ASTM E2677 Limit of Detection Web Portal
Supporting Information
ASTM subcommittee E54.01 has developed a Standard Test Method for the
determination of Limit of Detection (LOD) in trace explosive detectors.
The Method was developed following ISO-IUPAC guidelines that harmonize
concepts of detection limits1 and considers the observed
behaviors of response signals in a wide range of trace detectors. Here,
the LOD90 is defined as the lowest amount of a particular
substance for which there is 90% confidence that a single measurement will
have a true detection probability of at least 90% while the true
non-detection probability of a realistic process blank is at least 90%. The
LOD90 value is therefore a directly useful metric of trace
detector performance and reliability, since the value reflects the practical
detection capability of the detector system, influenced by the inherent
sensitivity, selectivity, and response variability of the system under
realistic deployment conditions.
To obtain the LOD90 value, you enter 0.10 for the confidence
limit in the web form. Although 0.10 is the default, you can request
a different confidence limit. If you specify a confidence limit of
0.05, this will generate the LOD95 value rather than the
LOD90 value. Typical values for the confidence are 0.10, 0.05,
and 0.01 corresponding to LOD90, LOD95 and LOD99,
respectively. For convenience, the documentation will simply refer to
LOD90.
The ASTM Standard Method was developed by a task group at NIST.2
The specific statistical method for LOD determination has been
published.3 In order to facilitate the calculations involved
for users of the Method, the task group has developed a web-based
LOD calculator hosted on a NIST cloud server. The LOD application utilizes
the NIST Dataplot program4 and was coded into HTML by Alan Heckert
and Keith Kwiatek.
The web-based calculator has been developed to provide the best estimate
of the LOD90 value for a particular analyte given the quality
of the inputted data. Ideally, data should include a large number of
process blank replicates as well as detector responses from a large number
of replicates from two mass levels closely straddling the actual
LOD90 mass level. By this strategy, the Method is insensitive to
many pitfalls that are encountered in commercial trace detectors, including
detector response saturation, truncated response distributions, and response
heteroscedasticity, i.e. changes in response variation with signal level.
Practicalities dictate, however, that a limited number of replicates be
analyzed and that mass levels be selected that are wide enough apart to
guarantee the straddling of the unknown LOD90 value. By setting the
minimum number of required replicates to ten, and by utilizing a short
sequence of mass levels that increase by a factor of three, data for the
calculation of an adequate LOD90 estimate may be obtained. The
web-based calculator has been tested and validated with real and simulated
data possessing several types of error structure.
To see a description of the minimum data requirements for the LOD
calculator, please see the LOD data input
page.
For questions regarding the ASTM Standard Method, please contact michael.verkouteren@nist.gov.
References
- Currie, L.A. (1999) Analytica Chimica Acta 391, 103-134.
- Verkouteren, R.M. et al. NIST Special Publication
(in progress).
- Rukhin, A.L.; Samarov, D.V. (2011) Chemometrics and Intelligent
Laboratory Systems 105, 188-194.
- Filliben, J.J.; Heckert, N.A.
http://www.itl.nist.gov/div898/software/dataplot/