Bone tumor boundary demarcation based on multimodal image fusion

The delineation of bone tumor boundaries is a critical issue in the field of medical image segmentation due to the unique positioning of these tumors and the complexity of the associated surgical procedures. In recent years, researchers have made significant strides in determining tumor boundaries through computer version methods. Doctors can achieve precise localization of bone tumors using the U-Net network or the improvements. However, using tumor boundary segmentation models based on single-modal images makes it challenging to capture the complete characteristics of the tumor. In this paper, we propose a multi-modal image fusion network to achieve more accurate segmentation of tumor boundaries. Experimental results show that the accuracy of the segmentation results can be improved by about 6.8% when using fused images for segmentation. Therefore, this image processing process is of great significance for improving the accuracy of bone tumor boundary demarcation and the efficiency of clinical diagnosis.