Hello, 

I don't think it could be the surface of a gyrus. The data are smoothed and it would never produce such sharp edge. 

It was actually caused by the mask around cerebellum.

O.

===============================
Ondrej Zika
DPhil Student (St. Anne's College)
pain&mind research group
Functional Magnetic Resonance Imaging of the Brain (FMRIB)
Nuffield Department of Clinical Neurosciences
Level 6, West Wing
John Radcliffe Hospital
OX3 9DU, Oxford
________________________________________
From: [log in to unmask] [[log in to unmask]]
Sent: 10 October 2017 19:50
To: Ondrej Zika; [log in to unmask]
Subject: RE: [SPM] Sharp edge in activation on glass brain preview

Yeah, but I wouldn't worry about that. Remember the glass brain takes the
max of across that whole view. Part of that difference is the sharp inner
edge is probably the surface of the parietal cortex, while the outer bit is
the occipital cortex, where less 'activity' is occurring. Its an illusion
based on the fact that that cortical edge slopes down the axial view.


Colin Hawco, PhD
Neuranalysis Consulting
Neuroimaging analysis and consultation
www.neuranalysis.com
[log in to unmask]

-----Original Message-----
From: Ondrej Zika [mailto:[log in to unmask]]
Sent: October-10-17 12:47 PM
To: [log in to unmask]; [log in to unmask]
Subject: RE: [SPM] Sharp edge in activation on glass brain preview

Hello,

I am fine with the sharpness of the outer edge of the brain. I am referring
to the what I circled inside where there is a clear boundary between higher
and lower t values that is very rapid, as if there was a brain edge within
the masked area. Does that make more sense?

Ondrej

===============================
Ondrej Zika
DPhil Student (St. Anne's College)
pain&mind research group
Functional Magnetic Resonance Imaging of the Brain (FMRIB) Nuffield
Department of Clinical Neurosciences Level 6, West Wing John Radcliffe
Hospital
OX3 9DU, Oxford
________________________________________
From: [log in to unmask] [[log in to unmask]]
Sent: 10 October 2017 16:47
To: Ondrej Zika; [log in to unmask]
Subject: RE: [SPM] Sharp edge in activation on glass brain preview

I am not sure I understand what you issue is? What do you mean 'sharp'? Are
you concerned because the areas outside the brain are masked out? because
yes, SPM does apply a mask, even if you select the no masking option. Any
regions with 'poor' signal in any session will be masked out.

You can view the mask in the mask.nii file.

I have always found the SPM masking to be too strict so I have been known to
reduce the threshold. It is defined in spm_defaults.m, line 69 (at least
line 69 in my version).

defaults.mask.thresh    = 0.5

I older versions it was 0.8 which is quite harsh I have gone down to 0.2 on
occasion, and it can be set to zero if there is a compelling reason to do
so.

I am not sure this is what was the issue for you though.


Colin Hawco, PhD
Neuranalysis Consulting
Neuroimaging analysis and consultation
www.neuranalysis.com
[log in to unmask]



-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On
Behalf Of Ondrej Zika
Sent: October-10-17 11:13 AM
To: [log in to unmask]
Subject: [SPM] Sharp edge in activation on glass brain preview

Hello,

Potential issue:
http://users.ox.ac.uk/~ndcn0259/pain_learningI/our_actual_data_artifact2.png

The image shows a contrast activation set to p = 1 in the glass brain SPM
preview. I wanted to simply check the coverage of my signal because I
previously had issues with parts of the midbrain missing which I then fixed
using bias field correction on the EPI images.

The task has 4 separate sessions which are modelled in the GLM.  My initial
thoughts were that there is a problem with the registration. However, check
reg clearly shows that that isn't the case. The registration and
normalisation worked top notch. Furthermore, the data were smoothed with 8mm
kernel, so the edge of the brain simply isn't this sharp. I suspect that
during model estimation SPM creates a mask that is specific to each session
but I couldn't find any information as to whether this is actually the case.

Would anybody have any ideas? I include description of preprocessing and
acquisition below.

Thank you,

Ondrej



EPI sequence: Multi-band, acceleration x 3, PAT x 2, TR = 1570 ms, TE = 30
ms, FOV = 216 mm.

Preprocessing:
1) Brain extraction (BET, FSL)
2) Bias Field correction - EPI segmented, BF estimated and applied to each
volume using imcalc
3) Pre-smoothing (4mm for next step)
4) AROMA ICA (FSL), this involves pretty much the whole FSL preprocessing
pipeline to get the epi2struct and struct2norm matrices.
5) Temporal reslicing (SPM)
6) Reslicing and Unwarping, incl FieldMap (all SPM)
7) Coregistration of mean EPI to struct (SPM)
8) Segmentation of struct (SPM)
9) Normalisation (standard SPM, not DARTEL)
10) Smoothing 8mm (SPM)