Hi,
I studied the FreeSurfer code for MRISinflateBrain and came across a discrepancy in how the sulcal depth is computed. This explains why the values are much smaller than expected.
My question @Bruce is if this is intentional or (as the function names below indicate) possibly a bug in the earlier FS code which has not been fixed later as otherwise FS sulcal depth measures would no longer be comparable to previous versions ?
The FS source code file utils/mrisurf.c contains two functions
- mrisTrackTotalDisplacement - mrisTrackTotalDisplacementNew
The first (old) one is used by MRISinflateBrain. It sums up the normal components of the node forces. The second one sums up the actual node displacements in normal direction. These are identical for a usual forward Euler integration without momentum. The difference is caused by the Euler integration with momentum, where the integration update is
v->odx = delta_t * v->dx + momentum*v->odx ; v->ody = delta_t * v->dy + momentum*v->ody ; v->odz = delta_t * v->dz + momentum*v->odz ;
The values integrated by MRISinflateBrain (i.e., mrisTrackTotalDisplacement) are
nc = v->dx * v->nx + v->dy * v->ny + v->dz * v->nz ;
but the mesh nodes are actually moved by the displacement [v->odx; v->ody; v->odz]. Given a momentum of 0.9, the difference in absolute displacement is quite noticeable.
Thanks, Andreas
On 26 Oct 2015, at 16:30, Simon Vandekar simonv@mail.med.upenn.edu wrote:
Hi Bruce,
The units are not standardized because the standard deviation across the surface does not equal 1.
From line 227 in mris_inflate.c it says that the sulc file is demeaned before it is written out:
MRISzeroMeanCurvature(mris) ; /* make sulc zero mean */
However I am still unclear on what the units are. They are too small to be millimeters. Given other published measurements of sulcal depth I think they might be in centimeters, but can you confirm?
Thanks again, Simon
Hi Simon
I just took a look at the code and I believe that we make it 0 median (not mean) and unit standard deviation. The curvatures can get quite large in regions where the vertex spacing is small and the median is robust to this.
cheers Bruce
On Sat, 24 Oct 2015, Simon Vandekar wrote:
Hi Bruce,
I think that unnormalized sulcal depth would be useful. However, I am also curious what the units are in the standard output, because they appear to not be z-score standardized in each hemisphere.
Thanks again, Simon
H I Simon Do you need the unnormalized sulc? It would be pretty easy to put something in to generate it Bruce
On Oct 22, 2015, at 3:23 PM, Simon Vandekar simonv@mail.med.upenn.edu wrote:
Hi Bruce,
I chose a random subject from the ADNI dataset and computed the mean and variance for the left and right hemispheres separately. It looks like the mean is close to 0, however the variance is not 1. Do you think it is possible that the surfaces are demeaned, but the units are still something meaningful (i.e. not unitless)?
sulc = read.table('rh.sulc.asc', header=FALSE, col.names=c('v', 'x', 'y', 'z', 'sulc'))
mean(sulc$sulc)
[1] -1.236923e-08
var(sulc$sulc)
[1] 0.3276958
quantile(sulc$sulc)
0% 25% 50% 75% 100%-1.72503 -0.42128 -0.03963 0.40072 1.80980
Thanks again, Simon
Hi Simon
they are originally mm, but I believe we z-score normalize them so that the sulc file is 0 mean and unit variance.
cheers Bruce On Wed, 21 Oct 2015, Simon Vandekar wrote:
Dear Freesurfer experts,
I am reading the manuscript "Cortical Surface-Based Analysis II: Inflation, Flattening, and a Surface-Based Coordinate System" by Fischl, Sereno, and Dale (1999) where the estimation of sulcal depth is described on page 199, however from the description I am still unsure of the exact units of sulcal depth. Are the units for sulcal depth in mm or are they in arbitrary units?
Thanks in advance, Simon
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
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.