The use of soft computing techniques for medical image analysis is reviewed in this paper. These techniques involve application of fuzzy logic, neural network and evolutionary algorithms. We introduce the different imaging modalities and motivation for using soft computing techniques for medical image analysis. We review different soft computing based approaches for image processing and segmentation, which can improve the diagnostic value of the images. We also present some results with MRI images. Based on these approaches we suggest an architectural framework for soft computing based application development. © 2003 Taylor & Francis Group, LLC.