Abstract (EN):
The urinary bladder on T2-weighted MR images has a high signal intensity appearance, which can be clearly identified from the neighboring structures. However, due to the complex imaging background, the appearance of the bladder is frequently influenced by noise and partial volume effect. Different algorithms have been proposed to segment the bladder. Nevertheless, these algorithms are at their beginning phases, and considerable improvements are needed to obtain an effective automatic segmentation algorithm. In this paper, the performances of four algorithms are evaluated using a case study, from which the effectiveness and the differences between the algorithms are discussed. Quantitative analysis of the segmentation results are presented to measure the deviations of the segmentation results and reflect the aspects that need to be improved. © 2012 Taylor & Francis Group.
Language:
English
Type (Professor's evaluation):
Scientific