Remove rows in variable dimension

Is there a way to remove the N last rows of a variable dimension dim in a multi-dimensional matrix. For example, if M is a 4x4x4x4 matrix, N=2,and dim=2 I want the result to be a 4x2x4x4 matrix. If N=1 and dim=4 I want a 4x4x4x3 matrix, and so on. Kind of like an inverse of cat. The problem is that I don't know the number or size of dimensions beforehand, or which dimension to remove from.

My current solution is to use the following switch statement. I have assusmed that I will never work with more than 8 dimensions, and then this works fine. I'm just wondering if there is a better way, as this is (in my opinion) very ugly and not at all modular enough to be satisfying.

switch(dim)
    case 1
        M2 = M(1:(end-N),:,:,:,:,:,:,:);
    case 2
        M2 = M(:,1:(end-N),:,:,:,:,:,:);
    case 3
        M2 = M(:,:,1:(end-N),:,:,:,:,:);
    case 4
        M2 = M(:,:,:,1:(end-N),:,:,:,:);
    case 5
        M2 = M(:,:,:,:,1:(end-N),:,:,:);
    case 6
        M2 = M(:,:,:,:,:,1:(end-N),:,:);
    case 7
        M2 = M(:,:,:,:,:,:,1:(end-N),:);
    case 8
        M2 = M(:,:,:,:,:,:,:,1:(end-N));
end

Answer 1

[Matt J]

idx=repmat({':'},1,ndims(M));

idx{dim}=1:size(M,dim)-N;


M2=M(idx{:})

[Martin Hultman]

Brilliant! I had no idea you could use cell arrays to index like that!

Thanks!

[Matt J]

See also Comma-Separated Lists.