About the MICRON Research Hubs

What these pages are

These pages group a selection of the research literature that incorporate images and/or data from MICRON imaging systems. The papers are grouped by the species studied and the condition under investigation. Each hub is one species and one condition. Every figure on a hub page is computed from the full cohort of papers on that page; on large cohorts the table at the foot lists only the most recent papers, and says so.

The corpus, and how a paper enters it

The hubs are built from reSEARCH, the Phoenix-Micron database of published, preclinical, in vivo, small animal research. reSEARCH holds more than 1,000 peer-reviewed MICRON papers, and 10's of thousands of additional papers. Papers are found through the public literature indexes, such as PubMed and Europe PMC, with bibliographic records completed from Crossref and OpenAlex. A paper joins the MICRON set when its text names a MICRON product together with the company, by a deterministic rule applied to the stored text. Our corpus does not claim to be an exhaustive index of the field or to incorporate all papers that incorporate MICRON data.

The hubs cover 1,107 distinct papers across 111 hubs. A paper appears on every hub whose species and condition it studied, so the hub tables together hold more rows than there are papers.

What is recorded for each paper

Four things are recorded per paper, each from the paper's own text: the species studied, the conditions under investigation, the imaging and experimental techniques used, and the MICRON products cited. Each is a controlled name from a fixed vocabulary, so "Diabetes Mellitus, Type 1" on one hub is the same term on every other.

How a species' role is decided

For each paper and each species on it, the role of that species is recorded: a disease model of the hub's condition, a model of a related condition, healthy animals used for pharmacokinetics, delivery or tolerability, or excised tissue. A disease-model role counts toward a hub only when the paper itself names that condition; a role assigned without that evidence is shown as a related-condition model instead. Where a condition is induced rather than inherited, such as choroidal neovascularization produced by laser, the link between the paper and the condition is a relation record that quotes the passage of the paper that states it, and that passage is checked word for word against the stored text on every build.

What the counts mean, and do not mean

The number in each table is a count of papers. Technique counts are paper-level co-occurrences and do not necessarily indicate that a technique was applied to the focal species or focal disease model. On hubs with more than a few dozen papers, the table at the foot lists the 20 most recent papers in the cohort, and the page says so; the counts above it still cover the whole cohort. Our corpus does not claim to be an exhaustive index of the field.

How the pages are built and checked

Every hub page is generated from frozen data: a fact sheet per hub, produced from reSEARCH at a tagged content release, and a library of reviewed statements. Nothing on a page is typed by hand. Before a page is written, software checks that every count on it equals the fact sheet, that every paper in the table is on the sheet, and that every quoted evidence passage matches the stored text word for word. Before the site is published, a second set of checks runs on the built pages: the text on the page equals the reviewed text, every number in the generated blocks equals the number in the page's own tables, and every link resolves. A page that fails a check is not published.

How the statements were reviewed

The statements on these pages were drafted with AI assistance, reviewed by a second AI model acting as scientific reviewer, verified by software against each cohort's own papers, and approved by Phoenix-Micron before publication. Publication is reversible: a correction is a rebuild of the page from the corrected source, never an edit in place, and every page shows the date it was built and the content release it was built from.

Reporting an error

If a paper is misplaced, a count looks wrong, or a statement does not match a paper, write to micronhelp@phoenixmicron.com with the page address and the paper's DOI or reSEARCH link. Every paper in a hub's table opens in reSEARCH without an account.