Overview - ISO/TR 21900:2018 (Guidance for uncertainty analysis)
ISO/TR 21900:2018 provides guidance on evaluating uncertainty when applying ISO/TS 10974 to assess MRI safety of active implantable medical devices (AIMDs). The report describes illustrative methods to quantify the confidence interval of RF‑induced power deposition predictions and measurements. It emphasizes that multiple legitimate approaches exist and that the examples are guidance only, not prescriptive requirements.
Key topics and technical requirements
- Scope of uncertainties: Divides uncertainty into random and systematic components and stresses eliminating systematic error where possible.
- Uncertainty terms: Covers principal uncertainty contributions required by ISO/TS 10974 - experiment uncertainty (u_exp), AIMD model uncertainty (u_predict), clinical uncertainty (u_clinical) and power‑conversion uncertainty (u_power) - and how to combine them into a final uncertainty budget.
- Two evaluation methods:
- Method 1 (lumped approach) - evaluate the assembled measurement/simulation system as a whole by repeated tests to estimate standard deviation.
- Method 2 (component approach) - dissect the system into constituent sources, evaluate each uncertainty separately, then combine (e.g., root‑sum‑square, RSS) using GUM Type A/B approaches.
- Experimental elements addressed:
- Measurement probe uncertainty (SAR and temperature probes): calibration, linearity, isotropy, noise.
- Probe position uncertainty: highly sensitive due to steep spatial gradients (example: a 1 °C change may correspond to