.png){kind=icon}
.png){kind=icon}
Return
Artificial intelligence-enabled predictive system for Escherichia coli colony counting using patch-based supervised cytometry regression: A technical framework
.png){kind=icon}
PRE
AI
DOI：10.1016/j.microc.2025.113206.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Biologists often rely on counting circular objects, such as cell colonies, to derive valuable insights. However, this critical task remains manual, time-consuming, and prone to human error and subjective interpretation. Additionally, traditional methods, such as cytometers, can introduce inaccuracies that compromise data analysis. To address these challenges, researchers are increasingly turning to automated counting techniques. In response, we have developed an artificial intelligence-enabled predictive system for automated cell counting called the Patch- Based Supervised Cytometry Regression (PB-SCR) system. This system provides high accuracy and ease of use and is designed for universal cell counting applications, including Escherichia coli (E. coli), significantly reducing the workload for researchers. By integrating machine learning algorithms for data collection and analysis, we optimized a USB camera for image capture. We adopted patch learning combined with ensemble approaches, implementing five regression models: ridge regression, gradient boosting regression, support vector regression (SVR), k-nearest neighbor regression, and adaptive boosting regression. Among these models, SVR demonstrated the best performance, achieving a root mean squared error of 0.53, R-square of 0.87, mean squared error of 0.34, and mean absolute error of 0.38. The system achieved an average accuracy of 95.40 % +/- 3.01 %. A state-of-theart comparison between our proposed framework and the OpenCFU solution demonstrated an 8.5 % improvement in performance. Finally, our PB-SCR system shows promising potential for future applications, including cancer cell recognition.
Keywords:
Artificial intelligence (AI)
Automated analysis
Cell counting
Image processing
Patch-Based Supervised Cytometry Regression (PB-SCR)
Support Vector Regression (SVR)
Journal
IF:
5.1
Papers:
1.8W
Citations:
3.7W
View more organizations