A structure-aware splitting framework for separating cell clumps in biomedical images

Splitting clumps of convex objects has many practical applications in biomedical and industrial fields. In this paper, a novel framework is proposed for splitting overlapping or touching cells in biomedical images. The method mainly contains two parts: candidate splitting points extraction and structure-aware splitting. In candidate splitting points extraction, we extract concave points and centroids from the input image to characterize the contained cell clumps. Next, in structure-aware splitting, we first use the extracted candidate splitting points to identify the clump structure and then construct the split line by using the corresponding splitting strategy. To further improve the robustness of our splitting results, we propose a post-processing method and add it in our splitting framework. Experiments on three datasets from the Broad Bioimage Benchmark Collection are conducted. The obtained experimental results demonstrate the superior capacity of the proposed approach.