K-means clustering applied to a biomedical image of Escherichia coli.

This study evaluates two image-processing methods—K-means clustering and Otsu thresholding—for identifying and measuring Escherichia coli colony surface area in biomedical images. Accurate estimation of colony size is critical for linking bacterial growth models with experimental data, thereby supporting simulation studies and bioreactor design. Both algorithms were applied to digital colony images, and their outputs were compared using visual inspection and statistical analysis. Otsu thresholding consistently underestimated colony areas by 13–19% relative to actual measurements. In contrast, K-means clustering produced measurements closely aligned with true values, demonstrating higher accuracy and reliability for colony detection, counting, and surface quantification. When colonies from both methods were paired. K-means produced significantly larger areas, the P-value factor is much less than 0.05 which confirms there is significant meaning in the obtained results. These results confirm the potential of K-means clustering as a robust and practical tool for microbial image analysis in research and biotechnological applications