Comparison of three-dimensional and two-dimensional computed tomographies in the classification of acetabular fractures

Purpose This study compared the accuracy and timeliness of two-dimensional computed tomography (2DCT) and three-dimensional computed tomography (3DCT) in the diagnosis of different types of acetabular fractures and by different groups of interpreters using the Letournel and Judet classification system. Methods Twenty-five fractures cases, five each of five common types of acetabular fractures, were selected. Nineteen interpreters with different levels of experience (ten graduate trainees and nine radiologists) individually classified the fractures using multiplanar 2D and standardized 3DCT images. The 3DCT image set was comprised of 39 images of rotational views of the entire pelvis and the disarticulated fracture hip. Consensus reading by three experts served as a reference standard. Results Classification accuracy was 66% using 2DCT, increasing to 73% (p = 0.041) when 3DCT was used. Improvement occurred in the interpretation of transverse and posterior wall-type fractures (p < 0.01 and p = 0.015, respectively), but not in T-type, transverse with posterior wall, or both-column fractures. The improvement was noted only in the graduate trainee group (p = 0.016) but not the radiologist group (p = 0.619). Inter-observer reliability in the graduate trainee group improved from poor to moderate with 3DCT, but remained at a moderate level in both 2DCT and 3DCT in the radiologist group. The overall average interpretation time per case with correct diagnosis was 60 s for 2DCT but only 32 s for 3DCT. Conclusions Standardized 3DCT provides greater reliability and faster diagnosis of acetabular fractures and helps improve the accuracy in transverse- and posterior wall-type fractures. In addition, it helps improve the accuracy of less experienced interpreters.