A python implementation of the model is freely available in GitHub: github.com/fragrussu/bo...

This was a massive team effort made possible by the support of my @vhio.bsky.social colleagues – Raquel Perez-Lopez above all -, of researchers at the VH Hospital and at the Institut de Diagnòstic per la Imatge, and of international collaborators such as Marco Palombo from Cardiff University.

We test the model on a pilot cohort of cancer patients suffering from primary or metastatic cancer in the liver, showing that it detects cancers aggressiveness (by agreeing with Ki67 proliferation immunostains) and that it distinguishes primary cancer types.

Through the analysis of a total MRI-histology correlation score, we find that fitting a model of restricted intra-cellular diffusion (with neglible extra-cellular signal contributions) to high b-value data maximises radiological-histological correlations.

However, while cancers spread frequently to the liver, models tailored for liver application and feasible in the clinic are still sought. In this article, we studied co-localised histological and MRI data in mice and cancer patients to compare a set of candidate biophysical dMRI models.