Investigating microarchitectural characteristics of woven bone exposed to ionizing radiation: comparative analysis of two segmentation methods and their impact on trabecular measurements in an animal study

The study aimed to evaluate bone formation in surgical defects exposed to ionizing radiation (IR) and to compare two image segmentation methods: manual and standardized delimitation of the region-of-interest (ROI). Bone defects were created in the tibias of Wistar rats using a 2.3 mm trephine. After seven days, the animals were exposed to IR (30 Gy single dose). Two weeks later, the animals were euthanized, and their tibias were scanned using X-ray microtomography. ROI definition was performed by a single operator using two methods: (a) interactive segmentation, where the woven bone was manually outlined within the bone defect region, and (b) a standardized, predefined rectangular ROI. Bone volume fraction (BV/TV), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), trabecular number (Tb.N), bone surface (BS/BV), and fractal dimension (FD) were assessed for both segmentation methods. Significant differences were observed for BV/TV (p<0.001), Tb.N (p<0.001), and Tb.Sp (p<0.001), but not for Tb.Th (p=0.71), BS/BV (p=0.79), and FD (p=0.35). The manual method was approximately 3.5 times more time-consuming than the predefined ROI method (446.2±49.5 vs 131.2±38.8 s, p<0.001). None of the evaluated parameters showed significant differences when the IR was considered. Within the limitations of this study, the finding indicate that IR does not interfere with the bone repair process when delivered post-surgery. While manual segmentation is more time consuming, it retrieves greater bone volume and reveals differences in trabecular structure, showing more trabeculae number and less trabeculae separation in irradiated bone.