Zaineb Amor

Bone segmentation and extrapolation in Cone-Beam Computed Tomography

Abstract

This work was done within the french R&D center of GE Medical Systems and focused on two main tasks: skull bone segmentation on 3D CBCT data and skull volumetric shape extrapolation on 3D CBCT data using deep learning approaches. The motivation behind the first task is that it would allow in- terventional radiologists to visualize only the vessels directly without adding workflow to their procedures and exposing the patients to extra radiations. The motivation behind the second task is that it would help understand and even- tually correct some artifacts related to partial volumes. The skull segmenta- tion labels were prepared while taking into account imaging-modality related considerations and anatomy-related considerations. The archiecture that was chosen for the segmentation task was chosen after experimenting with three different networks, the hyperparameters were also optimized. The second task explored the feasability of extrapolating the volumetric shape of the skull out- side of the field of view with limited data. At first, a simple convolutional autoencoder architecture was explored, then, adversarial training was added. Adversarial training did not improve the performances considerably.