Skip to content

Research, validation, and publications

Adelpha MRI is an open research platform. This page states what you can honestly write in a methods section today. It does not invent a clinical trial or a phantom benchmark that is not in the repository.

What you can claim from the code

Claim Basis
Desktop observer + operator for low-field MRI Tauri app, three workspaces
Twin state with explicit provenance DTAM QuantitySource on timestamped quantities
Thermal → \(B_0\) model with configurable \(\alpha_T\) dtam models + configs/models.yaml
Optional thermal PINN forecast Artifact in sidecar/dtam/data/models/pinn/ when trained
MRI4ALL sequence/recon path on desktop FastAPI façade + in-process pipeline
MaRCoS ping and bitstream helpers marcos_boot.py, TCP 11111

What you should not claim

  • That Adelpha MRI is a medical device, or that images are for diagnosis.
  • That EMI/RF twin channels are calibrated EMC measurements.
  • That PINN predictions are validated against a published phantom study in this repo — they are an optional estimator with a fallback heating rate.
  • That systemd acq/recon Start/Stop on System Status works on macOS (they stay Unknown).
  • That the FALLBACK sequence catalog is the same as a fully imported SequenceBase environment.

If you run a validation, publish the protocol, raw data, and software commit hash. This documentation will not pre-write your results.

Citing the software

Use CITATION.cff and the Citation and licensing page. Cite MRI4ALL separately when the console or sequences are material to the work. Cite MaRCoS / Red Pitaya when hardware I/O is material.

  • MRI4ALL console overview (upstream): YouTube
  • DTAM in-tree docs (developer): sidecar/dtam/docs/ — twin math, PINN, Twin API
  • Adelpha operator docs: this site

When a peer-reviewed Adelpha MRI paper exists, add the DOI here. Until then, cite the software version from GitHub Releases (for example the installer version shown on Download).