top of page

Artificial Intelligence for Label-Free Cellular Characterization

Example_prediction.jpg

Our lab harnesses AI to advance microscopy-based cell analysis, with a focus on non-invasive, quantitative characterization of live cell cultures. We develop machine learning models, including conditional GANs, to extract molecular and phenotypic information directly from transmitted light images such as brightfield and phase contrast microscopy.

A primary application is deciphering the intra-population heterogeneity of MSCs, a key challenge in regenerative medicine and cell therapy. Our AI-powered platform quantifies extracellular marker expression and localization in real time, enabling robust, high-throughput, single-cell, and multi-marker analysis, without the need for labeling or destructive assays. This approach streamlines drug screening and quality control for cell therapies, offering a rapid, cost-effective, and scalable solution.

Recent publications:

Epithelial Mechanobiology and Collective Behavior

Screen Shot 2022-09-20 at 11.26.56 AM.png

Epithelial cells form the body’s essential barriers, coordinating solute transport, secretion, protection, and sensory functions. Their collective behavior underlies normal physiology and the development of many diseases, including most human cancers. Yet, traditional studies focusing on isolated cells or small colonies fail to capture the emergent, tissue-level properties of fully confluent epithelia

Our lab addresses this gap by systematically dissecting how cell-cell adhesion and cell-substrate interactions jointly regulate epithelial phenotype and function in fully confluent monolayers. We employ engineered substrates with tunable stiffness and microfabricated patterns to decouple the effects of matrix mechanics from intercellular adhesion. We also combine biophysical approaches with live-cell imaging, genetic manipulations, and quantitative image analysis to interrogate the dynamic organization, mechanotransduction, and phenotypic plasticity of epithelial sheets.

Recent publications:

Next-Generation 3D Tissue Models and Single-Cell Analysis through Engineering

lncap_fig1_edited.jpg

The urgent need for physiologically relevant tissue models and scalable single-cell analysis tools is driving innovation in biomedical engineering, drug discovery, and disease modeling. Our lab addresses this need by developing advanced 3D tissue engineering and instrumentation platforms that more accurately recapitulate human tissue and tumor microenvironments. By integrating spheroids and organoids from stem cells or patient samples, we model in vivo-like cell-cell and cell-matrix interactions. To overcome barriers in tissue maturation and throughput, we engineered programmable Microwell Flow Devices (MFDs) that deliver controlled laminar shear, improving tissue complexity and enabling robust quantitative drug response assessment. Further, our recent work introduces rapid-prototyping of large-area acoustic microfluidic devices via laser manufacturing for sub-wavelength, programmable manipulation of cells and particles. We also pioneered scalable, high-density single-cell acoustic trapping and selective release, supporting parallel isolation and downstream analysis of thousands of individual cells with high viability. Collectively, these technologies are transforming the landscape of 3D tissue modeling, high-content screening, and single-cell analysis.

Recent publications:

  • Twitter Classic
  • Facebook Classic
bottom of page