Hello,
I am running into a problem when applying the HbO, HbR and HbT montages on some of our data.
Here you can see that the trial file has all of these values. We can see the waves for the HbO, HbR and HbT values.
However, when I run the pipeline to apply the montage (standardize–> apply montage –> HbO), no values are shown on this file. It is simply blank with a flatline.
Because it is doing this, when we extract the values, they all come out as NaN.
I was wondering if anyone had run into this issue or would know why it would be doing this.
Hi @annikafleming12,
It seems a figure was missing in the original post.
Indeed, that is strange, the NIRS montages are just for selecting channels, they should not change the values on the data. I tested on my side and it looks good.
You mentioned that the problem is present for "some of our data".
Can you share an example of data that presents this problem?
@edelaire, have you encounter this behaviour?
Hello,
I only saw it when @annikafleming12 showed it to me few weeks ago. So the original file looks fine; but then after apply montage it’s only NaN.
Indeed, having access to the subject file would be useful for testing.
Hi @edelaire & @Raymundo.Cassani ,
Sorry for the delayed reply!
Here is one of the subject protocols that is giving us this issue. For this subject specifically, we are running into the NaN problem with HbO and HbT, but not HbR.
KIN209 Protocol
Let me know if this link works or if it’s better for me to share it another way.
Thank you so much for your help!
Annika
Hi @annikafleming12,
Thanks for sharing the data. I was able to figure out what is going on.
Summary
Yes, for the data A1 / A1 (#1)2-150 the channels S18D16HbO and S18D16HbT are all NaN values.
- When the montage HbO is applied in the GUI, there is this message in the Matlab command window:
BST> Error: Matrix F contains NaN values, replacing them with zeros. Please check your data files.
- When using the process, these
NaN values are not replaced, so applying the montage (matrix multiplication) results in all NaN.
Tracking the origin of those NaN
Checking the history of the data A1 / A1 (#1)2-150, it can be seen that the NaN values for that were introduced by the process_nst_motion_correction (Motion Corrected (TDDR)), so the file KIN209_hb_02 / Raw (0.00s,1657.56s) | Hb [Topo] | band(0.005-0.1Hz) | Motion Corrected (TDDR) already has NaN values.
Exploring further the data, I noticed that before the motion correction (and before the band pass), the three channels S18D16{HbO, HbR, HbT} are flat and several orders of magnitude smaller than the time series for other source-detectors.
It seems there is a problem with these channels even since the raw data, where the channels S18D16WL762 and S18D16WL841 (both highlighted in the figure below) are flat the entire recording.
1 Like
