Python code for data collection from Neurosky Mindwave Mobile headset device

mindwave.py

import numpy as np
import pandas as pd
import sys
import json
import time
from telnetlib import Telnet

# Initializing the arrays required to store the data.
attention_values = np.array([])
meditation_values = np.array([])
delta_values = np.array([])
theta_values = np.array([])
lowAlpha_values = np.array([])
highAlpha_values = np.array([])
lowBeta_values = np.array([])
highBeta_values = np.array([])
lowGamma_values = np.array([])
highGamma_values = np.array([])
blinkStrength_values = np.array([])
time_array = np.array([])

tn=Telnet('localhost',13854);

start=time.clock();

i=0;
tn.write('{"enableRawOutput": true, "format": "Json"}');


outfile="null";
if len(sys.argv)>1:
	outfile=sys.argv[len(sys.argv)-1];
	outfptr=open(outfile,'w');

eSenseDict={'attention':0, 'meditation':0};
waveDict={'lowGamma':0, 'highGamma':0, 'highAlpha':0, 'delta':0, 'highBeta':0, 'lowAlpha':0, 'lowBeta':0, 'theta':0};
signalLevel=0;

while time.clock() - start < 30:
	blinkStrength=0;
	line=tn.read_until('\r');
	if len(line) > 20:	
		timediff=time.clock()-start;
		dict=json.loads(str(line));
		if "poorSignalLevel" in dict:
			signalLevel=dict['poorSignalLevel'];
		if "blinkStrength" in dict:
			blinkStrength=dict['blinkStrength'];
		if "eegPower" in dict:
			waveDict=dict['eegPower'];
			eSenseDict=dict['eSense'];
		outputstr=str(timediff)+ ", "+ str(signalLevel)+", "+str(blinkStrength)+", " + str(eSenseDict['attention']) + ", " + str(eSenseDict['meditation']) + ", "+str(waveDict['lowGamma'])+", " + str(waveDict['highGamma'])+", "+ str(waveDict['highAlpha'])+", "+str(waveDict['delta'])+", "+ str(waveDict['highBeta'])+", "+str(waveDict['lowAlpha'])+", "+str(waveDict['lowBeta'])+ ", "+str(waveDict['theta']);
		time_array = np.append(time_array, [timediff]);
		blinkStrength_values = np.append(blinkStrength_values, [blinkStrength]);
		lowGamma_values = np.append(lowGamma_values, [waveDict['lowGamma']]);
		highGamma_values = np.append(highGamma_values, [waveDict['highGamma']]);
		highAlpha_values = np.append(highAlpha_values, [waveDict['highAlpha']]);
		delta_values = np.append(delta_values, [waveDict['delta']]);
		lowBeta_values = np.append(lowBeta_values, [waveDict['lowBeta']]);
		highBeta_values = np.append(highBeta_values, [waveDict['highBeta']]);
		theta_values = np.append(theta_values, [waveDict['theta']]);
		lowAlpha_values = np.append(lowAlpha_values, [waveDict['lowAlpha']]);
		attention_values = np.append(attention_values, [eSenseDict['attention']]);
		meditation_values = np.append(meditation_values, [eSenseDict['meditation']]);
		print (outputstr);
		if outfile!="null":	
			outfptr.write(outputstr+"\n");		

person_name = input('Enter the name of the person: ')
blink_label = input('Enter left or right eye blink(1 for left, 2 for right): ')
time_starting = input('When does TGC start: ')
lefty_righty = input('Is the person left-handed or right-handed: ')
time_blinking = input('The time of the blink: ')

# Data Recorded for a single person
data_row = pd.DataFrame({'Name': person_name, 'attention': [attention_values], 'meditation': [meditation_values], 'delta': [delta_values], 'theta': [theta_values], 'lowAlpha': [lowAlpha_values], 'highAlpha': [highAlpha_values], 'lowBeta': [lowBeta_values], 'highBeta': [highBeta_values],
                          'lowGamma':[lowGamma_values] , 'highGamma': [highGamma_values], 'blinkStrength': [blinkStrength_values], 'time': [time_array], 'LOR': blink_label})

# Reading the data stored till now
dataset = pd.read_csv('data_eeg.csv')

from numpy import nan as Nan

dataset = dataset.append(pd.Series([blink_label, person_name, [attention_values], [blinkStrength_values], [delta_values]
	, [highAlpha_values], [highBeta_values], [highGamma_values], [lowAlpha_values], [lowBeta_values], [lowGamma_values], [meditation_values], 
	[theta_values], [time_array], time_starting, lefty_righty, time_blinking], index=['LOR', 'Name', 'attention', 'blinkStrength', 'delta', 'highAlpha', 'highBeta', 'highGamma', 'lowAlpha', 'lowBeta', 'lowGamma', 'meditation', 'theta', 'time', 'time_start', 'LTYRTY', 'time_blink']),  ignore_index = True)

#Appending and storing the data in the same csv
#dataset.append(data_row)
dataset.to_csv('data_eeg.csv')      
    
tn.close();
#outfptr.close();

Hey, this is an amazing piece of code! I'm trying to get raw-wave data (1D data that has not been segregated into various frequency spectrums) from my headset. I tried putting 'enableRawOutput' as false, but it doesn't seem to effect anything. I tried looking into data that the headset sends over and found a few new categories like 'familiarity' and 'mental_strength' but couldn't find raw-wave output/EEG_signal. I know that it is possible via MATLAB(I tried, it worked but ran into a lot of technical issues with thinkgear.dll) but I'm specifically looking for python implementation. If you happen to know something, can you guide me in the right direction?

https://github.com/MexsonFernandes/EEG-controlled-Robot-using-Neurosky-Mindwave-Mobile2 kindly check this repo for raw EEG. I have coded it in C# and had got better results with good response time. So try to use it and post your experience.

Hey, I am trying to build an EEG controlled bionic arm. I am trying to run this code but I am stuck at tn=Telnet('localhost',13854);. The error message reads - "Name or service not known". Could you please help me out?

@svr1998 that would mean that the ThinkGear Connector is probably not running and/or unable to connect to the headset.

For anyone who makes it to this script, I've built something more formal here (use it with --output path/filename --no-lsl to get all the data into a CSV file without running LSL): https://github.com/gmierz/mindwave-lsl

How to increase the samples per second ? As of this code i m hardly getting 1 sample per second . I need around 250+ samples per second .

@kesavsivakumar those are the frequency measurements which return data at 1 Hz. You want to look at the raw data to get the higher sampling rate.

@kesavsivakumar those are the frequency measurements which return data at 1 Hz. You want to look at the raw data to get the higher sampling rate.

@gmierz How to extract the raw data , lately i have been trying to extract raw eeg data using java script (node-think-gear-socket) but I feel like ,I always lead to a dead end .

@kesavsivakumar check out the code I wrote here: https://github.com/gmierz/mindwave-lsl/blob/master/mindwavelsl/outlet.py#L19

There's some examples of how you can use this in test_*.py at the root of the repo.

When you create the MindwaveLSL object, initialize it with run_lsl=False so you can pull the data without needing the LSL bits.

You'll want to use the make_sample function, and then check the data at index 8 of the sample, that will be the raw eeg value. You can see the ordering of the fields here: https://github.com/gmierz/mindwave-lsl/blob/master/mindwavelsl/constants.py#L17

@gmierz Thanks mate!

Hey, I have questions to ask as i am noob in python (have some knowledge) as i am making fully functional car with raspberry pi which will have rc ,egg, as well internet, (auto pilot) anticrash so lets move to questions no 1) will i have to change the script for eeg control( mean to put a value )2) will there is any errors come when i will run 3 python script simultaneously 3) how to make python script for obstrical avoidance (as i did nt find the way no where )

someone please help

It should be
tn.write(str.encode('{"enableRawOutput": true, "format": "Json"}'))

Does anyone solve this error?

@urmi2927 change '\r' to b'\r'

can you explain this line please
tn=Telnet('localhost',13854);

That's the telnet connection used to communicate data between the device and your computer. I suggest using this package so you don't get bogged down by these details: https://github.com/gmierz/mindwave-lsl

I tried this link thank you for shearing it, but i still not understand how can it work and how I can connect this app with my headset? is there any information should I add it? i hope someone help me in this Issue

You need to get the Thinkgear connector working first, then that'll provide you with a host and port to use in the package above (see the help section of the command).

WHEN i trien the modul I get an error in related with pyxdf

is it possible to get raw eeg data

Yes, but you need to make sure you do a write like this to enable it: https://github.com/gmierz/mindwave-lsl/blob/21827fe80f034e0a1d6b5da6b9165c6d0574ac23/test_mouse_movements.py#L19

is it possible to enable raw eeg data from mindwave neurosky 1, the headset with RF dongle connecting through usb port? and if how can i do that?

Yes, but you need to make sure you do a write like this to enable it: https://github.com/gmierz/mindwave-lsl/blob/21827fe80f034e0a1d6b5da6b9165c6d0574ac23/test_mouse_movements.py#L19

but I tried it is not possible to get raw EEG data

I'm using another single electrode FT$S mind link EEG Headband, is it any method to receive live raw EEG data from it