Fig. 6.

An example test image from each of the three brain cortical plate datasets, the corresponding ground truth segmentation, and segmentations produced by models trained on each of the three datasets separately and by a model trained on all three datasets. A model trained on each one of the datasets does not segment the other two datasets accurately. For example, a model trained in CP- younger fetus has large errors on CP- newborn test images. On the other hand, a model trained on all three training datasets accurately segments test images from all three datasets.