Inter laboratory comparison on Computed Tomography for industrial applications in the slaughterhouses: CIA-CT comparison

An intercomparison on X-ray Computed Tomography (CT) for industrial applications in the slaughterhouses was organized by the Centre for Geometrical Metrology (CGM), Department of Mechanical Engineering, Technical University of Denmark (DTU) and carried out within the project “Centre for Industrial Application of CT scanning - CIA-CT”. In the comparison, 4 laboratories from 4 countries were involved, and CT scanned two synthetic phantoms, which were used instead of real pig carcasses. A phantom consists of several polymer components as Poly methyl methacrylate (PMMA), Polyethylene (PE) and Polyvinyl chloride (PVC). The polymer materials PMMA, PE and PVC represent tissue types as respectively: meat, fat, and bone. The one phantom represents a skinny pig carcass, when the other one represents a fat pig carcass with a higher content of fat (PE). The phantoms were produced through milling and cutting processes. The phantoms circulated among four participants and a total of six clinical CT scanners in Europe. The circulation took place between May 2011 and May 2012. Different volume measurands are considered, encompassing PMMA, PE, and PVC. The results of each participant are kept confidential. Each participant can identify their own results in this report using an anonymous identification number provided by the coordinator. Measuring instructions distributed by the coordinator were followed by all participants without problems. Participants carried out measurements and sent their results to the coordinator. Reference values of both phantoms were measured by Danish Meat Research Institute (DMRI) before the circulation and determined by the coordinator using the principle of water displacement. A stability investigation on the phantoms was performed through 3 reproduced measurements over a 4 month period on a clinical CT scanner under the same conditions at DMRI. Investigations confirmed that the mean variation between the three time periods were quite small, below 30 mL. ANOVA tests demonstrated that the reproduced measurements were not significant (α=0.05), and the materials were stable enough. Depending on phantom and material, reference expanded uncertainties (k=2) ranging from approx. 0 mL up to approx. 10 mL were estimated. The most participants did not have any experience of how to outline uncertainty budgets. The expanded uncertainties stated by the participants are in the range 0-18 mL for both phantoms and all materials. Results by the single participants were compared with the reference values provided by the coordinator through the En value, where |En| < 1 indicates agreement between measurement results while |En| ≥ 1 shows disagreement. Out of a total of 6 single results obtained by the participants using CT scanning, 0% of the measurements yield |En| values less than 1, and 100% larger than 1. Systematic errors were detected for some participants on some of the measured volumes. It could be due to the specified tolerances defined by the participants for segmentation of the polymer materials. It was found that scale error correction particularly should be considered for some participants. The comparison shows that CT scanning on phantoms, generally speaking, is connected with uncertainties in the range 1-1090 mL, as compared to an uncertainty range of 0-10 mL using the principle of water displacement. Each participant can use the comparison results in the report to investigate the presence of systematic errors or an underestimation of uncertainties. Statistics related to the used equipment and procedures show that participants, in general, have followed state of the art procedures for their measurements. The phantoms are suitable artefacts for CT measurements of this kind.