[Simnibs-discuss] Targeting subject specific locations on the scalp

Guilherme B Saturnino guilhermebs at drcmr.dk
Wed Jul 15 07:57:26 CEST 2020


Hi Dan,


This transformation is embedded inside the NifTi files. It's often 
called the "affine". Normally the X axis corresponds to left/right, Y to 
posterior/anterior and Z to bottom/top, but it can vary depending on how 
the image was acquired and how the subject was positioned inside the 
scanner, so it is not so reliable. However, the co-register images, 
located in the m2m_{subID}/toMNI folder, are sure to follow this 
convention, if the co-registration worked (it generally does). We also 
use the co-registrations to calculate the 10-10 EEG positions, including 
FP1, which are stored in 
m2m_{SubID}/eeg_positions/EEG10-10_UI_Jurak_2007.csv.



Best Regards,

Guilherme


On 14/07/2020 23.12, Mccalley, Daniel wrote:
> Hi all,
>
> Thanks again for the help!
>
> Oula, would you be able to point me toward the header file that may 
> have information on the voxel-to-world transformation? It would be 
> useful to know which direction is up/down, left/right, 
> forward/backward if X Y and Z aren't always accurate
>
> Best,
>
> Dan McCalley
> Medical University of South Carolina
> College of Graduate Studies, Neuroscience
>
> ------------------------------------------------------------------------
> *From:* Oula Puonti <oupu at dtu.dk>
> *Sent:* Monday, July 13, 2020 2:44 AM
> *To:* Mccalley, Daniel <mccalled at musc.edu>; Guilherme B Saturnino 
> <guilhermebs at drcmr.dk>; discuss at simnibs.org <discuss at simnibs.org>
> *Subject:* Re: [Simnibs-discuss] Targeting subject specific locations 
> on the scalp
> *
>
> CAUTION: External
>
> *
>
> Hi Daniel,
>
>
> Thanks for the clarification.
>
>
> I could see two problems:
>
>
> 1) The Nz coordinate from SimNIBS is not correctly placed. This can 
> happen due to segmentation/registration errors. You should open the 
> MRI scan in a viewer, for example freeview, and check that the Nz 
> coordinate that SimNIBS outputs matches the nasion location on the scan.
>
>
> 2) The header, i.e., voxel-to-world transformation, of the MRIs is 
> different between subjects. If you always substract the X and Y 
> distances you might end up moving in a different direction than intended.
>
>
> Best,
>
> Oula
>
> ------------------------------------------------------------------------
> *From:* Mccalley, Daniel <mccalled at musc.edu>
> *Sent:* Friday, July 10, 2020 6:43:03 PM
> *To:* Oula Puonti; Guilherme B Saturnino; discuss at simnibs.org
> *Subject:* Re: [Simnibs-discuss] Targeting subject specific locations 
> on the scalp
> Hi all,
>
> Thank you so much for the fast replies!
>
> Let me clear up a few things about our experiment --
>
> When our participants come in, the experimenter measures the 
> individual's head (nasion to inion, tragus to tragus, circumference). 
> Each person's measurements are of course different and there is some 
> amount of error/rounding made by the experimenter in each measurement. 
> *Our participants are never fitted with an EEG cap* to locate our 
> cortical target (FP1).
>
> Instead, starting at Nz, we take 10% of the distance for Nz to Inion 
> (generally ranges from 2-4 cm, I referred to this as Y in my initial 
> example) and move up and we take 5% of the total head circumference 
> and move that distance left (towards FP1, I referred to this as X in 
> my initial example). The experimenter makes a mark on the persons head 
> with sharpie, but *unfortunately we did not use a fiducial for this 
> study*.
>
> Given that the FP1 location is determined in a patient-tailored manner 
> and we expect some degree of experimenter error in making these 
> measurements, I'd like to as closely as possible recreate an e-field 
> at the site of stimulation for each person using these measurements. 
> Thus far, I have used the coordinates for Nz that SIMNIBs produces and 
> made adjustments from there based on X and Y displacement for each 
> subject. This almost always produces a result close to FP1, but in 
> some instances it seems the result remains near the eye or in entirely 
> unexpected head locations
>
> Does that information help?
>
> Thank you all again for your help!
>
> Best,
>
>
> Dan McCalley
> Medical University of South Carolina
> College of Graduate Studies, Neuroscience
>
> ------------------------------------------------------------------------
> *From:* Oula Puonti <oupu at dtu.dk>
> *Sent:* Friday, July 10, 2020 3:35 AM
> *To:* Guilherme B Saturnino <guilhermebs at drcmr.dk>; Mccalley, Daniel 
> <mccalled at musc.edu>; discuss at simnibs.org <discuss at simnibs.org>
> *Subject:* Re: [Simnibs-discuss] Targeting subject specific locations 
> on the scalp
> *
>
> CAUTION: External
>
> *
>
> Hi,
>
>
> Just to add to what Guilherme said: Are you locating the nasion 
> manually or using coordinates that SimNIBS outputs automatically? If 
> the latter, please check that the fiducial placements look correct.
>
>
> Best,
>
> Oula
>
>
>
> ------------------------------------------------------------------------
> *From:* Simnibs-discuss <simnibs-discuss-bounces at drcmr.dk> on behalf 
> of Guilherme B Saturnino <guilhermebs at drcmr.dk>
> *Sent:* Friday, July 10, 2020 9:31:07 AM
> *To:* Mccalley, Daniel; discuss at simnibs.org
> *Subject:* Re: [Simnibs-discuss] Targeting subject specific locations 
> on the scalp
>
> Hello,
>
>
> SimNIBS 3 offers two ways of getting 10/10 EEG positions out of the box.
>
>
> 1) Use the eeg positions calculated using MNI transformations, which 
> are stored in m2m_{SubID}/eeg_positions/EEG10-10_UI_Jurak_2007.csv. 
> Those can also be accessed in the GUI 
> <https://urldefense.com/v3/__https://simnibs.github.io/simnibs/build/html/tutorial/gui.html__;!!Ab1_Rw!VmHiKnZCbukEHQmiiCDWTKL0HVw7v3spr5BG5nz_3hj_QfY9egavym9kBSlfQpNA$>, 
> which loads it automatically, or in scripts (MATLAB example 
> <https://urldefense.com/v3/__https://github.com/simnibs/simnibs/blob/v3.1.2/simnibs/examples/simulations/TMS.m__;!!Ab1_Rw!VmHiKnZCbukEHQmiiCDWTKL0HVw7v3spr5BG5nz_3hj_QfY9egavym9kBXnVXhnJ$>)
>
>
> 2) Using the eeg_positions 
> <https://urldefense.com/v3/__https://simnibs.github.io/simnibs/build/html/documentation/command_line/eeg_positions.html__;!!Ab1_Rw!VmHiKnZCbukEHQmiiCDWTKL0HVw7v3spr5BG5nz_3hj_QfY9egavym9kBd53sE1Z$> 
> command line tool with the Nz, LPA, RPA and Iz as an input
>
>
> Those two were compared and validated here 
> <https://urldefense.com/v3/__https://link.springer.com/chapter/10.1007/978-3-030-21293-3_1__;!!Ab1_Rw!VmHiKnZCbukEHQmiiCDWTKL0HVw7v3spr5BG5nz_3hj_QfY9egavym9kBSy4FP92$>.
>
>
> I'm not sure why it is not working in your case though, might be due 
> to head size, as you are moving by constant amounts?
>
>
> Best Regards,
>
> Guilherme
>
>
>
> On 09/07/2020 19.48, Mccalley, Daniel wrote:
>> Hi all,
>>
>> Our lab targets the FP1 with TMS. We use a simple measurement system 
>> (http://clinicalresearcher.org/F3/ 
>> <https://urldefense.com/v3/__http://clinicalresearcher.org/F3/__;!!Ab1_Rw!VmHiKnZCbukEHQmiiCDWTKL0HVw7v3spr5BG5nz_3hj_QfY9egavym9kBcP0FOvN$>) 
>> to locate exactly where we want to place the coil for each subject. 
>> Typically, we measure the subject's head, and then, starting from 
>> nasion, we'll move up Y mm and left X mm (where X and Y are 
>> determined from each subject's given head size).
>>
>> Ideally, we'd like to create an electric field model for each subject 
>> that accounts for the TMS parameters as well as the exact location 
>> used. To implement this in SIMNIBS, I've tried the following:
>>
>> 1) Open a subjects m2m file and find the coordinates for nasion (say, 
>> for example, that these coordinates are X= -2.78, Y= 117.59, Z= -20.63).
>>
>> 2) Determine Y and X used to locate FP1 for the given subject (say, 
>> for example, that starting from nasion, we moved 37 mm in the Y 
>> direction, and over 32 mm in the X direction).
>>
>> 3) Subtract the displacements in the Y and X directions from the 
>> original nasion coordinates, resulting in X= -34.78 and Y=80.69
>>
>> 4) Add a new position wherein the reference coordinates match the 
>> original nasion coordinates (and the Define Reference Coordinates box 
>> is checked) and the coordinates calculated in step 3 are input in the 
>> position fields (Z is left blank for SIMNIBs to calculate)
>>
>> 5) Adjust for coil orientation and TMS parameters
>>
>> This seems to have worked well for 90-95% of our subjects, however 
>> there are a few subjects in which the resulting coil placement 
>> appears to be over the eye (Note: Ordinarily I'd attach examples of 
>> the working and not-working subjects, but I'm not sure if the skin 
>> reconstruction of our participants falls under PHI).
>>
>> Am I going about this the wrong way? Could this reflect a problem 
>> with segmentation/meshing? Or maybe a problem with the experimenter 
>> measurements of the subjects head? Any advice would be greatly 
>> appreciated. Thank you all so much!
>>
>> Best,
>>
>> Dan McCalley
>> Medical University of South Carolina
>> College of Graduate Studies, Neuroscience
>>
>>
>> _______________________________________________
>> Simnibs-discuss mailing list
>> Simnibs-discuss at drcmr.dk  <mailto:Simnibs-discuss at drcmr.dk>
>> https://mailman.drcmr.dk/mailman/listinfo/simnibs-discuss  <https://urldefense.com/v3/__https://mailman.drcmr.dk/mailman/listinfo/simnibs-discuss__;!!Ab1_Rw!VmHiKnZCbukEHQmiiCDWTKL0HVw7v3spr5BG5nz_3hj_QfY9egavym9kBfqFDV5E$>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.drcmr.dk/pipermail/simnibs-discuss/attachments/20200715/50f9da96/attachment-0001.html>


More information about the Simnibs-discuss mailing list