Abstract
Introduction: Articular fractures, especially in weight-bearing joints of the lower extremity, require anatomical reduction for good long-term function. If anatomic reduction is achieved 90% of patients have good results. Accurate anatomical reduction of acetabular fractures is made difficult y the complicated anatomy of the acetabulum and pelvis. Plain films and computerised tomography are very useful in the assessment, classification and planning for surgery of these fractures, but conventional imaging only provides two dimensional images of these fractures. While interpretation of traditional imaging techniques becomes better with experience novel techniques may assist in the understanding of these complex injuries. This project presents the use of three dimensional life size models which can be rapidly manufactured from CT scans to facilitate fracture classification and operative planning.
Patients and Methods: Standard 4 mm CT cuts were reformatted to 2 mm and saved as DICOM files for a series of 15 consecutive patients with complex acetabular fractures (not including isolated posterior wall injuries). These were imported into Mimic (Materialise) data conditioning software, which allowed manipulation and thresholding of images so that a three-dimensional computer model could be built. The data was exported to the Selective Laser Sintering (SLS) system and a life size three-dimensional model constructed. Three consultant orthopaedic surgeons and three senior trainees were asked to classify, using the system of Letournel, each fracture using 1) conventional radiographs (AP pelvis, Judet views and CT scans). 2) Using a model. As no absolute of fracture type was available the Kappa statistic was used to evaluate inter and intraobserver agreement. This compared an observed level of agreement with the level of agreement that would be expected by chance alone.
Results: Interobserver agreement was not absolute using either conventional radiographs or the sintered model. For the consultants the kappa statistic using conventional radiographs was 0.61 while the kappa value using the model was 0.76 (p=< 0.05). For the trainees the kappa was 0.42, using conventional radiographs and 0.71 using the model (p=< 0.01)>
Discussion: The complex three-dimensional anatomy of the pelvis and acetabulum make assessment and classification of fractures of these structures notoriously difficult. The value of accurate fracture classification is well recognised particularly in the treatment of intra-articular injuries. The use of this real life model of a fractured pelvis has been shown in this study to significantly reduce the degree of interobserver variability in the classification of these injuries. This effect is particularly evident for less experienced surgeons.
Conclusion: The technique is available and relatively inexpensive. These models may be of great benefit for preoperative planning, classification and assessment of prognosis in acetabular fractures. Its use should prove invaluable as a tool to aid clinical practice.
The abstracts were prepared by Raymond Moran. Correspondence should be addressed to him at the Irish Orthopaedic Assocation, c/o Cappagh National Orthopaedic Hospital, Finglas, Dublin 11, Ireland.
