A new algorithm for segmenting a multimodal grey-scale image is proposed. The image is described as a sample of a joint Gibbs random field of region labels and grey levels. To initialize the model, a mixed multimodal empirical grey-level distribution is approximated with linear combinations of Gaussians, one combination per region. Bayesian decisions involving expectation maximization and genetic optimization techniques are used to sequentially estimate and refine parameters of the model, including the number of Gaussians for each region. The final estimates are more accurate than with conventional normal mixture models and result in more adequate region borders in the image. Experiments show that the proposed technique segments complex multimodal medical images of different types more accurately than several other known algorithms.