Deep learning-enabled imaging flow cytometry for high-speed Cryptosporidium and Giardia detection.

Area of Science:

• Environmental Science
• Biotechnology
• Microbiology

Background:

• Imaging flow cytometry generates vast data, challenging analysis of similar microorganisms.
• Accurate detection of waterborne pathogens like Cryptosporidium and Giardia is crucial for public health.

Purpose of the Study:

• To develop a high-throughput system for rapid and accurate detection of Cryptosporidium and Giardia in drinking water.
• To leverage deep learning for enhanced bioparticle image analysis.

Main Methods:

• Integration of imaging flow cytometry with an efficient artificial neural network (MCellNet).
• Development of a deep learning model for high-speed classification of microorganisms.
• Validation of the system's performance in detecting specific waterborne parasites.

Main Results:

• Achieved a classification accuracy exceeding 99.6% for Cryptosporidium and Giardia.
• Demonstrated high sensitivity (97.37%) and specificity (99.95%) in detection.
• Reached analysis speeds of 346 frames per second, surpassing existing deep learning algorithms.

Conclusions:

• The MCellNet-based system provides a rapid, accurate, and high-throughput solution for bioparticle detection.
• This technology has significant potential for environmental monitoring and biosensing applications.
• The system overcomes data analysis challenges posed by imaging flow cytometry for similar morphologies.