Advise on how to handle MRI Data

Hi, I am looking for a way to count and get volume measurements for catonized feratin labeled glomeruli in MRI images. The glomeruli are black dots in the kidney depending on the where you are in the kidney they are in the outter tissue. the kidney’s are 200 to 500 Images per stack. Because of some noise, I have not been able to find a good threshold that lets me segment the glomeruli. I am looking for some tips etc. that might aid me in sgementation of the glomeruli. I am attaching 2 Images, one unlabed and one with red dots roughly representing what I would like to segment A-1219-19-RK-0140StacktoImage.tif (230.2 KB) A-1219-19-RK-0140Labeled.tif (347.2 KB)

Hi Susan,

Since your glomeruli are in the cortical area, have you tried to first segment the kidney in cortical vs medulla and background? Once this is done, you can create a mask that only contains the cortex which should make the segmentation easier.
Afterwards, invert your image and use the Find Maxima function (Process>Find Maxima).

I get this:
image

M.

The issue is how to robustly define the medullary region. How did you define what was medullary and cortex for your example?

Hi Susan,

I used Weka Segmentation (Plugin>Segmentation>Trainable Weka…) but I guess Ilastik will work too.
Tip: train for 3 labels in this order: 1) background, 2) Medulla and 3) Cortex
Once trained, export the segmentation image. The values therein will be 1, 2 and 3. You can see them using Glassbey LUT. You then Threshold for the value that correspond to the cortex, Analyse Particles for the large area that is the cortex (i.e you’re filtering the small bits) and you then have a ROI for the cortex.
Project this ROI to your original image and clear outside (Edit>Clear Outside).
Go back to the binary image and invert it, use Analyse particles to find the area that correspond to the Medulla.
Project this ROI to your original image and fill (Edit>Fill)

Sincerely,

M.

Thanks, this is really useful! I’ll give it a try.