LSGA: combining level-sets and genetic algorithms for segmentation

A novel technique is presented to combine genetic algorithms (GAs) with level-set functions to segment objects with known shapes and variabilities on images. The individuals of the GA, also known as chromosomes consist of a sequence of parameters of a level-set function. Each chromosome represents a unique segmenting contour. An initial population of segmenting contours is generated based on the learned variation of the level-set parameters from training images. Each segmenting contour (an individual) is evaluated for its fitness based on the texture of the region it encloses. The fittest individuals are allowed to propagate to future generations of the GA run using selection, crossover and mutation. The GA thus provides a framework for combining texture and shape features for segmentation. Levelset-based segmentation methods typically perform gradient descent minimization on an energy function to deform a segmenting contour. The computational complexity of computing derivatives increases as the number of terms increases in the energy function. In contrast, here the levelset-based curve evolution/deformation is performed derivative-free using a genetic algorithm. The algorithm has been tested for segmenting thermographic images of hands and for segmenting the prostate in pelvic CT and MRI images. In this paper we describe the former; the latter is described in [11, 12]. The LSGA successfully segments entire hands on images in which hands are only partially visible. At the end of the paper we report experimental evaluation of the performance of LSGA and compare it with algorithms using single features: the Gabor wavelet based textural segmentation method [1, 9], and the level-set based segmentation algorithm of Chan and Vese [6].