Objective: This study presents a method to determine the volumetric density of pelvic bone and cortical bone thickness along critical regions of the S2 alar-iliac (S2AI) screw trajectory using bi-planar multi-energy X-ray (BMEX). Methods: Simulated BMEXs were generated from CT data from eight patients, with coordinate matching linking pixels to voxel density values. This dataset included pixel attenuation values, coordinates as independent variables, and voxel attenuation values (Hounsfield Units, HU) as the dependent variable for training a random forest regressor model. Results: The trained model revealed adequate trabecular bone density prediction (root mean square error: 32.8 mg/cm³) and cortical thickness accuracy (error ≤1.2 mm). Trabecular bone showed a minor tendency for density overestimation with a maximum difference of 83 HU, while cortical bone exhibited an underestimation of up to 118 HU. Conclusions: The improved prediction of bone density and the capability to estimate cortical bone thickness signify a significant advancement towards a comprehensive modality for predicting bone quality in implant placement planning.