Machine learning assisted microfluidics dual fluorescence flow cytometry for detecting bladder tumor cells based on...

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Cancer Diagnostics

Background:

• Bladder cancer (BC) is a leading cause of cancer death worldwide.
• Current urinary cytology for BC detection relies on manual microscopy, limiting sensitivity and efficiency.
• Automated cell differentiation is crucial for improving early BC diagnosis.

Purpose of the Study:

• To develop an automated, high-throughput platform for urinary cytology analysis.
• To differentiate bladder cancer cells from normal urothelial cells using machine learning.
• To enhance the sensitivity and efficiency of early bladder cancer detection.

Main Methods:

• Development of a machine learning-empowered dual-fluorescence flow cytometry platform (μ-FCM).
• Definition of a characteristic parameter (CP) for cell and nucleus size differentiation.
• Optimization of a support vector machine (SVM) algorithm for data analysis and false positive rate (FPR) compensation.

Main Results:

• The μ-FCM platform achieved high-throughput differentiation of BC cells from urothelial cells.
• SVM algorithm provided 84.7% accuracy, 91.0% specificity, and 75.0% sensitivity for cell differentiation.
• False positive rate compensation allowed accurate detection of rare BC cells with 0.4% error.

Conclusions:

• The developed μ-FCM system demonstrates proof-of-concept for identifying exfoliated cells in urine.
• Integration of AI and microfluidics offers a scalable, efficient, and automated solution for biosensing and in vitro diagnosis of BC.
• This technology advances automated cell analysis for improved bladder cancer diagnostics.