Dear FreeSurfer Devs,

I'm following the fswiki (MailScanner has detected a possible fraud attempt from "secure-web.cisco.com" claiming to be https://surfer.nmr.mgh.harvard.edu/fswiki/CoordinateSystems) to map sEEG contacts to the template cvs_avg35_inMNI152.

I've got MNI305 coordinates for each contact, and I want to plot them on cvs_avg35_inMNI152.

According to the wiki, once I have MNI305 coordinates, I have two choices to calculate the transform from MNI305 to the scanner RAS for this template (instruction 2 and 8b, see below for comparisons).

Normally I use instruction 2 since it seems to be more general and can be applied to any template. For MNI152-based templates, instruction 8b seems to be more straightforward. According to your Wiki, these two instructions would yield the identical mapping.

However, a recent case suggested a result of ~5mm difference between these two methods.

My questions are:

* Is this error range expected/acceptable in the eyes of FreeSurfer?

* Which image defines MNI305 RAS. (For example, MNI152 RAS usually aligns with the scanner RAS of one of ICBM 152 Nonlinear atlases version 2009)

* If I need more accurate scanner->MNI305 mapping (instead of using talairach.xfm), what's your recommendations on FreeSurfer templates that are not sitting in MNI152a/b/c space?

Here's my case study:

Here is a screenshot of the native subject's brain. The two contacts (in red) are at the level of thalamus.

Choice 1: Following the instruction 2,

> I have an RAS point on the surface (tkrR tkrA tkrS) ("Vertex RAS" from tksurfer) and want to compute the MNI305 RAS that corresponds to this point:

> MNI305RAS = TalXFM*Norig*inv(Torig)*[tkrR tkrA tkrS 1]'

This is the method that I usually use. I can calculate MNI305 RAS from scanner RAS via talairach.xfm (in mri/transforms/). This transform is

1.0540    0.0028   -0.0071   -0.0761
-0.0105    1.0914    0.0090    1.7791
0.0201   -0.0599    1.1014   -5.4613

Result: the contacts are at the level of brainstem

Choice 2: instruction 8b

However, since the template already aligns with MNI152, as an alternative, I can follow the instruction 8b,

> I have an RAS from a voxel in MNI152 space which I want to convert to MNI305 space (fsaverage space). Create a vector of the MNI152 space point like v = [R A S 1]'. Multiply this vector by the matrix below (ie, M*v)

    1.0022    0.0071   -0.0177    0.0528
   -0.0146    0.9990    0.0027   -1.5519
    0.0129    0.0094    1.0027   -1.2012

Notice the translation on Z axis is -1.2mm instead of -5.5mm
Result: This mapping seems better

Best,

- Dipterix