Coupling anatomical and functional information for the computer-aided delineation of Phase-Contrast MRI images using active contours

Phase-Contrast (PC) velocimetry MRI is a useful modality to explore cardiovascular pathologies, but requires the automatic segmentation of vessels. Most existing segmentation approaches focus on the exploitation of anatomical information to guide algorithms, such as widely used active contours we consider in this paper, although acquisitions simultaneously provide functional information. Due to noisy regions surrounding vessels, it is difficult to integrate this information as a complementary guiding force for attracting contours towards vessel boundaries. As we illustrate, the recently proposed phase quality map is interesting but appears not appropriately formulated to build useful attraction forces. In this paper, a refinement is presented to overcome this issue, together with preliminary experimental results. We detail and illustrate how this refinement operates on phase information, and how the coupling of both anatomical and functional information can improve the efficiency of active contours, compared to the only use of anatomical information.