A novel approach for automated counting of tumor cell colonies.

BACKGROUND AND

OBJECTIVE

The heterogeneous response of cancer cells to radiation necessitates comprehensive investigations into cell survival, which are often conducted using clonogenic assays. These assays involve laborious and subjective manual enumeration of cell colonies. This paper presents an innovative digital image processing technique for automated cell colony counting.

METHODS

This research focused on HT-29 cells (colon cancer). The methodology uses Hough techniques to achieve precise detection, mathematical morphology, and establishment of the optimal minimum colony size using ROC curve. This approach seamlessly integrated into ImageJ software as a ColCounter plugin.

RESULTS

Images with resolution of 1200*1200 pixels achieved sensitivity and specificity of 92.88% and 92.62%, respectively. To determine the software´s effectiveness of image analysis of different resolutions compared with manual counting, an evaluation of the automated method's reliability was conducted. The findings revealed an overall intraclass correlation coefficient (ICC) of 0.89 (95% CI: 0.81-0.93). The limits of agreement were determined using the Bland-Altman method.

CONCLUSIONS

The proposed methodology demonstrated interchangeability with the conventional manual enumeration technique. The results confirm the software's elevated performance, even in altered contexts. ColCounter offers manual editing function, enhancing its utility as a streamlined, rapid, and precise tool for quantifying tumor cell colonies in clonogenic assays.