Tumor identification method

Abstract

The invention discloses a tumor identification method, which comprises the following steps: rough segmentation: sequentially performing organ segmentation and blood vessel segmentation on a CT image to obtain a CT image of an organ with blood vessel being removed; constructing a test sample data set: based on an the CT image of the organ, sequentially constructing the tumor area sub-graph sets for an arterial phase, a venous phase and a delayed phase to form a test positive sample data set and the organ normal area subgraph set constitutes the test negative sample data set. The training tumor classifier: extracts the characteristic data of the test sample data set, Good/bad identification and staging of the tumor classifier; feature extraction of CT images and tumor identification and staging: the arterial phase, venous phase and the delay of the CT images to be extracted feature data extraction to form the characteristic data to be measured Set, will feature data to be tested input tumor classification for benign / malign tumor classification and staging. The present invention can accurately distinguish between benign and malignant staging and staging of the segmented tumor.

Description

technical field [0001] The invention relates to the technical field of benign / evil tumor identification, and more specifically, the invention relates to a tumor identification method. Background technique [0002] The clinical manifestation of the tumor on the image is that there are several tissue regions of different sizes and irregular shapes in the organ parenchyma, and the boundaries of these regions are relatively fuzzy. The internal density of the lesion area of ​​small tumors is similar, mostly lower than that of the surrounding normal renal parenchymal tissue, while large tumors may have internal necrosis or Koreanization, making their density heterogeneous. Based on the clinical manifestations of tumors on images, the following two types of segmentation methods exist in the prior art. [0003] A class of segmentation methods based on pattern recognition techniques. Among them, the most commonly used in organ segmentation is an unsupervised algorithm—fuzzy C-means...

AI Technical Summary

Problems solved by technology

[0005] Although the above two segmentation methods can segment tumors in human organs, they cannot perform benign/malignant identification and malign

Images