Interactive 3D medical data cutting using closed curve with arbitrary shape

• We mix octree and quad-tree to make cutting operation in real-time.
• 2D mask image and coordinate transformation make users cut 3D volume data in 2D space.
• The cutting method could improve the accuracy of image segmentation algorithms.
• The cardiac images segmented results can be used to evaluate cardiac function quantitatively and be more conducive for diagnosis and treatment of cardiac disease.

Interactive 3D cutting is widely used as a flexible manual segmentation tool to extract medical data on regions of interest. A novel method for clipping 3D medical data is proposed to reveal the interior of volumetric data. The 3D cutting method retains or clips away selected voxels projected inside an arbitrary-shaped closed curve which is clipping geometry constructed by interactive tool to make cutting operation more flexible. Transformation between the world and screen coordinate frames is studied to project voxels of medical data onto the screen frame and avoid computing intersection of clipping geometry and volumetric data in 3D space. For facilitating the decision on whether the voxels should be retained, voxels through coordinate transformation are all projected onto a binary mask image on screen frame which the closed curve is also projected onto to conveniently obtain the voxels of intersection. The paper pays special attention to optimization algorithm of cutting process. The optimization algorithm that mixes octree with quad-tree decomposition is introduced to reduce computation complexity, save computation time, and match real time. The paper presents results obtained from raw and segmented medical volume datasets and the process time of cutting operation.