MATERIALS AND METHODS: 52 healthy volunteers were scanned in a 16-slice MDCT, and the volume of 104 sets of carpal bones was measured using a Syngo workstation (Both CT and workstation-Siemens Healthcare, Erlangen, Germany).
RESULTS: Male carpal bones were of higher volume compared to the female carpal bones (p<0.001). Area under the curve (AUC) assessment of responder-operator characteristics curves showed that the trapezium bone was best able to predict sex with an AUC of 0.986. At a trapezium bone volume of ⩾1.94cm(3), there was a 93.5% probability that the subject was male. Binary logistic regression analysis found that the highest accuracy was derived using the pisiform, trapezium and capitate bones. There was a strong relationship between sex prediction and grouping of the carpal bone volumes (Nagelkerke R(2)=0.923) with an overall prediction accuracy of 97%.
CONCLUSION: All 8 carpal bones exhibit sexual dimorphism to varying degrees. A binary regression analysis combining the 5 carpal bones with the highest predictive values for sex produces an accurate predictive model.