2D fourier transform filter

I am trying to remove low frequency bands from an image after transforming it using 2D fast fourier. I can identify individual components of the transformed image, but cannot yet work out how to filter only components higher than a certain frequency threshold and then reconstruct the original image using only those frequencies.

Say, for example, I have the following 5 x 5 image matrix, x:

1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5
1 2 3 4 5

I subtract out the mean and compute the FFT:
x = x - mean(mean(x)); X = fft(x);

But then how would I filter only frequencies of X above a a threshold, t, and reconstruct the image using only those frequencies to give Y?

Thanks for your help!