A Novel Approach Based on Joint Optimization of Alignment and Statistical Surface Representation with Wavelet Transform for CBCT Segmentation Conference Paper uri icon

abstract

  • Cone-beam computed tomography (CBCT) can provide true 3D information of anatomical structures, with advantages of much thinner slice thickness and significantly lowered effective dose of radiation. However, CBCT images are extremely low contract and noisy. It is very difficult to segment thin bones. It usually takes 4-5 hours to manually segment a set of CBCT data. To this end, we developed a novel approach based on the joint optimization of alignment and statistical surface representation with wavelet transform for segmentation of CBCT images. It included two main steps: customized wavelet base initialization (CWBI) and base invariant wavelet active shape model (BIWASM). We validated our approach with others by comparing the surface deviation between segmented shape to the ground truth. The results showed that our approach outperformed the others in accuracy and computing time. 2013 Springer-Verlag.

published proceedings

  • Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

author list (cited authors)

  • Chang, Y., Yuan, P., Kuo, T., Xiong, Z., Gateno, J., Xia, J. J., & Zhou, X.

citation count

  • 0

complete list of authors

  • Chang, Yu-Bing||Yuan, Peng||Kuo, Tai-Hong||Xiong, Zixiang||Gateno, Jaime||Xia, James J||Zhou, Xiaobo

publication date

  • September 2013