We present fully automated deep learning approaches to placental tissue segmentation on our dataset of 68 3D R2* images. Using this dataset, we employ different data schemes to get 4 new datasets consisting of full 3D images, full 2D slices, 3D patches and 2D patches. An unmodified U-Net architecture is trained and tested on these datasets to evaluate the robustness of the model when presented with different data. We find that by artificially increasing the size of the dataset, the model is able to perform better at the segmentation task.
This abstract and the presentation materials are available to members only; a login is required.