An Automatic Multiscale Region Growing Segmentation in Medical Image Based on SLIC and 2D OTSU
Fan Chen, Jianling Gao · 2018
Most of medical image processing is involved in segmentation generally. It had been known that the region growing algorithm can perform the segmentation effectively, but it depends on the position of the initial seed point and the rule of growing primarily. This research develops a scheme of selecting seed points and a multiscale growing, which both are applied to division of pathological tissue slide image. The scheme takes simple linear iterative clustering (SLIC) to extract region of interest (ROI) containing nucleus, which makes selection trapped in these areas. Then, 2D OTSU completely searches pixel sets in these regions as seed points. In this way, nucleus of which grayscale is inhomogeneous can be extracted completely rather than partially. To improve efficiency, seeds growing in multiscale is adopted. Larger scale is priority to be grown, then supplementary to smaller one. Compared to SLIC and 2D OTSU respectively, the experimental result shows effectiveness and superiority of the algorithm.