accuracy after hyperparameter optimization.
- The LSTM-based RNN model achieved 82.32% ± 0.11% accuracy.
- The CNN-based Inception Net demonstrated superior performance compared to other machine learning algorithms and deep learning models evaluated.
Conclusions:
- The developed contactless stethoscope offers a viable alternative for heart sound auscultation, particularly in situations requiring infection control.
- Optimized deep learning models, especially the CNN-based Inception Network, show significant potential for accurate and automated classification of valvular heart diseases.
- This approach facilitates improved real-time analysis and diagnosis of cardiovascular conditions, contributing to advancements in remote healthcare monitoring.
Abstract:
Cardiac-related disorders are rapidly growing throughout the world. Accurate classification of cardiovascular diseases is an important research topic in healthcare. During COVID-19, auscultating heart sounds was challenging as health workers and doctors wear protective clothing, and direct contact with patients can spread the outbreak. Thus, contactless auscultation of heart sound is necessary. In this paper, a low-cost ear contactless stethoscope is designed where auscultation is done with the help of a bluetooth-enabled micro speaker instead of an earpiece. The PCG recordings are further compared with other standard electronic stethoscopes like Littman 3 M. This work is made to improve the performance of deep learning-based classifiers like recurrent neural networks (RNN) and convolutional neural networks (CNN) for different valvular heart problems using tuning of hyperparameters like learning rate of optimizers, dropout rate, and hidden layer. Hyper-parameter tuning is used to optimize the performances of various deep learning models and their learning curves for real-time analysis. The acoustic, time, and frequency domain features are used in this research. The investigation is made on the heart sounds of normal and diseased patients available from the standard data repository to train the software models. The proposed CNN-based inception network model achieved an accuracy of 99.65 ± 0.06% on the test dataset with a sensitivity of 98.8 ± 0.05% and specificity of 98.2 ± 0.19%. The proposed hybrid CNN-RNN architecture attained 91.17 ± 0.03% accuracy on test data after hyperparameter optimization, whereas the LSTM-based RNN model achieved 82.32 ± 0.11% accuracy. Finally, the evaluated results were compared with machine learning algorithms, and the improved CNN-based Inception Net model is the most effective among others.
