Macrophage image segmentation by Thresholding and subjective surface Method

Seol Ah Park (1) , Tamara Sipka (2) , Zuzana Kriva (1) , Martin Ambroz (1) , Michal Kollar (1) , Balazs Kosa (1) , Mai Nguyen-Chi (3) , Georges Lutfalla (2) , Karol Mikula (4)

1 Department of Mathematics and Descriptive Geometry, Slovak University of Technology in Bratislava, SLOVAKIA
2 DIMNP - Dynamique des interactions membranaires normales et pathologiques
3 LPHI - Laboratory of Pathogen Host Interactions [Montpellier]
4 Department of Mathematics and Descriptive Geometry, Faculty of Civil Engineering

We introduce two level-set method approaches to segmentation of 2D macrophage images. The first one is based on the Otsu thresholding and the second one on the information entropy thresholding, both followed by the classical subjective surface (SUBSURF) method. Results of both methods are compared with the semi-automatic Lagrangian method in which the segmentation curve evolves along the edge of the macrophage and it is controlled by an expert user. We present the comparison of all three methods with respect to the Hausdorff distance of resulting segmentation curves and we compare also their perimeter and enclosed area. We show that accuracy of the automatic SUBSURF method is comparable to the results of the semi-automatic Lagrangian segmentation.

Mots clés: macrophages; thresholding; finite volume methods; level-set methods; segmentation; image processing

Domaines: Mathématiques [math]

Seol Ah Park, Tamara Sipka, Zuzana Kriva, Martin Ambroz, Michal Kollar, et al.. Macrophage image segmentation by Thresholding and subjective surface Method. Tatra Mountains Mathematical Publications, 2020, ⟨10.2478/tmmp-2020-0007⟩. ⟨hal-02390453⟩