Using deep learning for predicting the dynamic evolution of breast cancer migration

Area of Science:

• Oncology
• Cell Biology
• Bioinformatics

Background:

• Breast cancer (BC) metastasis drives mortality, necessitating better understanding of cell migration.
• Traditional 2D wound healing assays offer limited insights due to manual labor and data scarcity.
• Improved cell migration analysis is crucial for developing effective breast cancer therapeutics.

Purpose of the Study:

• To introduce an innovative approach for analyzing and predicting cell migration in breast cancer.
• To overcome the scalability limitations of traditional wound healing assays.
• To enhance the understanding of metastatic processes in breast cancer.

Main Methods:

• Development of the Prediction Wound Progression Framework (PWPF) using Deep Learning (DL) and artificial data generation.
• Initial training of the DL model on simulated wound healing data for MCF-7 breast cancer cells (monolayers and spheres).
• Fine-tuning the DL model using real-world wound healing data.

Main Results:

• The PWPF effectively analyzes and predicts cell migration dynamics in wound healing models.
• Enhanced usability of 2D cell migration models through AI-driven analysis.
• Significant contribution to understanding breast cancer cell migration and wound healing mechanisms.

Conclusions:

• Advancements in automated cell migration analysis enable more scalable and comprehensive future studies.
• The developed dataset, models, and code are publicly available to facilitate further research.
• The PWPF offers a powerful tool for investigating cell migration in various biological contexts.