Automated Segmentation of Drosophila RNAi Fluorescence Cellular Images Using Graph Cuts

Abstract

Recently, image-based, high throughput genome-wide RNA interference (RNAi) experiments are increasingly carried out to facilitate the understanding of gene functions in intricate biological processes. Effective automated segmentation technique is significant in analysis of RNAi images. Traditional graph cuts based active contours (GCBAC) method is impractical in automated segmentation. Here, we present a modified GCBAC approach to overcome this shortcoming. The whole process is implemented as follows: First, extracted nuclei are used in region-growing algorithm to get the initial contours for segmentation of cytoplasm. Second, constraint factor obtained from rough segmentation is incorporated to improve the performance of segmenting shapes of cytoplasm. Then, control points are searched to correct inaccurate parts of segmentation. Finally, morphological thinning algorithm is implemented to solve the touching problem of clustered cells. Our approach is capable of automatically segmenting clustered cells with low time-consuming. The excellent results verify the effectiveness of the proposed approach.