Development of machine learning software to quantitatively map telomere induced senescence in tissue sections during aging

Mayo Clinic Rochester, Rochester MN

Investigators

Linked publications, trials & patents

Abstract

Project Abstract Cellular senescence is a key driver of many age-related diseases, yet its in vivo characterization remains challenging due to cellular heterogeneity and the lack of universal biomarkers. Deep learning classifiers based on morphological features hold promise for senescence detection, but existing models are unable to accurately identify individual senescent cells in complex tissue environments. As part of a SenNet-funded Technology development award (TDA), we have developed SenoQuant, an advanced AI-based analysis tool for multimodal imaging datasets. Using SenoQuant, we integrated single-cell spatial proteomics technologies (4i and CODEX) to analyze senescence-associated markers in human skin. From these data, we trained a deep learning classifier capable of identifying p21+ senescent cells at single-cell resolution using only DAPI staining, achieving 87% accuracy. Unlike conventional classifiers trained on in vitro datasets, our model is the first to leverage spatial omics data to capture the complexity of senescent cells in situ. Building on this success, we aim to enhance the accuracy and expand the scope of our classifier to detect a broader range of senescent cell types (senotypes). Utilizing omics datasets from aged human skin and lung, including SenNet’s benchmarking project comparing spatial omics technologies within the same tissue, we will develop more robust models. Additionally, we will integrate our classifier into SenoQuant, providing researchers an easy-to-use, no-code required solution to analyze DAPI-stained images and automatically quantify senescent cell burden. These advancements will have profound implications for both fundamental research and translational applications. By enabling cost-effective, high-throughput senescent cell detection, our work will accelerate drug discovery, biomarker development, and diagnostics, ultimately advancing our understanding of aging and age-related diseases.