MR sequences

[kopachini]

Hi,

I tried to segment urinary sistem of one patient which had to do MR urography. As I am resident in the beginning and I still didn't have much real contact with MR  I know, you are probably wondering how is that possible, but that's how residency program goes in Croatia), I don't really know which sequences/studies are the best to visualise or better said for segmentation of urinary sistem?

Presume that they should be isovoxel/space sequences, right? Does anyone have any expirience with that? What do you use?

Thx.

[Dr. Mike]

Don't do much MR Urography, but I would guess that T2 sequences should be best. T2 sequences show water (like urine) as very bright. If you want to be ideal, maybe you can get a fat-suppressed T2 sequence that will make segmentation even easier.

 

Hope this helps.

 

[kopachini]

yeah... i know about T2, but i can't get that isovoxel look as on CT, tha is the problem :/

I'll try something on 3 T MR with my collegue and will let you know if there is improvement in looks.

thx anyway.

[mikefazz]

For segmentation I would try a region growing method.

 

The airway segmentation extention may work if you want to get connected thin parts.  GrowCut or FastGrowCut may work but I think they don't work as well with thin features.

 

 

[kopachini]

If you look at these images, you can see that resolution is good only on one section (transversal on the right pic), and other two section are distorted.

this is the main problem, how to get good resolution on all three planes.

[mikefazz]

From the two images the first scan shows some mis-ordered images (some of posterior slices connect to anterior slices).  This suggests the slices are in the coronal plane but not ordered correctly.  The second scan shows that the slices are in the transverse plane with large slice spacing.  This can be confirmed by looking at the volume dimensions in the volumes section (see picture).  My guess is the second scan has about 1mm pixel size and the out of plane is 3+mm (look at "Image Spacing" like in attached image).  Often clinical scans have large slice spacing as they are for a radiologist to look at not for making 3D models.  Sometimes the original scan data will be of higher resolution and you can go back and ask for a higher quality (lower slice spacing) scan data.

 

 

As to the first scan getting images ordered correctly can be difficult.  I'm not sure what 3DSlicer uses to order images (ImageNumber, ImagePositionPatient, file name, etc...).  A separate program like Fiji (ImageJ) can make an nrrd file but when I have tried it doesn't load into 3DSlicer since Fiji doesn't include all the header info that 3DSlicer is looking for.  I use some matlab code I wrote to fix scan data and make an nrrd file for 3DSlicer but not everyone has matlab...  

 

[kopachini]

thx for the reply @mikefazz. this study was on 1,5 T MR... next time i will try MR urography with one of my collegue on 3 T machine for better resolution an smaller slice spacing.