Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Adaptive enhancements #2258

Open
5 of 6 tasks
t-b opened this issue Sep 12, 2024 · 7 comments · May be fixed by #2302
Open
5 of 6 tasks

Adaptive enhancements #2258

t-b opened this issue Sep 12, 2024 · 7 comments · May be fixed by #2302
Assignees
@t-b

Comments

@t-b
Copy link
Collaborator

t-b commented Sep 12, 2024 •
edited
Loading

  • Don't do linear estimation for DASCale values when the slope is negative or zero, instead go in DAScaleNegativeSlopePercentage steps to larger DAScale values. If negative or zero with further increase in DAScale, this is the max DAScale that is acquired. Then fill in, as described next (setQC is passing in this case, independent of NumSweepsWithSaturation):
  • With negative slope, fill in between last DAScale with positive slope and first negative/zero slope DAScale based on DAScaleFillInPercentageNegativeSlope
  • Round to pA in SetDAScale
  • Add prefix (F or R or X) in "DAScale values left" to mark where the value came from
  • Add an analysis parameter to determine how many SCIs we look back for failing adaptive suprathreshold data. We currently only look at the very previous one.

negative means negative and zero.

  • Additional passing criteria: 1. two negative slopes in a row. 2. a FI slope QC passing and one negative slope in a row (in this order). If a negative slope is obtained, subsequent slope measurements are made from the last da scale that gave a positive slope (this is not reset after a possible positive slope as we want to ignore negative slopes always). After 1 or 2, one additional sweep is acquired with a DA scale that splits the difference between the last da scale with a positive slope and the first with a negative one. We still do frequency fillin as usual.

For all sweeps with negative slope we don't care about post pulse baseline QC. The purpose here is to minimize the number of sweeps acquired with depolarization block.

DAScaleNegativeSlopePercentage (1-100% of maxDASCaleStep) is applied after the first negative slope to set the DA scale.

New behaviour applies to RhSuAd data as well.
@t-b t-b self-assigned this Sep 16, 2024
@t-b t-b mentioned this issue Sep 17, 2024
Closed
@t-b t-b linked a pull request Nov 26, 2024 that will close this issue
Draft
8 tasks
@t-b

This comment was marked as outdated.

Sign in to view
@timjarsky

This comment was marked as outdated.

Sign in to view
@t-b

This comment was marked as outdated.

Sign in to view
@t-b

This comment was marked as outdated.

Sign in to view
@t-b
Copy link
Collaborator Author

t-b commented Jan 21, 2025

DAScaleNegativeSlopePercentage (1-100% of maxDASCaleStep) is applied after the first negative slope to set the DA scale.

I don't think restricting it to first here makes sense. I'll do it for each.

@t-b
Copy link
Collaborator Author

t-b commented Jan 21, 2025

@timjarsky Is it okay if I do

For all sweeps with negative slope we don't care about post pulse baseline QC. The purpose here is to minimize the number of sweeps acquired with depolarization block.

only for the current adaptive data and not from the previous SCI data (RhSuAd)? Doing it for the previous ones complicates it a lot.

@t-b
Copy link
Collaborator Author

t-b commented Jan 26, 2025

@timjarsky In Adaptive we have the following code

	// original data:
	// apfreq:  {5,  8,   9,   9,   15}
	// DAScale: {50, 100, 130, 160, 200}
	// numPoints = 5
	//
	// after filtering:
	// apfreq:  {5,  8,   9,   15}
	// DAScale: {50, 100, 130, 200}
	// deletedIndizes: {3}
	// numPoints = 4
	[WAVE apfreqFiltered, WAVE DAScalesFiltered, WAVE deletedIndizes, numPoints] = PSQ_DS_FilterFrequencyCurrentData(apfreqSorted, DAScalesSorted)

which was added in 9e1b1e1 (PSQ_DAScale: Change fitting logic and drop NumPointsForLineFit, 2024-07-10). I do think that with our negative slope (which includes zero) handling now, this filtering is actually detrimental to the result. Should I remove it?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@t-b t-b

Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Enhance adaptive suprathreshold AllenInstitute/MIES
2 participants
@t-b @timjarsky