Navigating new waters in proximal humerus fractures
Navigating new waters in proximal humerus fractures
One of the most challenging issues in the field of shoulder fractures is that fracture assessment is hampered by poor interobserver agreement: surgeons assess and interpret fracture patterns differently, leading to treatment variation and potentially inferior outcomes for patients.
In this thesis, Reinier Spek explores new approaches to address this intrinsic problem, using emerging innovations such as 3D-printed models and convolutional neural networks. However, we showed that 3D-printed models did not improve interobserver agreement and are therefore not recommended to be used in daily clinical practice. Whilst automated simple objectives such as fracture recognition on radiographs are on par with human observations, convolutional neural networks still underperform for complex diagnostic tasks.
Furthermore, we found that collar and cuff treatment does not improve fracture alignment when patients are treated non-operatively, and that pre-operative virtual planning software may be a valuable tool to enhance fracture reduction in patients undergoing surgical plate fixation.