A2: Whole-body molecular mapping

Current spatial transcriptomics methods fall into two architectural classes, and the constraints of each are tied to their methodology rather than being incidental engineering issues.

Barcoded-surface methods (10x Visium, Stereo-seq, Slide-seq) work by depositing tissue onto a flat substrate carrying position-encoded molecular barcodes. RNA from each cell binds the local barcodes, gets sequenced, and is reconstructed back to its position. This requires physically flat, thin sections — typically 5-20 micrometres — and freezing or fixation to immobilise the RNA. The methodology is intrinsically two-dimensional and destructive.

In situ hybridisation methods (MERFISH, seqFISH+, MERSCOPE) work by labelling RNA molecules with sequential rounds of fluorescent probes inside the tissue itself. These methods preserve 3D structure better but still require optical access — typically through cleared, fixed tissue — and the sequential imaging takes hours per sample, during which the tissue cannot be living.

Neither architecture has a path to living-tissue, whole-body operation. Scaling either to AIHS specifications would require an entirely different approach: either deep-tissue molecular detection at single-cell resolution (which existing physics doesn't support) or distributed engineered reporters (a different technology with its own constraints — see A1).

Multi-target MRI and PET are improving but face a fundamental multiplexing limit set by the underlying physics of each modality.

PET typically allows one tracer per session because the detector measures 511 keV annihilation photons regardless of which positron-emitting isotope produced them — the detector cannot distinguish photons from different isotopes when they emit at the same energy. Sequential PET with different isotopes allows ~3-5 targets serially, with each session requiring its own tracer dose, scan time, and decay window before the next.

MRI with chemical exchange saturation transfer (CEST) contrast can detect a handful of metabolites simultaneously by tuning saturation frequencies to specific exchangeable proton environments. Hyperspectral MRI with multiple hyperpolarised agents (carbon-13 pyruvate, fluorine-19 perfluorocarbons, xenon-129) extends the count slightly. Multi-spectral imaging with engineered nanoparticle libraries pushes the limit further still.

The realistic ceiling for these approaches is in the dozens of simultaneous targets — two orders of magnitude short of AIHS specifications. They are useful as components of an AIHS diagnostic suite but cannot serve as the primary mapping modality. The multiplexing physics is unforgiving regardless of detector sensitivity improvements.

A composite system combining several partial capabilities, each achievable on a 15-year horizon:

Whole-body imaging at organ-and-tissue resolution using current modalities plus near-future improvements — total-body PET (recently demonstrated by the EXPLORER system at UC Davis with ~40× the sensitivity of clinical PET), 7T MRI for high-resolution structural imaging, and emerging modalities like photoacoustic tomography for vascular and metabolic detail.

Single-cell-resolution sampling from accessible compartments. Blood, urine, breath, and ingested-capsule samples carry signatures of tissues throughout the body. Cell-free DNA methylation patterns identify the tissue of origin to within centimetres (this is the basis of the Galleri test, and the methodology is improving rapidly). Circulating tumour cells, exosomes, and cell-free RNA add complementary information.

Engineered sentinel cells deployed strategically (per A5) reporting on local conditions to externally-readable signals — fluorescent reporter cells in the gut, immune cells engineered to release detectable molecules on encountering specific targets, and so on.

Combined with sophisticated computational models of how signals from internal tissues reach observable compartments, this composite approach captures roughly 70-80% of the AIHS diagnostic value without requiring full whole-body single-cell mapping. This interim is the realistic target for Phase III deployment. The full A2 specification remains a Phase IV target conditional on breakthrough advances.