Accuracy, precision and inter-rater reliability of micro-CT analysis of false starts on bones. A preliminary validation study.

Micro computed tomography (micro-CT) has already been proposed as a useful technique for the qualitative analysis of false starts (FS) produced on human bones, although the reliability and the error rate of this technique have not been tested yet, neither for qualitative nor for quantitative assessments. The aim of the present study was to test the morphological agreement, accuracy, precision and inter-rater reliability of micro-CT analysis of FS on bones. The morphological agreement was assessed through the degree of concordance among the 3 independent blind raters in the identification of the shape of 24 FS manually produced on bones by 3 different saws (8 FS for each saw). The accuracy was calculated through the percentage of error in the automatic and manual measurement of the diameter of a reference object. The precision was calculated as CV% of multiple measurements performed by 3 independent blind raters on the reference object and one bone sample acquired 20 times. The inter-rater reliability was assessed as intraclass correlation coefficient (ICC) among measurements performed by 3 independent blind raters, assessing 24 FS produced using 3 different saws. The results demonstrated that both qualitative and quantitative analysis were reproducible and robust. Micro-CT analysis showed a 100% morphological agreement, a high level of accuracy (percentage error < 0,5%), precision (CV% < 5%) and inter-rater reliability (ICC > 0.995), when FS were analyzed by forensic pathologists and/or radiologists with adequate expertise. Obviously, further validation studies are needed, including a higher number of samples produced by a wider variety of saws and multiple operators.