AI-driven tracking of hallmarks of aging mechanisms during chronological aging

Abstract

Cellular aging, a complex phenomenon marked by the gradual degradation of essential cellular components, poses significant challenges in understanding its temporal orchestration and underlying mechanisms. The ”Hallmarks of Aging” framework outlines pivotal processes contributing to cellular aging, yet their dynamic emergence and interconnectedness remain elusive. This research adopts an integrative approach, leveraging phase contrast microscopy, yeast genetics, and deep learning. In the first phase, phase contrast images are processed using a UNet segmentation model, enabling the identification of yeast cell contours. Utilizing Transfer Learning with ImageNet, a model is trained to classify cells at distinct aging stages. The second phase integrates fluorescent imaging and a regression model to correlate morphometric changes with hallmark indicators. This interdisciplinary methodology not only unveils the intricate dynamics of cellular aging but also establishes a foundation for predicting bioactivity based on morphological features. The approach presents a novel dimension to aging research, holding promise for innovative interventions in cellular rejuvenation and advancing our comprehension of the aging process.