Quantitative measurement methods must be precise and accurate to justify their clinical use. The accuracy reflects the difference in measurement groups compared to another, often expressed in compliance percentages, standard measurement errors, coefficients of variation or Bland-Altman diagram. We propose a variance component analysis (VCA) to assess the impact of errors due to certain elements of a pet scan (scanner, time, observer, etc.) to express the composite uncertainty of repeated measurements and obtain relevant repeatability coefficients (RCs) presenting a unique relationship with Bland-Altman plots. Here we present this approach to evaluating intra- and inter-observational variations with PET/CT, illustrated by data from two clinical trials. Zaki R, Bulgiba A, Ismail R, Ismail NA. Statistical methods of measuring compliance with medical devices measuring variables continuously in comparative studies of methods: a systematic verification. PLoS ONE. 2012;7 (5):e37908. Repeatability in relation to reproducibility: repeatability establishes the proximity of the concordance between the measurements in the same condition, i.e. with the same laboratory, with the same observer and with the same equipment (scanner PET, image reconstruction software), at close intervals. Reproducibility is about the proximity of the agreement between actions under all possible conditions for identical themes, i.e. using different laboratories, observers or PET scanners or assessing daily variation.
Hoge, R. D. (1985). The validity of direct observational measures of student class behaviour. Review of Educational Research, 55, 469-483. Shrout, P.E., Spitzer, R. L., Fleiss, J. L. (1987). Comment: Quantification of compliance in the resumed psychiatric diagnosis. Archives of General Psychiatry, 44, 172-178.
In Study 1, we found a RC of 2.46 that corresponds to half the width of the Bland-Altman boundaries of the agreement. In Study 2, the RC for the same conditions (same scan, patient, time and observer) was 2392; For different scanners, the RC rose to 2543. The differences between the observers were negligible compared to the differences due to other factors; 1 and 2: 10 USD (95% CI: 352 to 332) and observers 1 to 3:28 (IC 95%: 313 to 370 USD). In addition to factors such as the observer, time and scanner, the quantification of the FDG PET is itself done by technical calibration (for example. B between the PET scanner and the dose calibrator, the paraveeur administration of FDG PET), biological (p.B.