Radiomic characterization of sinus lift grafts using texture analysis technique in cone-beam computed tomography images

Objective: This study aims to apply texture analysis (TA) to cone-beam computed tomography (CBCT) images as a predictive tool for characterizing Geistlich Bio-Oss® bone grafts used in maxillary sinus lift (SL) surgeries. Material and Methods: Eleven CBCT scans of patients who underwent bilateral sinus lift (SL) procedures with Bio-Oss® grafts were selected from the Radiology database of ICT-UNESP. The scans had been acquired using an I-CAT Next Generation scanner (voxel size: 0.25 mm; field of view: 16.0 × 8.0 cm) and exported in DICOM format to OnDemand 3D software. Five parasagittal slices from each case were saved in bitmap format. Texture analysis was performed using MaZda 3.20 software. A 4.0 × 4.0 mm region of interest (ROI) was positioned at the center of the graft, and texture features were extracted using the gray-level co-occurrence matrix (GLCM). All analyses were performed by a single calibrated evaluator. Data normality was assessed using the Shapiro–Wilk test. Since the variables did not present a normal distribution, data were transformed into ranks. Spearman’s correlation coefficient was used to evaluate the association between directions and distances of texture measurements. A mixed linear model was applied for each texture parameter, considering maxillary sinus (right/left), section (S0–S4), and their interaction as fixed effects, with patient as a random effect. Intra-examiner reproducibility was assessed using the intraclass correlation coefficient (ICC). Statistical significance was set at 5%. Results: Most texture parameters showed a homogeneous distribution between the analyzed sides and slices, except for the SumAverg parameter, which indicated higher density on the right side. Conclusion: The TA proved to be an effective tool for the quantitative characterization of grafts, revealing symmetrical and homogeneous bone integration of the biomaterial, with potential clinical application in surgical planning and monitoring.