A team from Guangyuan First People’s Hospital reconstructed conventional MRI images using a three-dimensional super-resolution generative adversarial network. The validation-set AUC for predicting lymphovascular space invasion in early-stage cervical cancer increased from 0.727 to 0.864, with the difference reaching statistical significance by the DeLong test. The study included 117 patients with FIGO stage IB to IIB disease, 21 of whom underwent image quality assessment; the AUC improvement with super-resolution T2-SPAIR was not statistically significant.
Source:BIOENGINEER.ORG · Read Original · 前沿热点
Leave a Reply