Imaging snakes automate medical-image segmentation

Many objects can be modeled by rigid primitives, either as contours, surfaces, or volumetric primitives. But these representations are not appropriate for natural objects, such as parts of the body. In such applications, active or deformable models can be used to represent either a class of objects of differing shape or objects that may change shape.

May 1st, 1997

Imaging snakes automate medical-image segmentation

Many objects can be modeled by rigid primitives, either as contours, surfaces, or volumetric primitives. But these representations are not appropriate for natural objects, such as parts of the body. In such applications, active or deformable models can be used to represent either a class of objects of differing shape or objects that may change shape.

Imaging snakes are energy-minimizing splines that deform themselves from a given starting point to conform with image contours. When provided with an initial starting point, snakes are useful in medical image segmentation where regions of images need to be automatically defined.

Barney Baldwin at New York University (New York, NY) has used snake techniques to develop an image search procedure to find anatomic structures in medical images. A set of images is first used to train the system to recognize shape information and image features. Once image features are extracted, the mean and variance of the image features are computed. These data can be used to perform snake-based image segmentation on similar medical images. For information contact Baldwin at baldwin@cs.nyu.edu.

More in Security, Surveillance, Transportation