Abstract: To compress a digitized medical image, a first wavelet including quadrants respectively having LL, LH, HL and HH frequencies is provided, preferably in parallel. L and H respectively represent low and high frequency ranges. L and H representations may be respectively provided by averaging, and taking the difference between, two (2) successive pixels values. The first and second letters in each quadrant respectively refer to the frequency range in a pair of co-ordinates. A second wavelet may include, in the first wavelet LL quadrant, quadrants respectively having LL, LH, HL and HH frequency ranges. Successive wavelets may be similarly provided. Each wavelet is provided from the LL quadrant of the previous wavelet because most of the frequency information resides in the LL quadrant. Individual compressions may be provided in parallel of the representations in the LH, HL and HH quadrants in the first wavelet, preferably on a lossy basis (e.g. discrete cosine transform and then Huffman coding).