3-Dimensional medical image compression: A first approach to the application of the ADCT-ISO

In this paper, an application of the ADCT-ISO method using a three-dimensional trans- form is presented. In this first approach, a block size 8 × 8 × 8 was chosen. The multidimensional DCT was developed from the 1D DCT fast algorithm. Quantized coefficient values are obtained using a linear transformed quantization step dependent on the spectral position of the coefficient. These steps are given by the 3D quantization matrix which has been derived empirically from the 2D matrix. Information on the value and position of the coefficients are transmitted performing entropy coding (Huffman´s Code). Performance was evaluated on a MRI and CT 3D isotropic image data base. Five compression ratios were defined : 0.20, 0.30, 0.40, 0.50, and 0.80 bits/pixel. The preliminary image fidelity results (SNRmse) obtained suggest that it is possible to obtain higher compression ratios in 3D than in 2D.