Python OEphys

python-oephys is a comprehensive Python toolkit for working with Open Ephys devices and electrophysiology data. From file loading to real-time ZMQ streaming, signal processing to machine learning, and visualization tools—everything you need for high-density neural data analysis in one package.

---

✨ Key Features

• 📁 File I/O: Robust support for Open Ephys Binary (.oebin) and .npz formats
• 🔴 Real-time Streaming: Seamless integration with the Open Ephys GUI via ZMQ
• 🎛️ Signal Processing: Filtering (Bandpass, Notch), Channel QC, and synchronization
• 🤖 Machine Learning: Hybrid CNN-LSTM models for real-time gesture recognition
• 📊 Visualization: Real-time EMG viewer, offline analysis, and trial segmentation tools
• 🚀 Performance: Optimized for low-latency real-time applications

---

📦 Installation

From TestPyPI (Current Development Release)

pip install --index-url https://test.pypi.org/simple/ --no-deps python-oephys

From Source

git clone https://github.com/Neuro-Mechatronics-Interfaces/python-open-ephys.git
cd python-open-ephys
pip install -e .

Optional Extras

• GUI: pip install 'python-oephys[gui]' (PyQt5, pyqtgraph)
• ML: pip install 'python-oephys[ml]' (PyTorch, scikit-learn)
• Docs: pip install 'python-oephys[docs]' (Sphinx)

---

🚀 Getting Started

Load and Filter Data

from pyoephys.io import load_open_ephys_session
from pyoephys.processing import filter_emg

# Load session
sess = load_open_ephys_session('path/to/recording.oebin')
data = sess['amplifier_data']
fs = sess['sample_rate']

# Apply filters
filtered = filter_emg(data, filter_type='bandpass', lowcut=10, highcut=500, fs=fs)

Real-time ZMQ Streaming

# Launch the live viewer (ensure ZMQ Interface plugin is active in GUI)
python -m pyoephys.applications._realtime_viewer --host 127.0.0.1 --channels 0:8

---

🗂️ Package Structure

pyoephys/
├── applications/     # GUI applications (Real-time & Offline viewers)
├── interface/        # ZMQ, LSL, and playback clients
├── io/               # Unified file loaders (.oebin, .npz)
├── ml/               # Gesture classification (CNN-LSTM)
├── plotting/         # Visualization utilities
└── processing/       # Signal filters, QC, and synchronization

examples/
├── benchmarks/       # Throughput and latency tests
├── interface/        # LSL, ZMQ, and hardware control
│   ├── hardware/     # Serial/UDP Pico integration
│   ├── imu/          # Sleeve IMU client & monitor
│   ├── lsl/          # LSL streaming & capture
│   └── zmq/          # ZMQ clients & plotters
└── machine_learning/ # Model training and evaluation

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Made with ❤️ by the Neuromechatronics Lab