Classification of acetabular fracture: Accuracy and interpreting duration by suing 3D-volume rendered and MPR images

Abstract

Objective: To retrospectively assess the time efficiency and accuracy of three-dimensional (3D) volume-rendered images obtained from a standard pattern of reconstruction in the interpretation of acetabular fracture. Methods and Materials: Two inexperienced radiology residents evaluated twenty acetabular fractures (8 elementary and 12 associated fractures) and independently classified the fracture with Multiplanar reformatted (MPR) images alone and 3D images alone by using Letournel and Judet classification system. Kappa value were calculated to assess agreement of reader with reference standard. The reference standard was a consensus reading of a musculoskeletal radiologist and orthopedic surgeon by using a combination of the 3D-volume rendered and MPR image findings and operative findings. Results: Comparing the reference standard with two different image sets, 3D images had a worse kappa value than MPR images (reader 1: k3D = 0.61 and KMPR = 0.86; reader 2: k3D = 0.58 and KMPR = 0.74). The associated fracture subgroup was frequently associated with disagreement (reader 1: k3D = 0.34 and KMPR = 0.62; reader 2: k3D = 0.065 and KMPR = 0.37). The average time required for diagnosis was significantly shorter with 3D images than with MPR images (reader 1: 52.50 vs140.5 seconds, P < 0.05; reader 2: 76.50 vs 82.50 seconds, P = 0.896; all readers 66.50 vs 105.00 seconds, P< 0.05 ). Conclusion: Although, 3D images reveal time reduction in the interpretation of acetabular fracture, the accuracy was worse than MPR images, especially in the associated fracture subgroup. This led up to conclude that 3D image is an important adjunct to axial and MPR images for the evaluation of acetabular fracture.