Figure 1
From: Automated neonatal nnU-Net brain MRI extractor trained on a large multi-institutional dataset
Flowchart describing the iterative brain masking process. * Studies were manually corrected. ** Iteration 1 used BET from FSL to generate automated brain masks. *** Iterations 2–4 used nnU-Net models to generate automated brain masks. Studies categorized as “borderline” were manually corrected. The nnU-Net models were subsequently retrained on the “acceptable” and newly corrected “borderline” studies, and new automated masks were regenerated for the “needs revision” studies. **** For iteration 5, all “borderline” and “needs revision” studies were manually corrected.
