The interdisciplinary team of Avinash Yaganapu (Computer Science), Amarnath Singam (Mechanical Engineering), Sai Phani Parsa (Computer Science), Calvinique Lee (Biomedical Engineering), Jennifer Zheng (Life Sciences), Sunjeet Saha (Mechanical Engineering), Seungman Park (Mechanical Engineering), and Mingon Kang (Computer Science) in collaboration with Jeong Hee Kim at Iowa State University published a original research paper, "Prediction of Oxidative Stress- and Anticancer Drug-Induced Cellular Senescence Using Holotomography and Interpretable Attention-Based Contrastive Learning," in Small Methods (IF: 8.7).
This paper introduces an AI-driven, label-free diagnostic framework that combines 3D holotomography (quantitative phase imaging) with an interpretable attention-based contrastive learning model to detect and predict cellular senescence. By capturing refractive index distributions within single cells, holotomography reveals biophysical and morphological changes, induced by oxidative stress or anticancer drugs. To interpret these complex morphologies without invasive staining or fluorescent tags, the authors implemented an attention-guided contrastive learning model that automatically highlights the specific sub-cellular regions driving the predictions. This approach will eventually offer a robust tool for evaluating therapeutic efficacy, screening anti-cancer drugs, and monitoring cellular aging dynamics in real time, by identifying drug- and stress-induced senescent states.
More broadly, this work demonstrates the power of artificial intelligence to extract meaningful biological insights from high-resolution imaging data, a frequent challenge across computational biology.
