Nasopharyngeal carcinoma necrosis prediction method based on AdaBoost feature fusion

Abstract

The invention discloses a nasopharyngeal carcinoma necrosis prediction method based on AdaBoost feature fusion, and the method comprises the steps: carrying out the image preprocessing through the scanning images of the first three modals of nuclear magnetic radiation of a nasopharyngeal carcinoma patient, drawing a tumor region as a mask by an experienced doctor, and extracting the radiomics features according to the two groups of images; for a dose image, features in the image are extracted manually, and feature screening is carried out in combination with radiomics features. And finally, performing model building, model testing and the like on the screening features by using a model fusion method AdaBoost based on a decision tree. The nasopharyngeal carcinoma necrosis condition is predicted according to the image data and the radiotherapy plan of the patient before radiotherapy, and the radiotherapy plan can be modified, so that the patient is prevented from being influenced by side effects of radiotherapy; by using a traditional feature extraction method and a traditional machine learning method, the model training time is reduced to a certain extent, and the interpretability between results and features is improved.

Description

technical field [0001] The invention relates to the technical field of nasopharyngeal carcinoma necrosis prediction, in particular to a nasopharyngeal carcinoma necrosis prediction method based on AdaBoost feature fusion. Background technique [0002] Nasopharyngeal carcinoma is a common malignant tumor of the head and neck. So far, radiotherapy is still considered to be the main treatment method for nasopharyngeal carcinoma because most of the pathological types of nasopharyngeal carcinoma are poorly differentiated squamous cell carcinoma, which has high sensitivity to radiation. It also ensures the integrity of the anatomical structure of the nasopharynx and neck, and improves the quality of life of patients. Since the development of deep beam radiotherapy for nasopharyngeal carcinoma, the local control rate of radiotherapy alone for nasopharyngeal carcinoma has been achieved through decades of radiotherapy technology. Some patients with early stage nasopharyngeal carcin...

AI Technical Summary

Problems solved by technology

Given that radiotherapy necrosis in nasopharyngeal carcinoma is not only affected by the characteristics of the tumor itself, it is difficult to make a good prediction based on omics characteristics alone.

Images