Sven

See attached our DTI protocol as part of a dementia protocol for 3T 750 MRI (not a wide bore)(page 7)

Aad van der Lugt

Dear Aad

Thanks a lot !!

Sven

Derek Jones suggests at least 31 diffusion directions, and 1 bzero (T2-weighted) volume for every 6 diffusion directions. I would also suggest 2mm isotropic voxels and a bvalue of 2000 or 3000. If you have access to it, you could also use multiband to reduce scanning time.

Dear Derek. 

Thanks a lot for the suggestions! In fact we have a problem that we cannot increase the number of slices >43. As I want a 2mm slice thickness, this means that we cannot cover the entire brain. I do not see a logical reason for this. Do you have the same problem? Or any idea to overcome this limitation?

Thanks

Sven

What sort of head coil and FOV are you using?

Regards,

Jerome

Hi Jerome

We do have the 32 CH head and 18 CH head&neck. We tried both coils and both did not work

For the FOV we used 250 mm

Thanks 

Sven

OK. What is your TR? Using the fat-sat may help as well.

Are you using ASSET? It reduces echo spacing, so technically you can get more slices in a given TR period.

Regards,

Jerome

Dear Jerome

Thanks for the answer, and I apologize for the long delay as I was away.

TR is 8966ms

We use fat sat. And ASSET

So nothing we could really do other than currently, correct?

Sven

Hi Sven,

Increase your TR to provide more slices (as TR is not an issue with EPI).

Regards,

Jerome

Thanks a lot !! Will try !

Hi Barbara

Thanks a lot for your answer! I would be very helpful if I could see the image quality that you got, so I can get an idea whether our image quality is in line with that of other centers !

Thanks

Sven

thanks a lot for sharing the data !!

the visual impression of these scans is comparable with the quality of our images, so I am happy about this

one question: you also have 5 b0 images, this seems to be a particularity of GE as I did not know this from Siemens or Philips. Is this really useful and improves the quality of the fit ?

Thanks

Sven

That's a nice protocol Barbara!

Is there a reason you chose a b-value of 1000 (rather than 2000 or 3000)?

Indeed, that article by Derek Jones is full of great advice.

If you have time, you may want to consider 10 b0 volumes (it has been suggested that 1 bo volume for every 6 directions is optimal).

Based on your in-plane resolution, I assume your matrix is something like 512x512? If so, then changing it to 256 x 256 should produce isotropic voxels.

Is your scanner a 750 or 750w? Did you use a GE 32 channel head coil or a NOVA head coil?

Regards,

Jerome

For a 1.5T, I would suggest 32 directions at a single bvalue of 1000 with 2 bzero volumes, and spatial resolution of 2.5mm. If you have a research key, then you can acquire isotropic voxels, and also interleave the bzero volumes. Which head coil and console DV software version are you using?

Jerome

4.7 is the AW. What version is on the console (e.g. DV28)?

A research key is something you apply to GE for. It opens up the scanner to research mode (by default the scanner is in clinical mode) and therefore allows you access to hundreds of CVs (control variables) that clinical mode does not show.

Thanks for the answers ! In the meantime I have changed my job, and now work with another vendor :-)

Sven