smburns47 / preprocessingfnirs Goto Github PK

fNIRSFilterPipeline is breaking for me at hmrIntensity2Conc, which doesn't seem to play nice with NaNs

I thought that moving d(:,SD.MeasListAct==0)=NaN; down below line 22 [dconverted, ~] = hmrIntensity2Conc(dfiltered, SD, samprate, 0.008, 0.2, [6, 6]); might fix the issue, with d being changed to dconverted.

I am not sure if it is problematic to be doing motion correction on bad channels and then NaN them out later, but not sure of another workaround to get hmrIntensity2Conc to work.

BaselineVolatilityCorrection is also converting some frames of some channels to NaNs (not sure how?), even when there are no bad channels, which seems like a bug.

Thanks for fixing the issue with BaselineVolatilityCorrection! I continue to get an error that prevents preprocessing from hmrIntensity2Conc when it is called up in fNIRSFilterPipeline. I found that removing d(:,SD.MeasListAct==0)=NaN; and adding dconverted(:,SD.MeasListAct==0)=NaN; after hmrIntensity2Conc gets applied fixed the problem, and then I was able to run.

I'm not sure why this error has occurred for me but not others. My reasoning behind why it might not work is that hmrInensity2Conc cannot accept NaN values. So by introducing NaN values into d too early on (and then d becomes dfiltered), they may be interfering with the ability of hmrInensity2Conc to run.

Just want to double check that it's reasonable to move it, without interfering with any preprocessing steps, as I'm not entirely certain about how it works, and don't know if certain calculations look across channels (in which case values that will not be used should not be included).