Feature/experiment

README.md

@@ -1,174 +1,118 @@
 
-[![DOI](https://zenodo.org/badge/80209610.svg)](https://zenodo.org/badge/latestdoi/80209610)
-
-# Muse LSL
-
-A Python package for streaming, visualizing, and recording EEG data from the Muse 2016 headband.
-
-![Blinks](blinks.png)
-
-## Requirements
-
-The code relies on [pygatt](https://github.com/peplin/pygatt) or [BlueMuse](https://github.com/kowalej/BlueMuse/tree/master/Dist) for BLE communication and works differently on different operating systems.
-
-- Windows: On Windows 10, we recommend installing [BlueMuse](https://github.com/kowalej/BlueMuse/tree/master/Dist) and using its GUI to discover and connect to Muse devices. Alternatively, if you have a BLED112 dongle you can try Muse LSL's bgapi backend (`muselsl stream --backend bgapi`).
-- Mac: On Mac, a **BLED112 dongle is required**. The bgapi backend is required and will be used by default when running Muse LSL from the command line
-- Linux: No dongle required. However, you may need to run a command to enable root-level access to bluetooth hardware (see [Common Issues](#linux)). The pygatt backend is required and will be used by default from the command line. and make sure to read the 
-
-**Compatible with Python 2.7 and Python 3.x**
-
-**Only compatible with Muse 2 and Muse 2016**
-
-_Note: if you run into any issues, first check out out [Common Issues](#common-issues) and then the [Issues](https://github.com/alexandrebarachant/muse-lsl/issues) section of this repository_
-
-## Getting Started
-
-### Installation
-
-Install Muse LSL with pip
-
-    pip install muselsl
-
-### Setting Up a Stream
-
-On Windows 10, we recommend using the [BlueMuse](https://github.com/kowalej/BlueMuse/tree/master/Dist) GUI to set up an LSL stream. On Mac and Linux, the easiest way to get Muse data is to use Muse LSL directly from the command line. Use the `-h` flag to get a comprehensive list of all commands and options.
-
-To print a list of available muses:
 
-    $ muselsl list
 
-To begin an LSL stream from the first available Muse:
+# Muse LSL - Experiment
 
-    $ muselsl stream  
+This repo is forked from [https://github.com/alexandrebarachant/muse-lsl](https://github.com/alexandrebarachant/muse-lsl) [1].
 
-To connect to a specific Muse you can pass the name of the device as an argument. Device names can be found on the inside of the left earpiece (e.g. Muse-41D2):
+The original code was modified and extended to allow streaming and recording of two or more muses simultaneously.
+We used two `2019 Muse 2 headsets` for developement and testing.
 
-    $ muselsl stream --name YOUR_DEVICE_NAME
+## Installation
 
-You can also directly pass the MAC address of your Muse. This provides the benefit of bypassing the device discovery step and can make connecting to devices quicker and more reliable:
+Install this repository as a pip package with
 
-    $ muselsl stream --address YOUR_DEVICE_ADDRESS
+    pip3 install .
 
-### Working with Streaming Data
+or just install the dependencies using:
 
-Once an LSL stream is created, you have access to the following commands.
+    pip3 install -r requirements.txt
 
-*Note: the process running the `stream` command must be kept alive in order to maintain the LSL stream. These following commands should be run in another terminal or second process*
+If you *did not* install the repo as a pip package, you need to replace `muselsl` with
+`python3 -m muselsl` in all following examples.
 
-To view data:
+## Hardware Setup
 
-    $ muselsl view
+We tested different hardware setups for multiple streams.
+The most reliable setup, which allowed streaming of all data-streams from two muses was the following:
 
-If the visualization freezes or is laggy, you can also try the alternate version 2 of the viewer. *Note: this will require the additional [vispy](https://github.com/vispy/vispy) and [mne](https://github.com/mne-tools/mne-python) dependencies*
+- Two Muses
+- Two Laptops (We used a MacBook with `BLED112 dongle` and a linux machine with integrated Bluetooth.)
+- Wireless (or wired) network with peer to peer traffic allowed (HPI network might not work, we made a WIFI hotspot using an android phone)
 
-    $ muselsl view --version 2
+One of the laptops acts as a *stream source*, while the other *records* its own stream and the stream from the stream source, which is sent over
+the network.
 
-To record EEG data into a CSV:
+## Stream Source
 
-    $ muselsl record --duration 60  
+The stream source is started before the recorder and must run until after the recorder has stopped.
+If the stream source crashes it should be restarted, the recorder can recover only if the stream source is restarted, otherwise the recorder will
+crash, when attempting to stop the recording.
 
-*Note: this command will also save data from any LSL stream containing 'Markers' data, such as from the stimulus presentation scripts in [EEG Notebooks](https://github.com/neurotechx/eeg-notebooks)*
+The stream source is started with the following command:
 
-Alternatively, you can record data directly without using LSL through the following command:
+     muselsl stream -pcg -a <MAC-ADRESS-OF-MUSE>
 
-    $ muselsl record_direct --duration 60
+Where `<MAC-ADRESS-OF-MUSE>` must be replaced with the mac-address of one of the muses.
+To find the mac addresses of your muses, you can turn on your muses and run
 
-_Note: direct recording does not allow 'Markers' data to be recorded_
+    muselsl list
 
-## Running Experiments
+to get a list of the available headsets.
+The last four letters (two bytes) of the MAC address are printed on each Muse headset.
 
-Muse LSL was designed so that the Muse could be used to run a number of classic EEG experiments, including the [P300 event-related potential](http://alexandre.barachant.org/blog/2017/02/05/P300-with-muse.html) and the SSVEP and SSAEP evoked potentials.
+In our setup one stream source is started on the *stream source* machine and one is started
+for the other muse, on the *recording* machine.
 
-The code to perform these experiments is still available, but is now maintained in the [EEG Notebooks](https://github.com/neurotechx/eeg-notebooks) repository by the [NeuroTechX](https://neurotechx.com) community.
 
-## Usage as a Library
+## View
 
-If you want to integrate Muse LSL into your own Python project, you can import and use its functions as you would any Python library. Examples are available in the `examples` folder:
+Before starting the recording, make sure all streams are coming through, and are of decent quality.
+This can be done using the viewer.
 
-```Python
-from muselsl import stream, list_muses
+    muselsl view -v 2 -a <MAC-ADRESS-OF-MUSE> -t <DATA-TYPE>
 
-muses = list_muses()
-stream(muses[0]['address'])
+Where data-type can be any of:
+- PPG
+- EEG
+- ACC
+- GYRO
 
-# Note: Streaming is synchronous, so code here will not execute until after the stream has been closed
-print('Stream has ended')
-```
+You can zoom in and out by holding your left mouse button and moving your mouse up or down.
 
-## Alternate Data Types
+> It is also a good idea, to at least monitor the EEG streams of all participant during the
+> experiment, to be able to detect artifacts early and fix them (eg. loose electrodes, muscle noise, etc.).
 
-In addition to EEG, the Muse also provides data from an accelerometer, gyroscope, and, in the case of the Muse 2, a photoplethysmography (PPG) sensor. These data types can be enabled via command line arguments or by passing the correct parameters to the `stream` function. Once enabled, PPG, accelerometer, and gyroscope data will streamed in their own separate LSL streams named "PPG", "ACC", and "GYRO", respectively.
+## Recorder
 
-To stream data from all sensors in a Muse 2 from the command line:
+When the stream source is running, the recorder script can be started with the following command:
 
-    muselsl stream --ppg --acc --gyro
+    muselsl record -d <data-path> -n <number-of-participants> [-i <trial-id>]
 
-As a library function:
+The script will ask for the `MUSE-ID` of each participant, which is written on each muse device.
+The `MUSE-ID` are the last two bytes (four letters) of the mac address.
 
-```Python
-from muselsl import stream, list_muses
+After all IDs are set, the recording starts automatically.
+During the recording, timestamps can be saved, by typing any text as a label and pressing enter.
+These timestamps can be used to mark when different parts of the experiment begin, or when the experiment has ended.
 
-muses = list_muses()
-stream(muses[0]['address'], ppg_enabled=True, acc_enabled=True, gyro_enabled=True)
-```
+The recording will continue, until it is stopped by pressing `CTRL` `+` `C`.
 
-To record data from an alternate data source:
+All recordings are saved in the specified directory.
+If no `trial-id` is provided, the current timestamp is used instead to create a new subdirectory,
+in which all data is stored.
+This directory contains the recorded markers, which are stored in `markers.csv` and a subdirectory for each participant.
 
-    muselsl record --type ACC
+Each participant directory contains the following files:
 
-*Note: The record process will only record from one data type at a time. However, multiple terminals or processes can be used to record from multiple data types simultaneously*
 
-## What is LSL?
+| File | Description |
+|------|-------------|
+|ACC.csv | Data from the Accelerometer |
+|EEG.csv| EEG Data|
+|GYRO.csv | Gyroscope Data |
+|PPG.csv | Heart rate measured with PPG |
 
-Lab Streaming Layer or LSL is a system designed to unify the collection of time series data for research experiments. It has become standard in the field of EEG-based brain-computer interfaces for its ability to make seperate streams of data available on a network with time synchronization and near real-time access. For more information, check out this [lecture from Modern Brain-Computer Interface Design](https://www.youtube.com/watch?v=Y1at7yrcFW0) or the [LSL repository](https://github.com/sccn/labstreaminglayer)
+## Other Setups
 
-## Common Issues
+We tried using Raspberry Pi's as stream source, however
+even the newest model 4 with a `BLED112 dongle`
+it was not able to stream reliably for more than 10 minutes.
+It might be possible, though to use Raspberry Pi's, if not all data
+streams are enabled (eg. if only recording ppg).
+See [this github issue](https://github.com/alexandrebarachant/muse-lsl/issues/55) for more information.
 
-### Mac and Windows
-
-1.  Connection issues with BLED112 dongle:
-
-- You may need to use the `--interface` argument to provide the appropriate COM port value for the BLED112 device. The default value is COM9. To setup or view the device's COM port go to your OS's system settings
-
-### Linux
-
-1.  `pygatt.exceptions.BLEError: Unexpected error when scanning: Set scan parameters failed: Operation not permitted` (Linux)
-
-- This is an issue with pygatt requiring root privileges to run a scan. Make sure you [have `libcap` installed](https://askubuntu.com/questions/347788/how-can-i-install-libpcap-header-files-on-ubuntu-12-04) and run `` sudo setcap 'cap_net_raw,cap_net_admin+eip' `which hcitool` ``
-
-2.  `pygatt.exceptions.BLEError: No characteristic found matching 273e0003-4c4d-454d-96be-f03bac821358` (Linux)
-
-- There is a problem with the most recent version of pygatt. Work around this by downgrading to 3.1.1: `pip install pygatt==3.1.1`
-
-3.  `pygatt.exceptions.BLEError: No BLE adapter found` (Linux)
-
-- Make sure your computer's Bluetooth is turned on.
-
-4.  `pygatt.exceptions.BLEError: Unexpected error when scanning: Set scan parameters failed: Connection timed out` (Linux)
-
-- This seems to be due to a OS-level Bluetooth crash. Try turning your computer's bluetooth off and on again
-
-5.  `'RuntimeError: could not create stream outlet'` (Linux)
-
-- This appears to be due to Linux-specific issues with the newest version of pylsl. Ensure that you have pylsl 1.10.5 installed in the environment in which you are trying to run Muse LSL
-
-## Citing muse-lsl
-
-```
-@misc{muse-lsl,
-  author       = {Alexandre Barachant and
-                  Dano Morrison and
-                  Hubert Banville and
-                  Jason Kowaleski and
-                  Uri Shaked and
-                  Sylvain Chevallier and
-                  Juan Jesús Torre Tresols},
-  title        = {muse-lsl},
-  month        = may,
-  year         = 2019,
-  doi          = {10.5281/zenodo.3228861},
-  url          = {https://doi.org/10.5281/zenodo.3228861}
-}
-```
 
+[1]
 > Alexandre Barachant, Dano Morrison, Hubert Banville, Jason Kowaleski, Uri Shaked, Sylvain Chevallier, & Juan Jesús Torre Tresols. (2019, May 25). muse-lsl (Version v2.0.2). Zenodo. http://doi.org/10.5281/zenodo.3228861
+[![DOI](https://zenodo.org/badge/80209610.svg)](https://zenodo.org/badge/latestdoi/80209610)

muselsl/__init__.py

@@ -1,4 +1,5 @@
 from .stream import stream, list_muses
-from .record import record, record_direct
+from .record import record
 from .view import view
+from .replay import replay
 __version__ = "1.0.0"

muselsl/__main__.py

@@ -29,18 +29,14 @@ def main():
                 -f --figure     Window size.
                 -v --version    Viewer version (1 or 2) - 1 is the default stable version, 2 is in development (and takes no arguments).
                 -b --backend    Matplotlib backend to use. Default: TkAgg
-
-    record   Record EEG data from an LSL stream.
-                -d --duration   Duration of the recording in seconds.
+    record   Recording an experiment.
+                -d --directory      Root-directory to store recorded data in.
+                -n --participants   The number of participants in this run.
+                -i --trial-id       The id of this trial. Data is stored in a subdirectory with this name.
+                                    If no id is provided, the current timestamp is used instead.
+
+    replay   Replay data from a recorded CSV file into a new LSL stream.
                 -f --filename   Name of the recording file.
-                -dj --dejitter  Whether to apply dejitter correction to timestamps.
-                -t --type       Data type to record from. Either EEG, PPG, ACC, or GYRO 
-
-    record_direct      Record data directly from Muse headset (no LSL).
-                -a --address    Device MAC address.
-                -n --name       Device name (e.g. Muse-41D2).
-                -b --backend    BLE backend to use. can be auto, bluemuse, gatt or bgapi.
-                -i --interface  The interface to use, 'hci0' for gatt or a com port for bgapi.
         ''')
 
     parser.add_argument('command', help='Command to run.')

muselsl/cli.py

@@ -51,53 +51,31 @@ def stream(self):
                args.interface, args.name, args.ppg, args.acc, args.gyro, args.disable_eeg)
 
     def record(self):
-        parser = argparse.ArgumentParser(
-            description='Record data from an LSL stream.')
-        parser.add_argument("-d", "--duration",
-                            dest="duration", type=int, default=60,
-                            help="Duration of the recording in seconds.")
-        parser.add_argument("-f", "--filename",
-                            dest="filename", type=str, default=None,
-                            help="Name of the recording file.")
-        parser.add_argument("-dj", "--dejitter",
-                            dest="dejitter", type=bool, default=True,
-                            help="Whether to apply dejitter correction to timestamps.")
-        parser.add_argument("-t", "--type", type=str, default="EEG",
-                            help="Data type to record from. Either EEG, PPG, ACC, or GYRO.")
+        from .run_experiment import ExperimentalRun
 
-        args = parser.parse_args(sys.argv[2:])
-        from . import record
-        record(args.duration, args.filename, args.dejitter, args.type)
-
-    def record_direct(self):
         parser = argparse.ArgumentParser(
-            description='Record directly from Muse without LSL.')
-        parser.add_argument("-a", "--address",
-                            dest="address", type=str, default=None,
-                            help="Device MAC address.")
-        parser.add_argument("-n", "--name",
-                            dest="name", type=str, default=None,
-                            help="Name of the device.")
-        parser.add_argument("-b", "--backend",
-                            dest="backend", type=str, default="auto",
-                            help="BLE backend to use. Can be auto, bluemuse, gatt or bgapi.")
-        parser.add_argument("-i", "--interface",
-                            dest="interface", type=str, default=None,
-                            help="The interface to use, 'hci0' for gatt or a com port for bgapi.")
-        parser.add_argument("-d", "--duration",
-                            dest="duration", type=int, default=60,
-                            help="Duration of the recording in seconds.")
-        parser.add_argument("-f", "--filename",
-                            dest="filename", type=str, default=None,
-                            help="Name of the recording file.")
+            description='Start recording of an experiment.')
+        parser.add_argument("-d", "--directory",
+                            dest="data_root", type=str, required=True,
+                            help="Root-directory to store recorded data in.")
+        parser.add_argument("-n", "--participants",
+                            dest="num_participants", type=int, required=True,
+                            help="The number of participants in this run.")
+        parser.add_argument("-i", "--trial-id",
+                            dest="trial_id", type=str, default=None,
+                            help="The id of this trial. Data is stored in a subdirectory with this name."
+                                 + "If no id is provided, the current timestamp is used instead.")
         args = parser.parse_args(sys.argv[2:])
-        from . import record_direct
-        record_direct(args.address, args.backend,
-                      args.interface, args.name, args.duration, args.filename)
+
+        experiment = ExperimentalRun(data_root=args.data_root,
+                                     num_participants=args.num_participants, trial_id=args.trial_id)
+        experiment.start()
 
     def view(self):
         parser = argparse.ArgumentParser(
-            description='View EEG data from an LSL stream.')
+            description='View data from an LSL stream.')
+        parser.add_argument("-t", "--type", type=str, default="EEG", dest="data_type",
+                            help="Data type to view. Either EEG, PPG, ACC, or GYRO.")
         parser.add_argument("-w", "--window",
                             dest="window", type=float, default=5.,
                             help="Window length to display in seconds.")
@@ -116,7 +94,22 @@ def view(self):
         parser.add_argument("-b", "--backend",
                             dest="backend", type=str, default='TkAgg',
                             help="Matplotlib backend to use. Default: %(default)s")
+        parser.add_argument("-a", "--address",
+                            dest="source_id", type=str, default=None,
+                            help="Device MAC address.")
         args = parser.parse_args(sys.argv[2:])
         from . import view
         view(args.window, args.scale, args.refresh,
-             args.figure, args.version, args.backend)
+             args.figure, args.version, args.backend, args.data_type, args.source_id)
+
+    def replay(self):
+        parser = argparse.ArgumentParser(
+            description='Replay data from a recorded CSV file into a new LSL stream.')
+        parser.add_argument("-f", "--filename",
+                            dest="filename", type=str, default=None,
+                            help="Name of the recording file.")
+        # parser.add_argument("-t", "--type", type=str, default="EEG",
+        #                     help="Data type to record from. Either EEG, PPG, ACC, or GYRO.")
+        args = parser.parse_args(sys.argv[2:])
+        from . import replay
+        replay(args.filename)