Advanced CT visualization improves the accuracy of orthopaedic trauma surgeons and residents in classifying proximal humeral fractures: a feasibility study

Osteosynthesis of proximal humeral fractures remains challenging with high reported failure rates. Understanding the fracture type is mandatory in surgical treatment to achieve an optimal anatomical reduction. Therefore, a better classification ability resulting in improved understanding of the fracture pattern is important for preoperative planning. The purpose was to investigate the feasibility and added value of advanced visualization of segmented 3D computed tomography (CT) images in fracture classification. Seventeen patients treated with either plate-screw-osteosynthesis or shoulder hemi-prosthesis between 2015 and 2019 were included. All preoperative CT scans were segmented to indicate every fracture fragment in a different color. Classification ability was tested in 21 orthopaedic residents and 12 shoulder surgeons. Both groups were asked to classify fractures using three different modalities (standard CT scan, 3D reconstruction model, and 3D segmented model) into three different classification systems (Neer, AO/OTA and LEGO). All participants were able to classify the fractures more accurately into all three classification systems after evaluating the segmented three-dimensional (3D) models compared to both 2D slice-wise evaluation and 3D reconstruction model. This finding was significant (p < 0.005) with an average success rate of 94%. The participants experienced significantly more difficulties classifying fractures according to the LEGO system than the other two classifications. Segmentation of CT scans added value to the proximal humeral fracture classification, since orthopaedic surgeons were able to classify fractures significantly better into the AO/OTA, Neer, and LEGO classification systems compared to both standard 2D slice-wise evaluation and 3D reconstruction model.