ChromAgeNet Detects Aging Signs in Blood Stem Cell Images

Researchers at IDIBELL, the Barcelona Supercomputing Center, and ISGlobal developed ChromAgeNet, an AI tool that identifies aging patterns in blood stem cell images, IDIBELL detailed in a research announcement. The study was published in Aging Cell.

The researchers trained a convolutional neural network on three dimensional images of mouse blood stem cell nuclei stained with DAPI. ChromAgeNet distinguished young cells from aged cells with a 77 percent probability of correct classification, outperforming a machine learning model based on previously defined chromatin features.

The analysis identified chromatin entropy, heterochromatin at the edge of the nucleus, and certain chromatin condensates as predictors of age associated state. The team also tested the model on aged cells treated with epigenetic drugs. The results detected changes associated with a younger state but did not show that the treatments restored cell function.

The researchers released ChromAgeNet and the image dataset for further study. Its use of a low cost stain and a model with relatively few parameters could support high throughput screening of compounds for their effects on blood stem cell aging.