Fig. 1: Results of fivefold cross-validation voxel-wise prediction models. | npj Digital Medicine

Fig. 1: Results of fivefold cross-validation voxel-wise prediction models.

From: Novel radiotherapy target definition using AI-driven predictions of glioblastoma recurrence from metabolic and diffusion MRI

Fig. 1: Results of fivefold cross-validation voxel-wise prediction models.

Random forest (RF) models were able to predict contrast-enhancing lesion (CEL) progression (top row) and non-enhancing lesion (NEL) progression (bottom row). Highest area under the curve (AUC) was achieved for predicting early progressors (t < 7 months), especially for CEL recurrence. ROC=Receiver Operator Characteristic.

Back to article page