Segmentation des tumeurs en imagerie médicale TEP basée sur la marche aléatoire 3D (original) (raw)

This article presents an automatic segmentation algorithm based on the random walk (RW) method. To deal with some drawbacks of the original algorithm such as the dependance to the choice of hyperparameter β, as well as the probability that a random walker going to a label, depending only on the gradient of intensity, we propose an approach allowing to solve these problems. Our approach consist in using an adaptive hyperparameter β and to integrate probability density of labels into the system of linear equations used in the RW. Based on this approach, we have developped a new version of the RW in order to segment the tumor in medical imaging using Positron Emission To-mography (PET). The results we obtained on a physical phantom and patients' data show that our method is better than the original algorithm and a new method proposed in the literature .