ISO/TR 21900:2018
(Main)Guidance for uncertainty analysis regarding the application of ISO/TS 10974
Guidance for uncertainty analysis regarding the application of ISO/TS 10974
This document provides guidance for some methods that could be used to evaluate the sources of uncertainty. It is important to note that there are many legitimate methods for analyzing the overall uncertainty and that the methods in this document are illustrative only.
Lignes directrices pour l’analyse de l’incertitude concernant l’application de l’ISO/TS 10974
General Information
Relations
Standards Content (Sample)
TECHNICAL ISO/TR
REPORT 21900
First edition
2018-09
Guidance for uncertainty analysis
regarding the application of ISO/TS
10974
Lignes directrices pour l’analyse de l’incertitude concernant
l’application de l’ISO/TS 10974
Reference number
ISO/TR 21900:2018(E)
©
ISO 2018
---------------------- Page: 1 ----------------------
ISO/TR 21900:2018(E)
COPYRIGHT PROTECTED DOCUMENT
© ISO 2018
All rights reserved. Unless otherwise specified, or required in the context of its implementation, no part of this publication may
be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting
on the internet or an intranet, without prior written permission. Permission can be requested from either ISO at the address
below or ISO’s member body in the country of the requester.
ISO copyright office
CP 401 • Ch. de Blandonnet 8
CH-1214 Vernier, Geneva
Phone: +41 22 749 01 11
Fax: +41 22 749 09 47
Email: copyright@iso.org
Website: www.iso.org
Published in Switzerland
ii © ISO 2018 – All rights reserved
---------------------- Page: 2 ----------------------
ISO/TR 21900:2018(E)
Contents Page
Foreword .iv
Introduction .v
1 Scope . 1
2 Normative references . 1
3 Terms and definitions . 1
4 Uncertainty background . 1
4.1 General . 1
4.2 Method 1 Evaluation . 2
4.3 Method 2 Evaluation . 3
5 Experiment Uncertainty (u ) . 3
exp
5.1 General . 3
5.2 Measurement tool uncertainty (probe) . 3
5.3 Probe position uncertainty . 3
5.4 Tissue simulating phantom . 4
5.5 RF field source . 4
5.6 Phantom position uncertainty . 4
5.7 AIMD influence . 4
5.8 Overall u consideration . 4
exp
6 AIMD model uncertainty (u ) . 5
Predict
6.1 Piecewise excitation method for deriving AIMD model . 5
6.2 AIMD model uncertainty (u ): Method 1 . 6
Predict
6.3 AIMD model uncertainty (u ): Method 2 .11
Predict
7 Clinical uncertainty (u ) .12
Clinical
8 Uncertainty in conversion to power (u ) .13
power
9 Final uncertainty (combination and application) .13
Bibliography .14
© ISO 2018 – All rights reserved iii
---------------------- Page: 3 ----------------------
ISO/TR 21900:2018(E)
Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards
bodies (ISO member bodies). The work of preparing International Standards is normally carried out
through ISO technical committees. Each member body interested in a subject for which a technical
committee has been established has the right to be represented on that committee. International
organizations, governmental and non-governmental, in liaison with ISO, also take part in the work.
ISO collaborates closely with the International Electrotechnical Commission (IEC) on all matters of
electrotechnical standardization.
The procedures used to develop this document and those intended for its further maintenance are
described in the ISO/IEC Directives, Part 1. In particular, the different approval criteria needed for the
different types of ISO documents should be noted. This document was drafted in accordance with the
editorial rules of the ISO/IEC Directives, Part 2 (see www .iso .org/directives).
Attention is drawn to the possibility that some of the elements of this document may be the subject of
patent rights. ISO shall not be held responsible for identifying any or all such patent rights. Details of
any patent rights identified during the development of the document will be in the Introduction and/or
on the ISO list of patent declarations received (see www .iso .org/patents).
Any trade name used in this document is information given for the convenience of users and does not
constitute an endorsement.
For an explanation of the voluntary nature of standards, the meaning of ISO specific terms and
expressions related to conformity assessment, as well as information about ISO's adherence to the
World Trade Organization (WTO) principles in the Technical Barriers to Trade (TBT) see www .iso
.org/iso/foreword .html.
This document was prepared by Technical Committee ISO/TC 150, Implants for surgery, Subcommittee
SC 6, Active implants.
Any feedback or questions on this document should be directed to the user’s national standards body. A
complete listing of these bodies can be found at www .iso .org/members .html.
iv © ISO 2018 – All rights reserved
---------------------- Page: 4 ----------------------
ISO/TR 21900:2018(E)
Introduction
Clause 8 of ISO/TS 10974:2018 describes methods (Tiers) for analyzing the RF power deposition for
active implantable medical device (AIMD). EM evaluations in a complex near-field exposure scenario
can be difficult and involve many uncertainty sources. Simulations requiring a model of the DUT and
clinical incident field have uncertainties that need to be carefully assessed.
The objective of the uncertainty analysis is to determine the confidence interval of the RF-induced
power deposition with respect to its true value. The acceptable level of uncertainty for an AIMD model
is relative to the safety margin afforded by the AIMD’s RF performance. For instance, if the expected
MRI RF induced AIMD power deposition in vivo is very low, it is less critical to have a highly accurate
model and more uncertainty can be tolerated in the model predictions.
© ISO 2018 – All rights reserved v
---------------------- Page: 5 ----------------------
TECHNICAL REPORT ISO/TR 21900:2018(E)
Guidance for uncertainty analysis regarding the
application of ISO/TS 10974
1 Scope
This document provides guidance for some methods that could be used to evaluate the sources of
uncertainty. It is important to note that there are many legitimate methods for analyzing the overall
uncertainty and that the methods in this document are illustrative only.
2 Normative references
The following documents are referred to in the text in such a way that some or all of their content
constitutes requirements of this document. For dated references, only the edition cited applies. For
undated references, the latest edition of the referenced document (including any amendments) applies.
ISO/TS 10974:2018, Assessment of the safety of magnetic resonance imaging for patients with an active
implantable medical device
3 Terms and definitions
For the purposes of this document, the terms and definitions given in ISO/TS 10974 apply.
ISO and IEC maintain terminological databases for use in standardization at the following addresses:
— ISO Online browsing platform: available at https: //www .iso .org/obp
— IEC Electropedia: available at http: //www .electropedia .org/
4 Uncertainty background
4.1 General
The uncertainties are divided into random and systematic uncertainties.
Random errors result in measured values being distributed about the mean value. Measurement
variations are often well approximated by normal or lognormal distributions. Many of the sources
n
of uncertainty for the measurements described in this document are the result of exponential or r
functions, e.g., the decay of power levels as a function of distance from the AIMD, and therefore can be
approximated by lognormal distributions.
In addition to random errors, systematic errors should also be considered. Systematic error is the
error remaining once the random error is removed as shown in Figure 1. Systematic errors should be
eliminated wherever possible.
© ISO 2018 – All rights reserved 1
---------------------- Page: 6 ----------------------
ISO/TR 21900:2018(E)
Key
X range of values Y range of occurrence
1 mean value 4 random error
2 individual value 5 systematic error
3 true value 6 distribution of values
Figure 1 — Relationship of measured, mean, and true values and association of random and
systematic errors
Uncertainty assessments of systems such as these can be dominated in magnitude by a small subset of
uncertainty sources. When independent uncertainty sources are combined smaller uncertainty sources
often contribute negligibly to the overall budget.
A variety of factors contribute to the uncertainty described in Clause 8 of ISO/TS 10974:2018. The
dominant sources of uncertainty are specific to the equipment, measurement methods, and numerical
simulation tools used for the assessment. Clause 8 of ISO/TS 10974:2018 requires an uncertainty
assessment for the measurement system (u ) and AIMD model (u ). There are two additional
exp Predict
sources of uncertainty being clinical uncertainty (from 8.6 of ISO/TS 10974:2018) and power to
temperature uncertainty (8.4.3 of ISO/TS 10974:2018). Techniques for evaluating these uncertainty
terms (u , u , u , u ) are described. As Clause 8 of ISO/TS 10974:2018 has multiple tiers
exp Predict Clinical Power
for evaluation of power deposition, the evaluation of each uncertainty source is specified per tier.
Two methods of uncertainty evaluations are developed in this document. In both methods, the
uncertainty of the entire assembly is determined. In one method, many of the components of the
assembly are grouped and a single uncertainty determination is made for many of them. In the second,
the sources of uncertainty in a system are identified and individually evaluated a priori and the
dominant sources are combined to obtain the system uncertainty.
[1]
GUM has provided approaches for evaluating the uncertainty of assemblies, regardless of component
count, and called their approaches Type A and Type B. Either or both Type A and Type B evaluations for
each method is appropriate.
4.2 Method 1 Evaluation
Method 1 determines the uncertainty of a complex measurement system by considering the variability
of the system as a whole. Method 1 is based on the assumption that a probability distribution of the
random variation of the evaluation results can be deduced from approximation of the measurement or
modelling system where the uncertainty is determined for an assembly or collection of many parts of
the system. In this approach, multiple elements of the system are assembled or ‘lumped’ together and
2 © ISO 2018 – All rights reserved
---------------------- Page: 7 ----------------------
ISO/TR 21900:2018(E)
their combined uncertainty is assessed. Estimates of the standard deviation of this distribution are
obtained by repeated evaluations and statistical analysis of the obtained values.
4.3 Method 2 Evaluation
Method 2 generally dissects the assembly into its constituent parts, determines the uncertainty of each
individually, and then determines the uncertainty of the group by combining the uncertainty of each
of the components. Method 2 is based on reasonably assumed probability distributions that account
for the available information about the quantities concerned, and the standard deviation of these
distributions. This type of evaluation is performed by evaluating the independent sources of uncertainty
of the components of the measurement or modelling system. In this approach, the components of the
system are separated, and the uncertainty of each component is determined. In a subsequent step, the
uncertainty of each is combined together. Techniques for handling the types of distribution of these
uncertainties and normalizing to a standard distribution from non-standard distributions (such as
[1]
rectangular, triangular, and U shaped) are well known . Root sum square (RSS) is a common method
for combining individual uncertainty components, however the method assumes that the terms are
independent of each other.
In practice, some level of overlap between methods 1 and 2 is likely to exist in uncertainty evaluations.
5 Experiment Uncertainty (u )
exp
5.1 General
The measurement system of Clause 8 of ISO/TS 10974:2018 is comprised of the RF field source, tissue
simulating phantom, and probes for measuring temperature rise, SAR or E-field. It also comprises DUT
fixturing to enable accurate positioning of the probe relative to the DUT. The parameter u accounts
exp
for the combined uncertainty of the RF field source, tissue simulating phantom, the measurement
probe, and the positioning of the measurement probe relative to the DUT.
5.2 Measurement tool uncertainty (probe)
For Clause 8 of ISO/TS 10974:2018, measurements of RF hotspots are done using SAR or temperature
probes.
For SAR measurements, absolute measurements are necessary and the absolute accuracy is a
contributor to the overall uncertainty. The absolute SAR uncertainty is determined from the calibration
of the SAR probe. Depending on its use, the probes linearity, isotropy, distortion of the field, and noise
level could influence its uncertainty.
For temperature probes, all temperature rise measurements are relative. Temperature probe placement
accuracy will likely have a greater impact on measurement uncertainty than the repeatability of the
temperature probe due to the spatial distribution of temperature.
5.3 Probe position uncertainty
Temperature or SAR decreases exponentially as a function of distance from the source of the RF
hotspot. Therefore, probe positioning is an important contributor to the overall uncertainty.
Assuming a 10 °C peak temperature, a 1 °C temperature change can be observed in less than 250 µm.
If ∆T measurements are accurate to within 0,1 °C, this is equivalent to better than 25 µm in probe
placement accuracy. Therefore, temperature probe placement accuracy is one of the dominant sources
of uncertainty and should be evaluated. When SAR probes are used, probe placement accuracy is
equally important.
© ISO 2018 – All rights reserved 3
---------------------- Page: 8 ----------------------
ISO/TR 21900:2018(E)
5.4 Tissue simulating phantom
In order to minimize differences between measurements and model predictions, it is desirable
to control the conductivity of the TSM (tissue simulating medium) in addition to the background
temperature. Background temperature control is necessary because conductivity and permittivity
have large temperature coefficients as described in ISO/TS 10974:2018 Annex H.1.2. Permittivity is
relatively insensitive to slight variation in mixing method and similar to conductivity, the variation
during experiment can be controlled by limiting the bulk TSM temperature rise. TSM with various
conductivity values within the tolerance range as specified in 8.3.2 of ISO/TS 10974:2018 can be used
to evaluate the weighting coefficient for its contribution to the uncertainty in the temperature or SAR
measurement.
5.5 RF field source
In order to minimize differences between measurements and model predictions, it is important
to know the E along the lead pathway matches the exposure assumed for prediction. It is typical
tan
to consider and control the RF incident fields along the lead pathway as well as the contribution of
phantom position uncertainty to the RF Field uncertainty. Uncertainty of the RF incident fields along
the lead pathway is more salient than field error at points in the phantom not near the lead pathways,
and hence more useful to control and quantify.
Small changes in the lead pathway can produce a significant change in the measured ∆T, particularly
for lead pathways that are producing significant phase cancelation or phase enhancement. The AIMD
mounting fixture(s) and probe measurement fixture(s) need to be carefully designed in order to
minimize the positional variation of the lead over the entire pathway. Small changes in the position of
any section of the lead can change the magnitude or phase of the incident E and result in a change in
tan
the measured ΔT or SAR.
The fixturing of the lead pathway on the RF field might distort an idealized simulation of this RF field
and lead to uncertainty in the RF field source. Incident field distortion caused by fixturing could be a
source of uncertainty, unless accounted for in the simulated incident fields from which E along the
tan
lead pathways are derived.
5.6 Phantom position uncertainty
Uncertainty in the phantom position can also cause differences between the measured and simulated
E exposures leading to differences in the simulated and actual E along the AIMD. This effect should
tan tan
be closely controlled or quantified as it can be a contributor to the overall uncertainty.
5.7 AIMD influence
The electric field induced in numerical human body models that do not contain AIMDs are used to
define RF heating test environment specified in Tiers 2, and 3 of Clause 8. These methods assume that
the perturbation of the induced electric field due to the AIMD can be neglected or is accounted for in the
RF heating model validation. Care should be taken to ensure that these assumptions are valid for the
AIMD being evaluated (particularly when multiple parts of the AIMD are in close proximity).
5.8 Overall u consideration
exp
The above description of a typical power deposition measurement system identified a number of
contributors to the overall uncertainty.
In using a Method 2 analysis, the uncertainty of each of the above terms is determined. A specific
experiment isolating each of these variables is evaluated, as much as reasonably possible. Then each of
these individual contributions are combined using a RSS method to create an overall u .
exp
In using a Method 1 analysis, an experiment or series of experiments are devised that combines
the equipment or measurement components that are contributors to the overall uncertainty. The
measurements from repeated experiments of these assemblies could use the probes (temperature
4 © ISO 2018 – All rights reserved
---------------------- Page: 9 ----------------------
ISO/TR 21900:2018(E)
and/or incident E field), over the range of measurement probe placements, and over the range of TSM
parameters, and the range of applied RF field are used to calculate a measurement variation, (i.e.
temperature measurement variation/uncertainty).
6 AIMD model uncertainty (u )
Predict
6.1 Piecewise excitation method for deriving AIMD model
Clause 8 of ISO/TS 10974:2018 does not describe a specific method for determination of the AIMD
model, but contains Formula (1) for power deposition, P , predicted by the AIMD model.
hotspot
2
L
PA= Sz Ez dz (1)
() ()
hotspothotspot tan
∫
0
where E (z) is the incident tangential E field along the length of the lead. The AIMD model consists of
tan
A, the calibration factor and S (z), the transfer function along an AIMD of length L.
hotspot
[5]
The piecewise excitation method is one among several methods that can be used to derive S (z)
hotspot
for each AIMD hotspot. The piecewise excitation method involves measurement of the induced scattered
complex E-field at the AIMD hotspot under test as the steady state response for each piecewise unit step
E applied at successive discrete locations along the length of the lead at the frequency of interest.
tan
A typical piecewise excitation system consists of a source of localized E that provides constant
tan
amplitude, constant phase piecewise excitation, an E-field sensor and test and measurement
instrumentation such as RF source, oscilloscope, network analyzer, etc. The piecewise excitation system
might be configured in a number of ways depending on the type of RF transmitter, sensor and test and
measurement instrumentation. A schematic of one such configuration that uses a dipole transmitter, a
coaxial monopole sensor and a vector network analyzer (VNA) is shown in Figure 2.
Methods used to measure and compute the calibration factor, A, that enables transformation of the
relative total induced E-field amplitude from S (z) to an estimate of the power deposited for
hotspot
E (z) are described in 8.4.3 and 8.4.4 of ISO/TS 10974:2018.
tan
The E (z) do not include any disturbance caused by the presence of the AIMD. This disturbance might
tan
depend on the proximity of one segment of the AIMD to the other and the construction of the AIMD.
Caution should therefore be taken when using pathways with close AIMD segments such as the phase
reversal (see ISO/TS 10974:2018, Annex M).
Either Method 1 or Method 2 may be used to determine uncertainty associated with the AIMD model,
regardless of the technique for determining the AIMD model itself.
© ISO 2018 – All rights reserved 5
---------------------- Page: 10 ----------------------
ISO/TR 21900:2018(E)
Key
1 RF source 6 coaxial antenna
2 port 1 of VNA 7 helical tip electrode
3 port 2 of VNA 8 length of the AIMD lead
4 dipole antenna 9 tissue simulating phantom
5 localized E (z) 10 AIMD device case
tan
NOTE The test and measurement equipment (VNA) (1) consists of a RF source (2) (port 1 of VNA) and a
sense input (3) (port 2 of VNA). The RF source is connected to a transmitting antenna [e.g., dipole antenna, (4)]
that sets up a localized E (z) (5) whose profile envelope can be approximated by a square. The scattered E field
tan
sense probe [e.g., coaxial antenna, (6)] is placed close to the AIMD hotspot of interest [e.g., helical tip electrode,
(7)] so that the scattered E-field response to the piecewise incident E is recorded as a S21 or voltage measure
tan
as a function of the location of the incident E along the length of the AIMD (8). This is the transfer function,
tan
S (z). The AIMD, RF transmitting antenna and the sense probe are all located within a tissue simulating
hotspot
phantom (9).
Figure 2 — A typical measurement setup used to obtain S (z) with the piecewise
hotspot
excitation method
6.2 AIMD model uncertainty (u ): Method 1
Predict
An equivalent electromagnetic model of the AIMD might have error in its predictions even when the
incident tangential electric field along the AIMD is completely and accurately known. This subclause
describes a method of determining the prediction error distribution and assigning a prediction
uncertainty, u , to the AIMD model output assuming perfectly known incident field model input.
Predict
When using the AIMD model, its output is an estimate of the correct response to the input RF exposure
along the AIMD, E (l). The uncertainty of the output can be calculated based on differences between
tan
model predicted, P , and corresponding measured, P , RF power deposition when the AIMD is
model meas
exposed to a set of known radiated test conditions. The uncertainty depends on the RF power deposition
measurement noise and on the choice of radiated test conditions among other factors.
The purpose of RF power deposition measurements is to estimate the correct response of the AIMD to
a variety of known radiated test exposure conditions. Experimental controls should be incorporated to
minimize measurement error as well as any differences between the assumed model input RF exposure
along the AIMD and the actual applied radiated exposure during measurement. Errors in the RF field
exposure, the surrounding tissue simulating media dielectric properties and the power deposition
measurement (typically dominated by AIMD and measurement probe placement uncertainty) can
6 © ISO 2018 – All rights reserved
---------------------- Page: 11 ----------------------
ISO/TR 21900:2018(E)
be quantified, minimized or otherwise controlled by making reference measurements of known or
independent standards (e.g. performing a gauge repeatability and reproducibility study). Random
measurement error contribution to the model prediction can be reduced by using an average or median
value of multiple measured power deposition observations at each radiated test exposure. Radiated
test exposure experimental designs which incorporate controls and replicate measurements as well as
randomization of any remaining sources of error minimize the error contribution to u .
Predict
An example set of 25 AIMD radiated exposure predictions and corresponding power measurements are
illustrated in Figure 3, with data values provided in Table 1. Each data point in the figure represents the
mean value of replicate measurements, avg{P }i, on the y-axis and the corresponding AIMD model
meas
prediction, P ,i, on the x-axis for each AIMD exposure E (l)i tested. One method to quantify the
model tan
uncertainty of the model prediction is to perform a linear regression between the measured dependent
variable avg{P } and the model predicted RF power independent variable P given by avg{
meas model
P }i = β0 + β1 P ,i + ϵi where β0 and β1 are unknown constants that must be estimated and ϵi
meas model
is a normally distributed random residual error for the AIMD exposure with zero mean and standard
deviation of σ. For the case where no power is anticipated, the intercept β0 should be zero and the
linear regression equation reduces to avg{P }i = β1 P ,i+ϵi. In the case in which βo is not within
meas model
the measurement uncertainty, the source of this systematic error should be examined. An example
zero-intercept regression line is shown in Figure 3 with a β1 of 1,020 1 and statistical coefficient of
2 2
determination (R ) of 0,938 2. The β1 and R values close to 1 demonstrate a deterministic and accurate
regression relationship between model prediction and measurement.
© ISO 2018 – All rights reserved 7
---------------------- Page: 12 ----------------------
ISO/TR 21900:2018(E)
Key
X P for each ideal exposure
model
Y avg{P } for each actual exposure
measured
a
Upper 90 % prediction interval bound.
b
Upper 90 % tolerance interval bound.
c
Lower 90 % prediction interval bound.
d
Lower 90 % tolerance interval bound.
e
Outlier.
f
95 % confident that there is 90 % chance the correct AIMD prediction lies between the Upper and Lower
Tolerance Interval bounds.
Figure 3 — Linear regression for example set of 25 AIMD radiated exposure predictions and
corresponding power meas
...
Questions, Comments and Discussion
Ask us and Technical Secretary will try to provide an answer. You can facilitate discussion about the standard in here.