Enhanced YOLOv8-seg-SPDConv for accurate Schatzker classification of tibial plateau fractures

Tibial plateau fractures are caused by violent direct or indirect injuries, and their accurate classification according to the Schatzker classification is important for determining appropriate treatment. The study developed and validated a YOLOv8n-seg-SPDConv model for automated classification of these fractures using CT images from 552 patients. The model achieved overall accuracy of mAP50 0.914 and increased processing speed to 226 FPS, enabling real-time deployment. Simultaneously, the number of model parameters was reduced from 3.01 million to 2.64 million. The improved model showed better results especially for complex fracture types with low contrast. The system is designed as a clinical decision support tool to assist physicians rather than replace expert judgment.