search

JosephOchieng / braintrust

Automatically exported from code.google.com/p/braintrust
0 stars 0 forks source link

Filters are reversed in the EEG viewer #100

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago 
What steps will reproduce the problem?
1. view and EEG 
2. Click filters
3. apply a low pass or high pass filter you receive opposite results

Original issue reported on code.google.com by libertys...@gmail.com on 13 Dec 2011 at 1:40

GoogleCodeExporter commented 8 years ago

Original comment by libertys...@gmail.com on 13 Dec 2011 at 1:41

Attachments:

GoogleCodeExporter commented 8 years ago

Original comment by snd...@gmail.com on 13 Dec 2011 at 3:03

GoogleCodeExporter commented 8 years ago 
I think this is a terminological confusion, not an error, and we should resolve.

"Pass lo" = low end of the band we are passing.  I am replacing with "Minimum 
frequency to pass".
"Pass high" = high end of the band we are passing. I am replacing with "Maximum 
frequency to pass".

A low-pass filter starts at 0 and goes up to a max.
A high-pass filter starts at a min and goes up to the maximum.
A band-pass filter starts at a min and goes up to a max.

I can also disable some of these based on what the user selects as the filter 
type...?

Original comment by zack.i...@gmail.com on 13 Dec 2011 at 3:25

GoogleCodeExporter commented 8 years ago 
As I understand it.  In a clinical EEG viewer the filters are band pass 
filters, which usually have a predetermined cut off frequencies of  
approximately 70%  which will result in a 30 % reduction of amplitude at the 
specified frequency.

High pass filters should allow frequencies slower than 0  to pass e.g. OFF, 
0.01, 0.05, 1.0, 5.0, 10, etc and user can customize 

Low pass filter generally allows the slow frequencies to pass.  OFF, 15, 35, 
70, 100, etc and custom cut off.

The way it appears in this snap shot is that Pass low is actually the setting 
of your Low Frequency/High Pass filter setting

and Pass High  is set at 500 which is what High frequency filter was set at 
during acquisition for this record.  I also do not understand what the "(max 
249.95 indicates if it defalts to 500  this example is from study 005

Original comment by libertys...@gmail.com on 13 Dec 2011 at 4:16

GoogleCodeExporter commented 8 years ago 
I don't understand what "slower than 0" means...?

High pass, at least according to what I remember from EE, means that 
high-frequency signals pass and low ones are removed.  Are you referring to 
some sort of bandstop (notch) filter?

Bandpass lets you pass a range with a low and a high frequency.  Shouldn't a 
bandpass with a 0Hz low cutoff and an X-Hz high cutoff should be the same as a 
low-pass filter with an X-Hz high cutoff?

The max 249.95 is the Nyquist frequency:  anything above one half the original 
sampling rate (in this case 500Hz) is unreliable due to aliasing.  Hence asking 
for more than that is somewhat misleading.

Original comment by zack.i...@gmail.com on 13 Dec 2011 at 4:58

GoogleCodeExporter commented 8 years ago 
I am forwarding this on to Joost and Brian to speak to this more.  I think that 
maybe the form is a bit confusing.

see other responses below...

Less then zero.  

Ok that makes sense.  then that frequency should not default to the setting as 
in study five auto fills to 500.

Original comment by libertys...@gmail.com on 13 Dec 2011 at 5:26

GoogleCodeExporter commented 8 years ago 
Hi Zack, Libby,

Yes, a high pass filter should remove low frequency components. (i.e. to see 
high frequency oscillations)
low pass filters should make the plot more smooth and remove high frequency 
noise.

I think the problem is more that I don't think the filters are applied 
correctly. If I set the filters, the changes in the window do not correspond 
with what I expect. Furthermore, clicking forward and back in time seems to 
undo any filtering and the time series looks the same as when I have no filter 
selected.

Original comment by JBWagenaar on 13 Dec 2011 at 5:39

GoogleCodeExporter commented 8 years ago 
One acceptable for the 12.1 release to temporarily get rid of the filters.

Original comment by snd...@gmail.com on 15 Dec 2011 at 7:04

GoogleCodeExporter commented 8 years ago 
Filters have been updated and appear to give plausible behavior (low-pass cuts 
out high frequencies, high-pass results in high frequencies only).

Original comment by zack.i...@gmail.com on 13 Feb 2012 at 6:53

GoogleCodeExporter commented 8 years ago 
Has it been deployed to QA?  The filters do not seem to do anything there.  I 
can attempt to change things but when I click ok nothing happens.

Original comment by libertys...@gmail.com on 13 Feb 2012 at 7:21

GoogleCodeExporter commented 8 years ago 
deploying it now...you can watch out for the "Fixed-In-QA" status to know when 
stuff is out.

Original comment by snd...@gmail.com on 13 Feb 2012 at 7:25

GoogleCodeExporter commented 8 years ago 
No deploying it now...

Original comment by snd...@gmail.com on 13 Feb 2012 at 7:38

GoogleCodeExporter commented 8 years ago

Original comment by jwfro...@gmail.com on 13 Feb 2012 at 7:57

GoogleCodeExporter commented 8 years ago 
The filters are not filtering out the selected frequency

Original comment by libertys...@gmail.com on 15 Feb 2012 at 4:55

Attachments:

GoogleCodeExporter commented 8 years ago 
terminology is as usually as follows:
maximum of 4 frequency cutoffs - 2 for "pass"mode filters, 2 for "stop" mode 
filters

pass mode filters:
low fc (low frequency cutoff) is the only number specified for a "high pass" 
filter
high fc (high frequency cutoff) is the only number specified for a "low pass" 
filter
both low fc and high fc are specified for a bandpass filter

stop mode filters:
generally both low fc and high fc are specified, the frequencies between them 
are attenuated by the filter
in theory one could have "high stop" and "low stop" filters, but these are the 
equivalent if "low pass" and "high pass" respectively and so this terminology 
is not used.

-matt

Original comment by benbrinkmann on 15 Feb 2012 at 7:18

GoogleCodeExporter commented 8 years ago

Original comment by snd...@gmail.com on 17 Feb 2012 at 5:28

GoogleCodeExporter commented 8 years ago 
Changed to Butterworth filter.

Original comment by zack.i...@gmail.com on 20 Feb 2012 at 11:56

GoogleCodeExporter commented 8 years ago

Original comment by jwfro...@gmail.com on 21 Feb 2012 at 7:40

GoogleCodeExporter commented 8 years ago 
Merging this into 157 to reflect general problems with the filters rather than 
them being reversed.

Original comment by jwfro...@gmail.com on 24 Feb 2012 at 9:35

GoogleCodeExporter commented 8 years ago

Original comment by samd.p...@gmail.com on 17 Mar 2012 at 3:42