A linear regression model using computed tomography of forearm osteology to predict radius and ulna characteristics for surgical planning

Abstract

Introduction: The radius and ulna are commonly fractured bones. The restoration of the native anatomy is the primary surgical objective but can be difficult due to a mismatch between the bones' shape and available implants. A thorough understanding of the underlying anatomical relationships between the radius and ulna could allow for a more accurate prediction of variables, thus enabling the surgeon to treat patients more effectively. Methods: A cross-sectional investigation of forearm computed tomography scans and measurements were conducted on 97 forearms. Pearson's correlations were used to evaluate relationships between variables, and those with a coefficient of r > 0.4 and P < 0.001, as well as those considered clinically relevant, were carried forward into a multiple linear regression for each outcome variable, namely: (i) radius length, (ii) radius of curvature, (iii) the minimum diameter of the radial canal, (iv) ulna length, and (v) the minimum diameter of the ulna canal. A stepwise approach was used for the multiple linear regression analysis, with a significance level of 0.05 for predictor variables. Results: Radius length: in the multiple linear regression model, only ulna length remained in the model (adjusted R2 = 0.85). The radius of curvature: the final model only included ulna length (adjusted R2 = 0.30). Radius canal minimum width: three measurements were included in the final model (adjusted R2 = 0.82). Ulna Length: six independent correlations between individual measurements and the ulna length were observed, with radius length and the radial neck length being included in the final model (adjusted R2 = 0.86). Ulna canal minimum width: the final regression model included four variables: the maximum diameter of the distal third of the radial canal, the minimum diameter of the radial canal, and the minimum diameter of the proximal and middle third aspects of the ulna canal (adjusted R2 = 0.80). Conclusion: The results of this investigation illustrate that anatomical predictions for bone size can be made using other anatomical landmarks except for the radius of curvature. The clinical application and implementation of this statistical model need further research.