DDAUnet-based esophageal tumor segmentation method

Abstract

The invention belongs to the technical field of image segmentation, and particularly relates to a DDAUnet-based esophageal tumor segmentation method. The DDAUnet-based esophageal tumor segmentation method comprises the following steps of data acquisition, data preprocessing, model construction, model storage and model evaluation, wherein in the data acquisition, related CT scanning images of an esophageal cancer patient are acquired, image labeling is performed, and construction of a data set required by model training is completed; and the data preprocessing comprises data amplification, data division, image scaling and normalization. The invention provides a full-automatic end-to-end esophageal tumor segmentation method based on a convolutional neural network CNNs, the network is called as dilated intensive attention Unet, namely DDAUnet, space and channel attention gates in each dense block are selectively concentrated on a decisive feature map and a region, and an expanded convolutional layer is used for managing a GPU memory and increasing a network acceptance field. The method is used for image segmentation.

Description

technical field [0001] The invention belongs to the technical field of image segmentation, and in particular relates to an esophageal tumor segmentation method based on DDAUnet. Background technique [0002] Esophageal cancer is one of the least studied cancers, and it is fatal to most patients. Rapid and accurate delineation of target volume in CT images is very important for treatment and disease control. High uncertainty, especially at the cranial and caudal margins of tumors, to overcome these complexities, physicians correlate CT imaging with clinical history, endoscopic findings, endoscopic ultrasound, and other imaging modalities such as positron emission tomography Combined, however, acquiring these additional modalities is a time-consuming and expensive process. [0003] Causes of problems or deficiencies: Manual or automated delineation of esophageal tumors in CT images is very challenging. This is due to low contrast between the tumor and adjacent tissues, chang...

AI Technical Summary

Problems solved by technology

Obtaining its additional information is time-consuming, and at the same time the results are error-prone and may lead to uncertain results. In recent years, the application of deep learning in medical image analysis has attracted much attention. limited applications, and even less in esophageal tumor segmentation

Images