Recording live data

Recording live data#

This code allows you to record live data from the FinalSpark LiveMEA webview and save it to an HDF5 file.
It provides both an API and a CLI for ease of use.

The code is available at the following link.

Note

The code is usable even if you are not a NeuroPlatform user, and is freely available for anyone to use.
Please reuse the provided code in your own projects and modify it as needed; the current implementation merely provides a simple way to record data from the live view.

We only request that you mention FinalSpark as the source of the data where you use it.

Data specification#

The live view gives, roughly every second, 4096 samples of data from 32 electrodes.

In the following code, each “duration” unit corresponds to 4096 values; the sampling frequency is 30kHz downsampled via wavelet transform by a factor of 8.

Therefore, a duration of 1 corresponds in reality to ~1.09 seconds of data.

Expect a disk usage of around 1.04 MB, and an average wait time of at least 1.09 seconds per second of recording.

Installation#

We recommend using a virtual environment or a conda environment to install the required packages. For anyone unfamiliar with or new to Python, we recommend installing miniconda to create a new environment.

  1. Clone the repository:

    git clone https://github.com/FinalSpark-np/LiveMEA.git
cd live-mea

  1. Create and activate a conda environment:

    conda create --name livemea-env python=3.11
conda activate livemea-env

  2. Install the required packages:

    pip install -r requirements.txt

API Usage#

Caution

Using the record() method in Jupyter will not work.
You will have to use await record_async() instead.
See the Example below for more details, you can also use the CLI instead.

LiveMEA Class#

Initialization#

from MEA_live import LiveMEA
from pathlib import Path

live_mea = LiveMEA(save_path: str | Path, recording_duration: int = 5, mea_id: int = 1)

  • save_path (str | Path): Path to save the HDF5 file.

  • recording_duration (int, optional): Duration of recording in “seconds” (chunks of 4096 samples).
    Defaults to 5.

  • mea_id (int, optional): MEA ID to use. Defaults to 1, must be between 1 and 4.

Methods#

  • record() : Starts recording live data and saves it to the specified HDF5 file.

  • record_async() : Starts recording live data asynchronously and saves it to the specified HDF5 file. Used within Jupyter notebooks.

  • plot_data(h5f_path: str | Path): Plots data from an HDF5 file. Requires Matplotlib as an additional dependency.

Example#

from MEA_live import LiveMEA

# Initialize the LiveMEA instance
live_mea = LiveMEA(save_path="live_data.h5", recording_duration=10, mea_id=1)

# Record data
# IF YOU ARE USING A PYTHON SCRIPT :
live_mea.record()
# IF YOU ARE USING JUPYTER :
data = await live_mea.record_async()

# Plot recorded data
LiveMEA.plot_data("live_data.h5")

CLI Usage#

You can also use a command-line interface (CLI) to record live MEA data.

Command#

From the repository root directory, run the following command:

python live-mea -d <duration> -p <path> -m <MEA>

Arguments#

  • -d, --duration (int, optional): Duration of the recording in seconds. Defaults to 5.

  • -p, --path (str, optional): Path to save the recorded data. Defaults to “live_data.h5”.

  • -m, --MEA (int, optional): MEA ID to record data from. Defaults to 1.

Example#

python live-mea -d 10 -p "live_data.h5" -m 1

This command will record live MEA data for 10 seconds and save it to live_data.h5 using MEA 1.

Note

The CLI works natively in Jupyter notebooks or ipython.
Simply use ! before the command within your notebook cell.

Accessing the saved data#

Each time point is saved under its own group in the HDF5 file. The data for each electrode is then saved as a dataset within the group.

The structure of the HDF5 file is as follows:

live_data.h5
├── timestamp_0
│   ├── electrode_0
│   ├── electrode_1
│   ├── ...
│   ├── electrode_31
├── timestamp_1
│   ├── ...
│   ├── electrode_31
├── ...

You can access the saved data using the h5py library in Python.

import h5py

# Open the HDF5 file
with h5py.File("live_data.h5", "r") as f:
    # List all timestamps
    timestamps = list(f.keys())
    print("Timestamps:", timestamps)

    # Access data for a specific timestamp
    timestamp = timestamps[0]
    data_group = f[timestamp]

    # List all electrodes
    electrodes = list(data_group.keys())
    print("Electrodes:", electrodes)

    # Access data for a specific electrode
    electrode_data = data_group["electrode_0"][:]
    print("Electrode 0 data:", electrode_data)

Note regarding future updates#

Warning

We provide the current code to obtain live data as-is, and we do not guarantee its maintenance or future updates.

The provided code may stop working without notice, as our infrastructure is still in development, and constantly evolving and changing.
In the event that the code is non-functional, please reach out to us, e.g by opening an issue on the repository.

Of course, feel free to open issues or pull requests anytime if you encounter any problems or have suggestions for improvements.