Medical image compression by applying 3D DWT in PACS

The process of image compression in a practical Picture Archiving and Communication System (PACS) was discussed with detail. To fully reduce the inter-slice correlation existing in the volumetric image sets generated by CT and MR, 3D Discrete Wavelet Transformation (DWT) was introduced in our application. Instead of using a fixed quantizer of lossless low frequency and distinct loss of high frequency, an adaptive quantizer was devised taking MSE as the performance index. In the low frequency subband, DPCM was replaced withS+P transform to facilitate coding computation. Compared with JPEG or 2D DWT, our method is 20%–50% more efficient. Furthermore, preliminary tests showed that 33 dB may be the maximal distortion threshold for CT images. Biography: WANG Si-xian(1944-), male, Associate professor. Research direction: image processing.