Print

Print
Dear John,

Many thanks for this. The data was indeed acquired on a 1.5T scanner. I was provided with eight files from the fieldmap sequence, and was told that they were two "modulus" images (of different echo times), two "real" images, two "imaginary" images and the two phase images. I've attached a pdf of these four image types in order (just for one echo time each). 

I was advised to use the two modulus images as the magnitude images. This is different to what I'm used to because I have previously always just had one magnitude and one phase image to work with. Is it possible that the modulus image is not the correct image to use for this (I've not heard of this type of file before)? I tried using the "real" images as the magnitude instead and the results look even more strange (in my novice opinion!). I've attached these as well.

Or is this a situation where skull stripping of the modulus image could help?

Many thanks again,
Antoinette




On Tue, Oct 15, 2013 at 6:26 PM, John Ashburner <[log in to unmask]> wrote:
The presence of the scalp in the images is messing things up a bit.  Phase data needs to be unwrapped (in degrees, the values go ... 357, 358, 359, 0, 1, 2, 3 ... so - depending on context - the unwrapping needs to figure out that these need to be changed to ...357, 358, 359, 360, 361, 362, 363, 364..., or maybe -3, -2 -1, 0, 1, 2, 3).  The unwrapping has problems in regions where the magnitude is close to zero (eg skull) because the phase is pretty noisy.

Low magnitude regions are ignored by the unwrapping, which generally works OK when there is little or no scalp visible in the images.  For some reason though, your magnitude images contain scalp.

I'm more used to seeing magnitude images that look a bit more like the typical EPI data.  I relayed your email to Nik Weiskopf, who heads up the Physics Group in our department.  His response was:

I think the skull and scalp appears slightly strange in the magnitude
and phase maps. Were these data acquired at 3T or 1.5T? If it is 1.5T
data, then a fat-water cancellation effect may have occurred, since
the echo spacing is more appropriate for 3T (but still not perfect) to
avoid differential cancellation between the two echoes.

However, I am not sure if this will affect the actual distortion
correction results significantly. It is difficult to tell from the
unwarped EPI images on the pdf printout.

All the best,
-John



On 15 October 2013 14:25, Antoinette Nicolle <[log in to unmask]> wrote:
Dear experts,

I'm attempting to use the fieldmap toolbox with fieldmaps acquired from a Philips Achiever scanner. I have two magnitude images and two phase images, with echo times of 4.9 ms and 7.6 ms.

I don't get any error messages, and the toolbox seems to run through so that I get vdm files. However, the vdm, fpm and bmask images look quite strange, almost as if they are not being masked properly. I've attached the output from the fieldmap toolbox ("fieldmap_generation.pdf") as well as the original images ("fieldmap_images.pdf", with my two magnitude images on the top row, then my two phase images on the middle row, and then my mask and vdm on the bottom row).

Do these images look strange to anyone else and has anyone encountered such a problem before? I wonder if perhaps I am using the wrong magnitude images, or if I should adjust any parameters for the masking?

Many thanks,
Antoinette