Convolutional Neural Network Can Recognize Drug Resistance of Single Cancer Cells

Area of Science:

• Oncology
• Bioinformatics
• Artificial Intelligence in Medicine

Background:

• Circulating tumor cells (CTCs) are crucial for cancer diagnosis and prognosis.
• Current methods for analyzing CTCs and predicting drug response are challenging.
• Deep learning, specifically Convolutional Neural Networks (CNNs), shows potential in medical image analysis.

Purpose of the Study:

• To investigate if an AI system can classify anticancer drug sensitivity based on single-cell morphology.
• To develop and validate a CNN model for predicting drug efficacy at the single-cell level.

Main Methods:

• Construction of a CNN model utilizing the VGG16 architecture.
• Training the AI model on cell morphology data during drug treatment.
• Evaluating the model's accuracy in predicting drug effects on individual cells.

Main Results:

• The developed AI system achieved approximately 0.80 accuracy in identifying the effects of antitumor drugs.
• The model demonstrated the capability to classify drug sensitivity at the single-cell level.
• Cell morphology changes were successfully correlated with drug response.

Conclusions:

• AI systems, particularly CNNs, can effectively analyze single-cell morphology to predict anticancer drug sensitivity.
• This technology holds promise for advancing precision medicine by enabling personalized drug selection.
• Future applications may involve extracting CTCs for AI-driven classification of effective antitumor drugs for individual patients.