Journal: Frontiers in Neurology
Article Title: Feasibility of Multimodal MRI-Based Deep Learning Prediction of High Amino Acid Uptake Regions and Survival in Patients With Glioblastoma
doi: 10.3389/fneur.2019.01305
Figure Lengend Snippet: Results of the receiver operating characteristic analysis to predict 6-month progression-free survival by imaging (MRI, PET, and MRI-learned PET from U-Net 4 ) and clinical variables.
Article Snippet: To investigate the effect of different multimodal MRI protocols on the performance of the U-Net system, three different U-Net systems (U-Net 1(Siemens) trained by multimodal data of Siemens Protocol, U-Net 2(Philips) trained by multimodal data of Philips Protocol, and U-Net 3 trained by combined multimodal [Siemens+Philips] Protocol) were separately implemented using Google TensorFlow library ( www.tensorflow.org ).
Techniques: Imaging