Program system NEOCORTEX
Project summary
The aim of this project is to develop a software system that supports visualization, segmentation and 3D-reconstruction of comptuer tomography and electron microscopy data. This data represent a sequence of rectangular layers of the same size. Each layer consists of voxels with certain values of the density and may be perceived as a grayscale image. The software system developing in this project supports multiple methods for visualizing these data: using the 2D- and 3D-texture, as well as ray tracing. Before the reconstruction data are segmented to highlight features of interest. This can be done in two ways: completely automatic (using the k-means method with a given number of segments and range of voxels desity changing) and interactive (in this case user selects the region of interest, then it is localized by the active contour method). For the method of centers of gravity is written as serial and parallel to the GPU implementaton using OpenCL which gives good acceleration. The data can be processed in one of two modes: 2D (one layer) or 3D (all layers at once like a point cloud). The reconstruction algorithm is Ball Pivoting.
Project site
http://code.google.com/p/neocortex
Participants
- Elena Agapova
- Alexandr Reshetnikov
- Alexandr Kochenkov
- Alexandr Muravyov
Publications
-
Reshetnikov A. Parallel implementation of the computer tomograms segmentation by the gravity centers method using OpenCL.
Conference Proceedings of the 11th All-Russian conference «High-performance parallel computing on the cluster systems», P. 258-261, 1st-3rd of November, 2011, N. Novgorod, Russia (in Russian).
