[Homer-users] Questions about AtlasViewer on short separation channels and image reconstruction

Mon Dec 4 15:51:40 EST 2017

Please pardon the answer if it is at all incorrect.  I don’t use AtlasViewer much (at all) since I use my own code.  However, here are the answers.

1.
In the ImageRecon.m function (image_recon_Callback subroutine; line 355), the short-separation threshold is being used to mask out any short channels.  As I read it, this option should be used if short-separation regression was ALREADY applied during preprocessing prior to image reconstruction.  If you apply short-separation regression in preprocessing, then these channels will be “0” and so you don’t want to use them in the image recon and use a brain only model.  The alternative is to not regress in preprocessing and then use a brain+skin model for the image recon.  The ss-threshold should be the same distance cutoff used to specify the short distances in the preprocessing step.  I don NOT think you should use BOTH short-operation regression AND the scalp model.

2.  I am not sure on the notation in Custo paper, but the code is using alpha (in the hmrImageReconConc function) as the square-root of what would typically be denoted as “lambda” in a Tikhonov regularization.  IMG = A’ * inv( A* A’ + alpha^2 * I )* Y.   There are some ways to estimate alpha (or lambda; e.g. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3018091/ or an L-curve).  Under a Bayesian formulation of the model, alpha^2 is the ratio of the covariance of the parameter noise and the measurement noise (e.g. alpha is the signal to noise ratio).  1000 seems high to me (SNR~30) but...

3.  The scalp model is only used in the case that you do NOT do short separation regression as a preprocessing.    This should be a matrix as the result of the Monte Carlo model including a layer for the scalp based on the registered probe.  However, I am not sure where this comes from in the GUI.  (probably an option when you generate the forward model in the first place).

4.  It looks like this gets loaded through either the atlasViewer.mat structure or the folder structure in the getImgRecon.m function depending on what is available.  If loaded in the atlasViewer, you should be able to open the file, type atlasViewer.imgrecon=[]; and then resave/overwrite the file.  In the folder form, I think you just need to delete the “/imagrecon” folder.  I am not sure how to do it through the GUI.
I didn’t test this (again, I don’t really use AtlasViewer), but you may be able to trick the GUI at runtime by typing on the command line
>> global atlasViewer
>> handles = guidata(atlasViewer.handles.figure);
>> handles.hGroupList=atlasViewer.handles.hGroupList;
>> handles.ImageRecon=[];
>> atlasViewer.imgrecon=initImgRecon(handles);

This should call the initialization code again, but pretending like you don’t already have image recon data

5.  Sorry.  Not sure

6.  There probably is a better way to do this, but here is a hack that might work.

The surface map is in surface (Monte Carlo space).
With AltasViewer loaded.  In matlab type:

>> global atlasViewer
>> T_headvol2mc = atlasViewer.headvol.T_2mc;   % Colin to Monte Carlo Space t-form
>> T_labelssurf2vol = atlasViewer.labelssurf.T_2vol;  % MNI to colin
>> MNIcoor = xform_apply([point on surface], inv(T_headvol2mc*T_labelssurf2vol ) )

If you use the DataCursor button on the menu (looks like a cross-hair) it will tell you the value of the point on the surface where you click.

-Ted

Theodore Huppert, PhD
Associate Professor
Department of Radiology
Department of Bioengineering
Center for Neural Basis of Cognition
Clinical and Translational Science Institute

University of Pittsburgh
Email: huppertt at upmc.edu<mailto:huppertt at upmc.edu>
Phone: (412) 647-8459

On Dec 4, 2017, at 8:07 AM, Jörn M. Horschig <jorn at artinis.com<mailto:jorn at artinis.com>> wrote:

Dear Homer developers and users,

We would be grateful if someone could shed light on below questions. Also partial answers are more than welcome.

With best regards,
Jörn

--
Jörn M. Horschig, PhD, Software Engineer & Project Leader NeuroGuard XS
Artinis Medical Systems  |  +31 481 350 980
Sign up for our NIRS newsletter

-----Original Message-----
From: homer-users-bounces at nmr.mgh.harvard.edu<mailto:homer-users-bounces at nmr.mgh.harvard.edu> [mailto:homer-users-bounces at nmr.mgh.harvard.edu] On Behalf Of saskia at artinis.com<mailto:saskia at artinis.com>
Sent: Friday, November 17, 2017 12:04
To: homer-users at nmr.mgh.harvard.edu<mailto:homer-users at nmr.mgh.harvard.edu>
Subject: [Homer-users] Questions about AtlasViewer on short separation channels and image reconstruction

Dear Homer and AtlasViewer users,

I am Saskia Kruitwagen and I am doing my masters internship at Artinis Medical Systems. I am preforming NIRS measurements with a probe which includes short separation channels.
For the data processing and analysis, I am using Homer and AtlasViewer, which are fantastics tools for analyzing fNIRS data. However, I have a few questions about AtlasViewer:

1. I have a question about the processing step involving the short
separation channels. I can do this in Homer, using enAdaptiveSS or a GLM filter. However, in AtlasViewer there is also an option ‘short separation threshold’. Therefore, I was wondering, when to should I do the processing step with the short separation channels: in Homer or in AtlasViewer?
What is AtlasViewer exactely doing with the short separation channels?
In AtlasViewer, I can set a ‘short separation threshold’ when I do image reconstruction. However, this option gives me errors when this value is nonzero. So, is this option even working?

2. The second question is about image reconstruction in AtlasViewer. For
the image reconstruction I have to specify an alpha value. This is the regularization parameter. The default value is 1000. In Custo et al.
(2010) I looked up how to calculate alpha. They give a formula (following formula 3) how to calculate a tau, based on the sensitivity matrix. To my understanding alpha in AtlasViewer is the same parameter as tau squared in Custo et al. Is this correct? When I use the formula in Custo et al, alpha is 10, which is much lower than the default of 1000. Is this a sensible value?

3. My third question is about image reconstruction as well. In the
ImageRecon window I can either choose ‘brain only’ or ‘ brain and scalp’. For this second option, I need the file “fw/Adot_scalp.mat”.
Since I apparently do not have this file, AtlasViewer gives me an error.
How can I obtain this data file? And what is the difference between both options?

4. Next, when AtlasViewer has calculated the image matrix, with certain
thresholds, these values are stored and cannot be changed initially set.
The calculation dialog does not open a second time. If I want to change the thresholds, how can I perform this calculation again?

5. What is the unit and scale of the resolution images? In Aasted et al.
it says, “a linear scale in millimeter” for the localization error, but does that hold as well for the resolution profile? It seems to me that the scale is flipped, because resolution has the lowest values where I have a very dense probe profile.

6. My last question is: how can I obtain the MNI coordinates of my
activation map?

It would be nice if someone could help me out?

Best regards,

Saskia Kruitwagen
Master student Biomedical Engineering, UT Twente, The Netherlands _______________________________________________
Homer-users mailing list
Homer-users at nmr.mgh.harvard.edu<mailto:Homer-users at nmr.mgh.harvard.edu>
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/homer-users

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

---
This email has been checked for viruses by AVG.
http://www.avg.com

_______________________________________________
Homer-users mailing list
Homer-users at nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/homer-users