Compression of ultrasonic RF data

A compression algorithm is presented, which utilizes the special properties of ultrasonic radio frequency (RF) data. The compression is done in two steps: First, linear predictive coding (LPC) is applied, using an one-step-predictor. Further, the remaining error of the prediction is stored using only the necessary word length to store the signal. A lossy extension of the algorithm is presented, which stores only the upper bits of the error signal. The algorithm has been tested with both, data of a speckle phantom and in vivo data. The data could be compressed to approximately 30-55% of the original data size using the lossless algorithm. In comparison, a conventional compression tool achieves 65-75 % of the original data size.