Aim: To determine the optimal image segmentation protocol that minimizes the amount of manual intervention and correction required while extracting 3D model geometries suitable for 3D printing of aortic dissection (AD) using open-source software. Materials & methods: Computed tomography images of two type B AD cases were selected with images segmented using a 3D Slicer to create a hollow model containing the aortic wall and intimal tear. A workflow composed of filters, lumen extraction and outer surface creation was developed. Results & conclusion: The average difference in measurements at 14 anatomical locations between the Standard Tessellation Language file and the computed tomography image for cases 1 and 2 were 0.29 and 0.32 mm, respectively. The workflow for the image segmentation of type B AD was able to produce a high-accuracy 3D-printed model in a short time through open-source software.
A cascade algorithm combined Kohonen feature map with fuzzy C-means applied in MR brain image segmentation