An Automated Bone Removal Technique for Extraction of Cerebral Vessels from Head CT Angiography

Y. Lee, D.-Y. Tsai, S. Inomata, R. Yokoyama, T. Hara, H. Fujita, M. Kanematsu, T. Iwama, and H. Hoshi (Japan)


CT angiography, cerebral vessels, bone removal, Laplacian-like filter, computer-aided diagnosis


We propose a method for automatic removal of bone regions like calvaria and cranial base from head CT angiography (CTA), to extract only cerebral vessels. The method consists of binarization for first extraction of cerebral-vessel regions, Laplacian-like filtering to separate cerebral-vessel regions and bone regions, 2D and 3D- labeling to eliminate bone regions, and modification of extracted cerebral-vessels regions. These processes are repeated using reconstructed CTA images in the axial- sagittal- and coronal-view planes, to further improve the performance of bone removal. To validate the effectiveness of the proposed method, we apply the method to three clinical cases of head CTA. Our preliminary results show that the proposed method has the potential to automatically detect cerebral vessels from head CTA images with an acceptable accuracy.

Important Links:

Go Back