1887

Abstract

Background & Objectives Despite its long track record, segmentation in medical image computing still remains an active field of research, largely due to the complexities of in-vivo anatomical structures. We present a novel segmentation algorithm based on chaotic theory; the preliminary results show the potential of the technique. Methods: Henri Poincare first developed (later revisited by Lorentz) this chaotic model by observing a significant deviation in output through his "three-body-system" when the input is varied even slightly. This theory can be applied to image segmentation ensuring deterministic convergence by keeping initial conditions constant. The scenario is analogous to "iron particles moving randomly in a cell and a strong magnet is suddenly placed on its center". Though the process is chaotic, now it becomes periodic and is easy to solve. The probability of a particle to reach the magnet can be found by solving the magnetostatic version of Laplacian equation. The image is treated as a graph; a seed point is placed both on foreground and background individually and the node potentials are calculated in each case. The label map is built by considering the maximum of two probabilities at a node. Finally, gradient operation on the label map determines the coordinates that carry nonzero value as the desired contour coordinates. Results: The qualitative results shown in Fig. 1 reflect the method accuracy when compared to the ground-truth data. The liver CT datasets are collected from MICCAI 2007 workshop. For a single image, it takes nearly 3 sec to complete the operation on a system with 2GB RAM core2duo processor. Conclusions: This paper has presented a fast and simple graph-based segmentation algorithm based on chaotic theory which is deterministic and time efficient. In future, we intend to explore its behavior on different subjects and modalities as well.

Loading

Article metrics loading...

/content/papers/10.5339/qfarf.2012.BMP108
2012-10-01
2020-11-30
Loading full text...

Full text loading...

http://instance.metastore.ingenta.com/content/papers/10.5339/qfarf.2012.BMP108
Loading
This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error