How to find the nearest voxel of a 3D image based on some specific intensity value...

So I want to replace the abnormal intensity values from a 3D image. If the intensity is >1800 then I will replace it with nearest voxel value less than 1800.

I believe I can do it with hard coding... but is there any function to do that simply??

[Rik]

What is the nearest voxel? Any voxel that is not on the edge has 4 neighbor voxels. Which should be selected?

[banikr]

The voxel that has lowest value under 1800 should work.

[Rik]

Does the deletion of your answer mean you would like help on this question again?

[banikr]

The problem changed a little bit. I have created a segmentation label image based on the intensity that contains unwanted electrodes(label==20). I want to replace those voxels by nearest labeled(as there are some nearby unlabled==0 voxels) voxels.

Take a look at the file: lab_superior.nii.gz

you can read it using

lab = niftiread('lab_superior.nii.gz');

[Rik]

If you manage to select only the electrodes, you should be able to use watershed to find which label is closer to the invalid labels (store the locations and mark them with 0).

[banikr]

Could you demonstrate with some codes as you have the file?

The electrodes can easily be selected as they are segmented with label 20. And I want to remove them with nearest voxels.

[Rik]

It is not clear to me what indices are tissue and which are electrodes. Do you mean all electrodes are marked with label 20?

[banikr]

>> unique(lab)
ans =
     0
     1
     2
     3
     4
     8
    10
    20

Here label 20 is electrodes.

0 is for background and rest are different tissue labels.