Abstract: Compressed Sensing(CS) is a new signaln acquisition and processing theory.It can decrease the signal sampling time and computation cost by reducing the required data for signal recovery while maintaining good image quality.The CS theory has drawn a lot of attention and made great progress in medical imaging since it was proposed.This paper introduces the history of CS theory and the recent improvement in medical imaging. Moreover,we focus on the dictionary learning algorithm which is a new CS-based adaptive reconstruction algorithm.At last,the result of simulation is presented to convince the algorithm.