Remove points in a 3D matrix that lie inside a 2D polygon on each layer

I'm visualizing an MRI but am trying to remove a lot of artifacts that usually come in these images. I've implemented two solutions:

1. use knnsearch to remove points too far away from their neighbors (i.e. floating out in the middle of nowhere) - works very well
2. Have a user draw a polygon around remaining points from a topdown view (view 2) to remove the clusters of bad points from each layer.

The issue I'm running into is this: I can have a user create a polygon and get the points of the polygon with something like roi.Position, however I'm having trouble figuring out how to efficiently come through each Z layer of the matrix to remove the points within the polygon. Any ideas?

plot3(target_cluster(:,1,:),target_cluster(:,2,:))
roi = drawpolygon('Color','r');

My instinct is to run a for loop through each layer, but this seems inefficient. TIA

[Matt J]

If the process is going to have an interactive step in each slice, why are you fretting that it be fast?

[Cameron]

It won't be interactive per slice, I'm having the user draw a polygon from a top down view and effectively eliminate a vertical column group from the mri by adjusting the values of all points within that same polygon to 0 on every slice.

Answer 1

[Matt J]

and effectively eliminate a column from the mri by adjusting the values of all points within that same polygon to 0 on every slice.

If so, then once you've drawn the ROI, it would just be,

mriVolume=mriVolume.*(~roi.createMask);

[Cameron]

Wouldn't this just create a 2D matrix of ones matching the polygon, using a deprecated command?

[Matt J]

Is this not what you want? You have an MRI image volume and you want to zero all pixels inside a polygon and repeat that for the same set of pixels in each slice. As you can see below, the result my answer gives is not 2D.

mriVolume=rand(3,3,3) %Fake mri volume

mriVolume = 
mriVolume(:,:,1) =

    0.9117    0.3054    0.8743
    0.6790    0.5029    0.9291
    0.4444    0.7229    0.0910


mriVolume(:,:,2) =

    0.7140    0.6320    0.6661
    0.6696    0.3753    0.0985
    0.1696    0.9075    0.6249


mriVolume(:,:,3) =

    0.4784    0.9112    0.7842
    0.8221    0.8651    0.3624
    0.3684    0.6878    0.9057

imshow(mriVolume(:,:,1))
roi=drawpolygon('Position', [0,0; 2,0; 0,3]*1.7);

mriVolume=mriVolume.*(~roi.createMask) %Delete pixels inside polygon in each slice

mriVolume = 
mriVolume(:,:,1) =

         0         0    0.8743
         0         0    0.9291
         0    0.7229    0.0910


mriVolume(:,:,2) =

         0         0    0.6661
         0         0    0.0985
         0    0.9075    0.6249


mriVolume(:,:,3) =

         0         0    0.7842
         0         0    0.3624
         0    0.6878    0.9057

[Cameron]

Ah I see now, that's exactly what I'm looking for. Thanks!