About NucLex

What NucLex is

NucLex, the Nuclear Medicine Lexicon and Theranostics Ontology, is a publication and a project. As a publication it explains why concepts and the relationships among them matter, develops the philosophical and computational foundations of ontology, examines the case for investing in shared knowledge, and connects those ideas to precision nuclear oncology and future drug discovery. As a project it proposes a SNOMED CT-anchored, standards-governed, API-accessible knowledge infrastructure for nuclear medicine, theranostics, and computational nuclear oncology.

The publication comes first. The present release (W0) is a living prototype: essays, deep monographs, and a lexicon of concept entries, connected so that a reader can begin with an accessible essay, follow an unfamiliar term into a reference article, and inspect a concise concept entry without losing the thread of the original question. The technical platform is a separate development program, described on the roadmap.

The three pillars

The project rests on three connected functions, each discussed in the founding conversation and preserved here as a design direction rather than an implemented capability.

1. A nuclear medicine niche within SNOMED CT. Reuse existing clinical concepts wherever they exist and create only what is missing, in the domain of radiopharmaceuticals, radionuclides, nuclear medicine procedures, and theranostic treatments.
2. A curated registry of preferred external authorities. For chemistry, substances, medications, proteins, cancer concepts, measurements, and radiology terms, the project would point to the source that should be used (PubChem, FDA GSRS, RxNorm, UniProt, NCI Thesaurus, LOINC, RadLex) instead of recreating it.
3. A concept composer. Compound concepts built from existing concepts across sources, carrying provenance and passing review before anything becomes official.

A fourth idea runs through all three: interactions themselves (how molecule A behaves in relation to molecule B) are knowledge objects that a computational model should be able to retrieve with context, units, uncertainty, and provenance. The essay The three pillars of NucLex and the monograph Interactional ontology develop these ideas.

Editorial principles

• Nuclex.com is the source of truth. Other surfaces, if used later, are distribution only.
• Structured like a research publication, explained like a public educator. Every manuscript carries its sources, its review state, its version, and its correction route.
• AI proposes, humans approve, NucLex legitimizes. Machine-drafted text, definitions, or mappings remain proposals until a named person approves them. This website records who reviewed what, and says plainly when no one has yet.
• Honest content states. Every item is labeled as a brief, a draft, source-checked, or reviewed. Visual polish never implies expert review.
• Build the living prototype before bureaucracy. Unfinished in quantity, complete in soul. Stop and reflect before expanding.
• No patient data. Clinical illustrations are synthetic. The site offers no treatment recommendations and represents no prototype model as validated clinical software.

Current limits

This is a private W0 release. The 10 flagship essays and 20 deep monographs exist as a mix of AI-assisted drafts and writing briefs; none has been source-checked or reviewed by a named domain expert. The 50 lexicon entries are editorial drafts with local identifiers (NX-C001 to NX-C050) that do not claim membership in any released terminology. There are no terminology endpoints, no SNOMED CT extension, no atlas integration, and no simulation capability. The updates form records interest only; no newsletter is being sent. The roadmap lists what each later stage would need to demonstrate.

People and affiliation

NucLex was conceived by Babak Saboury, MD, MPH, a nuclear medicine physician and onco-radiologist, and is published by the Institute of Nuclear Medicine (INM), Bethesda, Maryland. Authors and reviewers are named on each page as review occurs; an empty reviewer field means review has not yet happened.