Assuming that your MRI images are in a 3D stacked volume, you can accomplish what you want by using interp3. The rows and columns of a slice will stay the same when choosing the sampling points, but the temporal or Z direction will simply double in size. So something like this, assuming that MRI is your volume:
[rows,cols,slices] = size(MRI);
[X,Y,Z] = meshgrid(1:cols, 1:rows, 1:slices);
[X2,Y2,Z2] = meshgrid(1:cols, 1:rows, 0.5:0.5:slices);
out = interp3(X, Y, Z, MRI, X2, Y2, Z2, 'linear', 0);
The above will generate a volume that has twice as many slices, keeping the rows and columns the same and using bilinear interpolation. The extra 0 ensures that if we are creating values that are outside of the original sampling points, we will extrapolate these points to 0.
If your images are not a 3D volume, you'll need to place this into a 3D matrix. Assuming that they're called MRI1 up to MRI9, you can do:
MRI = cat(3, MRI1, MRI2, MRI3, MRI4, MRI5, MRI6, MRI7, MRI8, MRI9);
You can then use the above code. Once you're finished, you can grab the intermediate slices by doing:
final_slices = MRI(:,:,1:2:end);
You can then access each intermediate slice with final_slices.
As a quick example seeing this working, let's assume that our volume is a bunch of random numbers in a 3 x 3 x 3 volume:
rng(123123);
MRI = rand(3,3,3)
MRI(:,:,1) =
0.3002 0.8302 0.1768
0.9946 0.7214 0.0678
0.2901 0.4627 0.5201
MRI(:,:,2) =
0.2323 0.8516 0.7838
0.3251 0.5326 0.6377
0.7220 0.4735 0.0717
MRI(:,:,3) =
0.3202 0.1259 0.3360
0.1004 0.9260 0.6287
0.6922 0.3191 0.9011
Running the above interpolation code, we get:
out(:,:,1) =
0 0 0
0 0 0
0 0 0
out(:,:,2) =
0.3002 0.8302 0.1768
0.9946 0.7214 0.0678
0.2901 0.4627 0.5201
out(:,:,3) =
0.2662 0.8409 0.4803
0.6598 0.6270 0.3527
0.5060 0.4681 0.2959
out(:,:,4) =
0.2323 0.8516 0.7838
0.3251 0.5326 0.6377
0.7220 0.4735 0.0717
out(:,:,5) =
0.2763 0.4887 0.5599
0.2127 0.7293 0.6332
0.7071 0.3963 0.4864
out(:,:,6) =
0.3202 0.1259 0.3360
0.1004 0.9260 0.6287
0.6922 0.3191 0.9011
As you can see, the code certainly does create intermediate slices correctly. You see that every even position is one of the original MRI images, while the odd positions are the interpolated results. The first slice doesn't mean anything as we are trying to extrapolate from outside the known volume. You probably want to concentrate on the third slice and its odd positions after this point up until the end of the new volume.
