Back

Cellular deconvolution of the brain with topological magnetic resonance image analysis

Vazquez, L. A.; Fromandi, M. B.; Hagemann, T. L.; Risgaard, R. D.; Gonzalez, J. M. G.; Singh, A. P.; Frautschi, P.; Hurley, S. A.; Sousa, A. M. M.; Dean, D. C.; Ulland, T. K.; Yu, J.-P. J.

2025-12-13 neuroscience
10.64898/2025.12.10.693426 bioRxiv
Show abstract

Magnetic resonance imaging (MRI) is foundational tool in neuroscience, enabling characterization of neuroanatomical markers of disease, behavior, and cognition. However, the precise cellular processes driving the structural and functional readouts provided by MRI remain opaque. Non-invasively assessing cell type, abundance, and location using MRI has the potential to revolutionize both basic science and clinical practice. To this end, we developed SpaTial Representation and Analysis using Topological Architecture (STRATA), an image-based gradient-boosted machine learning framework, which quantifies cell type proportions of neurons, astrocytes, oligodendrocytes, and microglia from MR images. Here we demonstrate and validate STRATA on diverse disease models, species, and regions of interest that together highlight the generalizability of the STRATA framework.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.