Prediction of femoral fracture load: cross-sectional study of texture analysis and geometric measurements on plain radiographs versus bone mineral density.

Abstract

To use standard radiographs to determine which combination of co-occurrence textural parameters, geometric measurements, and cortical thickness measurements from femur radiographs provided the best estimate of femoral failure load and to compare these with total hip dual-energy x-ray absorptiometry bone mineral density (BMD) evaluation. Digital radiographs of 40 pairs of excised femurs (24 women, 16 men; mean age, 82 years + or - 12 [standard deviation]) were obtained. Regions of interest in the femoral neck, greater trochanter, intertrochanteric area, and femoral head were then selected. Three textural parameters derived from a co-occurrence matrix were estimated with imaging software. Neck-shaft angle, femoral neck axis length, calcar femorale thickness, and internal and external femoral shaft thickness were assessed. The femurs were randomly allocated to single-stance (femoral neck fracture) or side-impact (intertrochanteric fracture) configurations for failure load measurement. Textural parameters correlated significantly with site-matched BMD. Stepwise regression analysis was performed, and total hip BMD explained 73% and 78% of the failure load in single-stance and side-impact configurations, respectively. Combining internal femoral shaft thickness with one or two textural parameters explained 72%-79% of failure load variance in the single-stance configuration and 63%-76% of failure load variance in the side-impact configuration. In these excised femurs, combining textural parameters with cortical thickness measurements had a performance comparable to that of BMD alone in the explanation of femoral failure load.