### Explanation about CAR spatial filter

Posted:

**16 Feb 2017, 08:59**Hello,

i do have some doubts about this filter that i hope someone could help me with. First of all, i am using the 3.6 beta version of bci2000.

Theoric part:

Well, as i do understand, readings are taken this way: It calculates the difference of voltage between reference points (usually neutral/not signal receiving electrodes) and each electrode. I will call this "normal reading".

What CAR filter does is first calculate the mean value of all the electrodes participating.

Now, i am not very sure what CAR filter application involves. It first take "normal readings" and then apply a filter over them ( it gives the difference between normal readings and this mean value) or it does not take "normal readings" and somehow takes some values from electrodes in the beginning, calculates their mean value and takes that as reference?

I am also not sure about this filter application in BCI2000. As i understand, it takes in account the value of all electrodes participating to calculate the mean value, but afterwards we can apply this filter just in some channels. Correct me if i am wrong.

Practic part:

I am obtaining readings and making graphs of them using matlab filter (signal processing module) and when i do not use any filter, it all seems fine. Now i want to see how readings are with CAR filter, so i go to the "filtering" option of Operator module and select CAR as spatial filter type.

Then, in spatialfilterCARoutput i put all the channels (as for some reason, if i leave this parameter in blank, all readings are null).

After doing this, lectures show exactly the same as when i did not used CAR filter.

For example, lets assume i have 3 channels, C1, C2 and C3.

Without any filters, i take 1 sample of each , with this values: 4, 5, 3.

Now i want to apply a CAR spatial filter and choose C1 and C2 channels as output. What i would expect to obtain in the same case is 0, 1, 3.

I am not sure if i should touch nothing in the "spatial filter" option here also.

I would appreciate some help with this. Thanks.

i do have some doubts about this filter that i hope someone could help me with. First of all, i am using the 3.6 beta version of bci2000.

Theoric part:

Well, as i do understand, readings are taken this way: It calculates the difference of voltage between reference points (usually neutral/not signal receiving electrodes) and each electrode. I will call this "normal reading".

What CAR filter does is first calculate the mean value of all the electrodes participating.

Now, i am not very sure what CAR filter application involves. It first take "normal readings" and then apply a filter over them ( it gives the difference between normal readings and this mean value) or it does not take "normal readings" and somehow takes some values from electrodes in the beginning, calculates their mean value and takes that as reference?

I am also not sure about this filter application in BCI2000. As i understand, it takes in account the value of all electrodes participating to calculate the mean value, but afterwards we can apply this filter just in some channels. Correct me if i am wrong.

Practic part:

I am obtaining readings and making graphs of them using matlab filter (signal processing module) and when i do not use any filter, it all seems fine. Now i want to see how readings are with CAR filter, so i go to the "filtering" option of Operator module and select CAR as spatial filter type.

Then, in spatialfilterCARoutput i put all the channels (as for some reason, if i leave this parameter in blank, all readings are null).

After doing this, lectures show exactly the same as when i did not used CAR filter.

For example, lets assume i have 3 channels, C1, C2 and C3.

Without any filters, i take 1 sample of each , with this values: 4, 5, 3.

Now i want to apply a CAR spatial filter and choose C1 and C2 channels as output. What i would expect to obtain in the same case is 0, 1, 3.

I am not sure if i should touch nothing in the "spatial filter" option here also.

I would appreciate some help with this. Thanks.