Figure 2
Examples of rVERDICT model parametric maps for a representative participant in our cohort: age late 60’s, PSA 5.21 left posterior lesion Gleason 3 + 4 maximum cancer core length (MCCL) 14 mm, left anterior lesion Gleason 3 + 3. A flowchart of the rVERDICT fitting using a fully connected deep neural network (DNN) is also shown. The DNN takes voxel-wise signals as input and outputs voxel-wise values of the eight rVERDICT model parameters. It is trained in a supervised fashion, using simulated noisy signals according to the rVERDICT model and the VERDICT MRI acquisition. Exemplar training data are shown in Supplementary Fig. S2.
