Automatic Tumour Segmentation Network Based on PLHN Architecture and Dynamic Contrast-Enhanced Magnetic Resonance Imaging

Tumor diagnosis often requires hemodynamic information to be judged. DCE-MRI technology can record and compare information well and has become an important tool in the diagnosis process. However, at this stage, the automatic segmentation technology for tumors is still not perfect, which is mainly due to the diversity of tumor morphology and signal intensity, and the insufficient performance of existing deep learning models in dealing with complex structures and image noise. In order to solve these problems, this paper proposes an automatic tumor segmentation method based on pyramid hierarchical network (PLHN) architecture and DCE-MRI. This method enhances the network's ability to recognize tumor boundaries and details through multi-scale feature extraction and hierarchical processing, and shows stronger robustness when dealing with dynamic changes in DCE-MRI. According to our experimental results on the data set, the method proposed in this paper can effectively improve the accuracy of tumor segmentation, which has strong research and practical significance.