Comparison of Data Augmentation Methods for Improving the Classification of Craniosynostosis

  • This project aims to investigate how different data augmentation methods influence the classification of craniosynostosis.

    Concerning the classification of craniosynostosis using 2D distance maps on a CNN, we test the influences of data augmentation using a Statistical Shape Model (SSM) and a using a conditional GAN. Concerning the classification of craniosynostosis on 3D geometric data, we aim at testing a conditional variational auto-encoder with an LDA classifier using the shape parameter vector of the SSM.