Microtomographic analysis of surgically-created bone defects in rats: manual versus predefined region-of-interest definition

Abstract Micro-computed tomography (micro CT) is a high-resolution radiographic modality that has been widely used in bone-tissue research. This study assessed whether there is an agreement between the manual and predefined region-of-interest (ROI) definition methods during the three-dimensional analysis of bone defects by micro-CT. Bone defects of 2.3 mm in diameter were created in the left tibiae of Wistar rats. After 14 days, the tibiae were collected and scanned. During the analysis by CtAnalyser® software, the ROI was defined in two different ways for each sample: manually (manualROI) and based on a predefined circular shape (circularROI). Data corresponding to bone volume fraction (BV/TV) and trabecular number (Tb.N) were analyzed. No difference was found between the groups for both BV/TV (p = 0.51) and Tb.N (p = 0.38), with an excellent correlation for BV/TV (r = 0.9). A non-significant correlation was observed for Tb.N (p = 0.1). The Bland–Altman analysis showed a low systematic error for BV/TV (1.2%) and Tb.N (0.2/mm). The manualROI was more time-consuming (512.8 ± 60.5 seconds) than the circularROI method (95.4 ± 9.4 seconds). Both methods provided similar results, therefore, predefined formats for ROI definition of regular lesions are indicated, considering their use is less time-consuming.