基于改进MaxViT的胰腺肿瘤图像分割

Pancreatic Tumor Image Segmentation Based on Improved MaxViT

近年来,基于深度学习的图像分割技术在医学图像处理中取得了良好的效果,而在腹部CT图像中分割胰腺肿瘤仍然面临着挑战。由于胰腺肿瘤图像存在大小不一、边界模糊、冗余信息繁杂等问题,现有的网络难以做到精准分割。基于此,本文将MaxViT多尺度信息提取注意力模块嵌入UNETR网络结构,用以改善基准网络模型所存在的不足。

In recent years,deep learning image segmentation techniques have shown promising results in medical image processing.However,segmenting pancreatic tumors in abdominal CT images faces challenging.Due to issues such as varying sizes,fuzzy boundaries,and complex redundant information in pancreatic tumor images,existing networks struggle to achieve precise segmentation.To address this,this study embeds the MaxViT multiscale information extraction attention module into the UNETR network structure to improve upon the deficiencies of the baseline network model.This paper proposes an image segmentation network based on an enhanced UNETR model.