Region growing within level set framework: 3-D image segmentation

We present a novel level set framework combined with seeded region growing algorithm for the automatic segmentation of complicated structures from volumetric medical images. Level set evolution methods combine global smoothness with the flexibility of topology changes and offer significant advantages over conventional statistical classification while region growing algori-thms provide pretty fast classification inside the target regions. The driving application is the segmentation of 3-D human cerebrovascular structures from magnetic resonance angiography (MRA), which is known to be a very challenging segmentation problem due to the complexity of vessels geometry and intensity patterns. The results demonstrate the potential of our approach. This framework should also be suitable for other 3-D image segmentation that the region of interest to be segmented has a relatively large size in width, height or both.