Cancer Cells Detection in Phase-Contrast Microscopy Images Based on Faster R-CNN

In biology and medicine research, detection and identification of cancer cells plays an essential role to further analysis of cell properties and developing new drugs experiments. However, owing to the adhesion among cells and great changes in morphology, it is a very challenging task to detect and locate the cells accurately, especially for the cells adhesion area. In this work, a deep detector for cells based on the framework of Faster R-CNN is proposed, and based on this, a Circle Scanning Algorithm (CSA) is presented for the redetection of adhesion cells. And then a series of experiments are achieved. The results show that the proposed deep detector can detect and identify all separate individual cells in an image, and that the hybrid method by combining Faster R-CNN with the proposed CSA can effectively detect and identify the adhesion cells under the conditions of the limited samples of adhesion cells.