MODELLING CANCER CELL MIGRATION IN TUMOR-ON-CHIP PLATFORMS AND FUTURE INTEGRATION WITH MACHINE INTELLIGENCE

Abstract

With the help of modern fabrication techniques, advanced biomaterials and cutting-edge data analysis based on machine learning, we developed a next generation tumor-on-chip platform to understand the migratory decisions of breast cancer cells under the influence of biophysical and biochemical cues. Gelatin-based hydrogels with precisely tuned degree of methacrylation was synthesized for 3D cell culture. We demonstrated that this method can independently control matrix elasticity without changing its porosity or density. Furthermore, when subjecting these 3D matrices to competing chemical gradients, the "decision making" process in single cells was captured and the results suggested a strong directional bias in cellular migration. Finally, a generative adversarial network was constructed to effectively extract features from cytoskeletal organization and reconstruct the missing nuclei information, as its positioning has been known to strongly impact cell migration. This link could serve as the key to decipher the differences between normal and diseased cell health.