ChromAgeNet uses nuclear images to identify aging blood stem cells
Researchers developed ChromAgeNet, an AI tool that analyzes the three-dimensional organization of chromatin in microscopy images of hematopoietic stem cells to detect aging-related patterns. The model was trained on DAPI-stained images of mouse blood stem cell nuclei using a convolutional neural network. Published in Aging Cell, the work aims to improve measurement of blood stem cell aging and support efforts to preserve or restore their function.
ChromAgeNet was created by teams led by Maria Carolina Florian at IDIBELL/ICREA and Paula Petrone at BSC-CNS/ISGlobal, with Pablo Iañez’s doctoral work central. It examines DAPI-stained mouse hematopoietic stem cell nuclei using a convolutional neural network. In tests, it classified cells as young or aged with 77% accuracy, beating an earlier machine-learning approach built on researcher-defined chromatin traits.
Key predictive cues included chromatin entropy, heterochromatin near the nuclear edge, and specific chromatin condensates. The researchers say the tool could complement epigenetic clocks and help assess blood stem cell aging, potentially aiding efforts to maintain or recover hematopoietic function.
This work may affect older adults and people with blood-production problems, as better measurement of stem cell aging could inform research on preserving or restoring hematopoietic function. Scientists studying aging and regenerative medicine could benefit from a tool that complements existing biomarkers. Its impact remains uncertain until validated beyond mouse cells and translated into clinical or therapeutic settings.