Presentation Information
[10a-C213-1]Machine Learning Reveals Polarization-Driven Biomechanical Signatures at Ferroelectric Cell Interfaces
〇ALEXIS BOROWIAK1,2, YOHEI KONO1,2, TAKESHI SHIMI1,2, TAKESHI FUKUMA1,2 (1.Kanazawa Univ., 2.NanoLSI)
Keywords:
AFM biomechanics,Machine Learning,Ferroelectric surface
This study presents a biocompatible, circuit-free approach that utilizes ferroelectric materials to generate localized electric fields, addressing biocompatibility concerns associated with traditional metallic electrodes. Focusing on cellular electromechanics, we cultured HeLa cells with Fluorescent Ubiquitination-based Cycle Indicator (FUCCI) on lithium niobate substrates with distinct polarization domains. Using advanced imaging techniques—Atomic Force Microscopy (AFM) with Quantitative Imaging (QI), Piezoresponse Force Microscopy (PFM), and fluorescence microscopy—combined with machine learning, we analyzed G1 phase cells. Over 150,000 force-distance measurements quantified cellular elasticity across varying polarization states. Data revealed that spontaneous polarization influences the ionic environment, modulating mechanical properties. Various ML models, including supervised classifiers and unsupervised clustering, were employed to uncover correlations between substrate polarization and elasticity metrics. Results indicate ferroelectric polarization governs cell membrane electromechanics, likely via ionic redistribution. These insights streamline the development of ML-integrated biocompatible platforms for cellular mechanobiology and bioelectric research.