Texture-based segmentation of the knee bones in MRI using phase information

This paper considers the problem of tissue classification in 3D MRI by the mean of texture analysis. The main contribution is the use of the phase of the complex MR image, to derive texture features. Phase can give information about tissue interfaces and provide a good discrimination between the bone and the surrounding tissues but is usually not used because of the phase unwrapping problem. We present a method to extract textural information from the phase that does not require phase unwrapping. The textural information extracted from the magnitude and the phase can be combined to improve tissue classification. The aim of this classification is to initialise an active shape model, leading to a more precise segmentation