3D reconstruction from CT-scan volume dataset application to kidney modeling

Organ segmentation and reconstruction are useful for many clinical purpose, like diagnostic aid or therapy planification. The segmentation literature is large. However, it is rare to find a method which performs both organ segmentation and reconstruction. Moreover, the major part of the literature focuses on the liver or lungs. We present a new method for kidney reconstruction from 3D CT scan. First, we perform a segmentation stage to extract the kidney volume from the greyscale image stack. Then, we refine this segmentation by analyzing the histogram of the kidney regions previously segmented. The refinement step eliminates the areas that was incorrectly considered as kidney region. A point cloud is extracted and is then reconstructed using the Poisson surface reconstruction. Thus, we obtain a 3D kidney model on which we apply a post-treatment in order to regularize it. Despite the reconstruction step, our method is fast and can be used real-time medical environment.