Signal quality in recorded file

[Benjamin]

Hello James,

I would like to know if it is possible know what was the connection quality like percentage for the whole recorded session for each sensors and an average ...I know that when look at the file online we can see Good Fit, Bad Fit and Jaw Clench but I think it would be nice to have those kind of data in order to estimate the quality of the recorded file.

Best regards

[James]

Interaxon already give 3 levels of signal quality in the recorded data. This is detailed in the CSV specs here: https://musemonitor.com/FAQ.php#csvspec
HSI_{TP9,AF7,AF8,TP10} -> 1=Good, 2=Medium, 4=Bad

I'm not sure how you would quantify a percentage. Perhaps by taking a rolling average min and comparing that to a rolling average max? That'd show you the "crispness" of the line which is approximately equal to signal quality. That's certainly something you could try out by loading the data into Excel and using my rendering Macro as a starting point.

[Benjamin]

I ask you that because I sent someone a Muse Monitor CSV files and he analyzed that with his software : NeuroVisual (it only run on PC) and it appears I had 1% percent of bad connection for 3 sensors and 2% on another one, he told me that only 0% percent results are relevant ; I am not a scientist but I would have guessed that on 15 minutes meditation 1 or 2% is relevant but he told me that I could make a huge change in brainwave analysis. This is why I ask you.

[James]

You'll never have perfect signal quality; there will always be some interference from something. The electrodes are designed to pick up brainwaves, but they also pick up the small electric signals used for muscle movement. Things like blinks make big obvious spikes in the data. When you get those blink spikes you dump the data and move on to the next viable section. Perhaps this is what he's talking about when he says to use 0% results. Otherwise I'm not sure, sorry. The only way you'd get perfectly clean data would be to make direct contact with the brain!

[Benjamin]

Thank you, that confirms what I have understand about the device.
He posted me results where according to his software he gets 0% of bad connection on 4,5h of meditation ; he made 2 meditations back to back and noticed that the graphs he gets are very different with small bad connections percentage than the ones at 0% he has on a very regular basis, he told me that it would come from Muse 2 which has more problems than Muse 1 on this matter ; I have to say I am quite lost now

[James]

For the best signal quality, keep facial muscle movement to a minimum. Also, the sensors require a thin film of sweat to form, to make electrical contact. That's about it.

[Benjamin]

That is what I do and I feel it is working well. I will see in the long run. Thank you very much.

[Benjamin]

That is what I do and I feel it is working well. I will see in the long run. Thank you very much.

[electronic monk]

hey james,whats up (we had a few conversations through my other email jerryashton81118@gmail.com about the dominant delta i produced back in august)

what i have been discussing with benjamin was the % of bad connections of each sensor,not the signal quality u r referring 2 as 2 the brainwaves and the possible interferences through muscle movements,open eyes creating fake delta or gamma etc

it can b seen in neurovisual in the bottom right corner where it tells u if the sensors had a good connection or not...

i was looking at benjamin's graphs and they didnt look right 2 me so i asked him 2 send me the CSVs,put them in neurovisual and saw that they had bad connections in the 2 front sensors or even all 4,so when i showed him the results he said that he believed such a small % wouldnt make a difference,or not a big enough difference,so i showed him some of my graphs where i tested back 2 back after seeing that beta was 2 high and didnt go along with my subjective feeling of being very calm which usually shows in high or even dominant delta,and those graphs had also just a 2-3% bad connection but it was enough 2 change the order of the brainwaves...

here r the graphs... muse 2 bad connection with high beta

which has 2-3% bad connections in all 4 sensors

and here is a muse1 meditation right after that where delta is strong

which has 0% bad connections

and here r the 4.5 hrs meditations he was talking about which i sent him 2 show him its not only possible but it should b 0% for the data 2 b accurate...

muse2


muse1


and i did it many times and it always had the same results...

but he still didnt think it was logical,maybe a coincidence,so he wanted 2 contact u 2 get another opinion on it...

[James]

Oh right. Well this is exactly what I was talking about in my first post mentioning the HSI values.
I guess NeuroVisual is doing something with those values to give you percentage of how many were bad vs good.

Getting 0% should be fairly easy, just don't start recording until the headband is well connected. You can check the CSV file to see where you're loosing optimal connectivity.

With the online graphing, I don't show a warning unless the HSI value is 4 (bad). If it's a 2 (medium), then I treat that as ok. Also if you're viewing the average minutes vs all values, you might not see a warning as if the majority were ok, it'll treat the set as ok.