Multi-Scale Tumor Localization Based on Priori Guidance-Based Segmentation Method for Osteosarcoma MRI Images

Osteosarcoma is a malignant osteosarcoma that extremely harmful to human health. Magnetic resonance imaging (MRI) technology one of the commonly used methods for examination osteosarcoma. Due large amount MRI image data and complexity detection, manual identification in images time-consuming labor-intensive task doctors, it highly subjective, which can easily lead missed misdiagnosed problems. AI medical image-assisted diagnosis alleviates this problem. However, brightness multi-scale make existing studies still face great challenges tumor boundaries. Based on this, study proposed prior guidance-based assisted segmentation method osteosarcoma, based few-shot technique fine fitting. It not only solves problem localization, but also greatly improves recognition accuracy First, we preprocessed using generation normalization algorithms reduce model performance degradation caused by irrelevant regions high-level features. Then, prior-guided feature abdominal muscle network perform small-sample tumors different sizes features processed images. Finally, more than 80,000 from Second Xiangya Hospital experiments, DOU value paper reached 0.945, at least 4.3% higher other models experiment. We showed our specifically has prediction lower resource consumption.