heart-murmur-analysis 0.3.0

Heartbeat analysis using Deep Learning, Signal Processing, and AI Agents.

🎵 Heart Murmur Detection with LSTM

A deep learning application that uses LSTM neural networks to detect heart murmurs from audio recordings. The application provides a user-friendly Streamlit interface for uploading audio files and getting real-time predictions with signal processing.

---

---

Quick Start

Prerequisites

• Python 3.9 or higher
• Windows 10/11 (PowerShell)
• Git (optional, for cloning)

Installation Steps

1. Create Virtual Environment

   python -m venv hvenv
   

2. Activate Virtual Environment

   hvenv\Scripts\activate
   

   You should see (hvenv) at the beginning of your command prompt.

3. Install the package

   pip install heart-murmur-analysis
   

4. Run the Application

   streamlit run test_file.py
   

   Download the test_file.py

5. Access the Application

   • Open your web browser
   • Navigate to http://localhost:8501
   • Upload a WAV or MP3 audio file
   • Get instant heart murmur predictions with signal processing !

🧠 Concepts & Terminologies

Heart Sounds

• S1 ("Lub"): Closing of the mitral and tricuspid valves.
• S2 ("Dub"): Closing of the aortic and pulmonary valves.
• Murmur: Abnormal sound caused by turbulent blood flow.
• Artifact: Noise unrelated to actual heart sounds (e.g., movement, microphone noise).

---

🛠️ Signal Processing Pipeline

1. Preprocessing

• Bandpass Filter (20–500 Hz): Retains important heart sound frequencies while removing noise.
• Envelope Detection: Smooths the waveform to highlight heartbeats for peak detection.

2. Extracted Metrics

The application extracts several clinical metrics to assist in diagnosis:

Metric	Description
BPM	Average heart rate in beats per minute.
IBV/IBI	Inter-Beat Interval – time between successive heartbeats.
HRV	Heart Rate Variability (SDNN, RMSSD, pNN50).
SNR	Signal-to-Noise Ratio (recording quality).
S1/S2 Ratio	Helps distinguish murmurs from normal beats.
Spectral Energy	Energy distribution below 200 Hz.

---

📊 Visualizations

• Waveform with Peaks: Visualizes detected beats.
• Spectrogram: Frequency vs. time representation.
• Poincaré Plot: Nonlinear HRV visualization.

---

🤖 Model Architecture

The LSTM model uses a hybrid CNN-LSTM architecture:

• Input: Raw audio data (52 timesteps).
• CNN Layers: 3 Conv1D layers for feature extraction.
• LSTM Layers: 2 LSTM layers for temporal modeling.
• Output: 3 classes (Normal, Murmur, Artifact).

---

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test thoroughly
5. Submit a pull request

Happy Heart Murmur Detection!