Deep learning reconstruction for liver T2-weighted and diffusion-weighted imaging：Improvement of image quality and lesion delineation

In this study, deep learning reconstruction (DLR) of liver fast spin echo T2-weighted (FSE-T2WI) and diffusion-weighted (DWI) was performed. The results showed the liver SNR, and lesion CNR were dramatically increased for both the FSE-T2WI and DWI with DLR compared with conventional reconstruction. The image quality of DLR FSE-T2WI even surpassed that of the routinely used PROPELLER T2WI. DLR didn’t impact the quantitative apparent coefficient derived from DWI. T2WI and DWI with DLR also improved the delineation of lesion structure due to improvement of image quality. Our study indicated DLR FSE-T2WI and DWI would be beneficial for liver disease diagnosing.