It is well known that an object viewed from multiple channels will generally convey more information than a single-channel observation [78-80]. A very successful application is remote sensing. Various sensors are designed to capture signals reflecting from surface of the earth in different bands. Since different objects on the earth have different spectrum profile, more details are usually detected by integrating the multibands information than that viewed with a single band. Similarly, in carotid plaque study, different imaging contrast weightings are often employed to detect the composition with bloodvessel wall , and these multiple contrast weighting (MCW) techniques play a more and more important role in finding the different tissue types in the studied subject and generally can provide a more comprehensive view.
To achieve the goal of image segmentation and also to take advantage of the information with multichannel data, a multidimensional MRF (mMRF)-based solution will be first discussed in this section, which integrates the information from all different channels with a dynamical weighting. However, because of the intolerable amount of computation involved in the optimization process and intrinsic interspectral independency requirements in mMRF model, this technique becomes unsuitable to the requirements of interactive MR image analysis application. As a compromise, a robust cluster based segmentation algorithm is then put forward in our study, which is with faster segmentation speed.
Was this article helpful?