Traditional diffusion weighted MR imaging(dMRI)usually uses a single shot acquisition method to suppress motion artifacts,which inevitably has disadvantages.In multishot diffusion imaging,the ghost artifacts from motion induced phase errors are usually corrected in the image domain.However,this correction requires image and navigator signals be geometrically matched.In this study,a SYnergistic iMage reconstruction using PHase variatiOns and seNsitivitY(SYMPHONY)is proposed for multishot diffusion imaging.In this new theory,phase variations are not treated as a source of imaging artifacts,but as a kind of image encoding.Then the k-space correlation of different shots and channels can be used for data interpolation.We also applied this technique in simultaneous multi-slice(SMS)imaging for multi-shot diffusion imaging.Experiments demonstrate that SYMPHONY can effectively suppress artifacts caused by phase variations and provide diffusion images with high resolution and less distortion for both brain and spine.Compared with the traditional image domain method,SYMPHONY is less sensitive to the mismatch between the image and navigator signals.In addition,the new technique can be seamlessly integrated with 2D GRAPPA to form a 3D recon framework for SMS imaging.In conclusion,by using the inherent k-space correlation of different shots and channels,SYMPHONY can robustly reconstruct diffusion images with reduced artifacts from multishot scans.