Skip to main content
. 2024 Apr 4;30:100574. doi: 10.1016/j.phro.2024.100574

Fig. 1.

Fig. 1

Graphical depiction of the workflow where the distortion corrected DWI images (1) are first fitted to the NG-IVIM model (2), which yields four parameters maps 3. Next, supervoxels are created based on the four parameter maps 4. Then all supervoxels from all patients are clustered (5) to gain insight in the prevalence of certain combination of parameters (identifying certain phenotypes) (6) in different tumors.