AI-based Apoptosis Cell Classification Using Phase-contrast Images of K562 Cells

Area of Science:

• Cell biology
• Biotechnology
• Computational biology

Background:

• Manual cell classification, especially for apoptosis, is time-consuming and prone to human error.
• Automating this process is crucial for efficient research and diagnostics.

Purpose of the Study:

• To develop and evaluate artificial intelligence (AI) models for automated cell classification using phase-contrast microscopy.
• Specifically, to identify apoptosis in K562 cells with high accuracy.

Main Methods:

• K562 cells were induced into apoptosis and stained for DNA fragmentation and caspase activity.
• Phase-contrast and fluorescence microscopy images were acquired.
• Two AI models (Lobe(R) and ResNet50) were trained and validated using five-fold cross-validation.

Main Results:

• Both AI models accurately classified cells into three distinct apoptosis-related groups.
• Classification relied on subtle phase-contrast image variations indicative of apoptosis progression.
• The server-based ResNet50 model showed improved performance with repeated training.

Conclusions:

• AI-assisted phase-contrast microscopy offers a powerful, automated solution for cell classification, particularly for apoptosis research.
• This approach can significantly expedite high-throughput screening for drug development and medical diagnostics.
• It reduces manual workload and enhances classification accuracy, promising advancements in cancer research.