Medical ultrasound image compression using multiresolution decomposition and vector quantization

Medical ultrasound image compression based on multiresolution decomposition and vector quantization (VQ) is presented. The wavelet transform enables one to decompose the ultrasound image into a set of sub-images with different resolution levels. On each level, the image has been split into four components: horizontal, vertical and diagonal detailed, and an approximation component. Statistical analysis shows that the coefficients of vertical components represent the most detailed information for ultrasound image, thus allows the authors to achieve higher compression ratio by applying VQ to the coefficients in vertical components. Experiment results show that ultrasound images can be compressed to more than 72:1 with the performance of PSNR=27.05 dB.