Method and device for automatically identifying nasopharyngeal carcinoma primary tumors

Abstract

The invention relates to the field of image processing, and provides a method and device for automatically identifying nasopharyngeal carcinoma primary tumors. The method comprises the steps: employing a ResNeXt152 network as a backbone network in a Mask RCNN deep semantic segmentation network, and defining the prediction types of semantic segmentation branches into three types; training the deepsemantic segmentation network by utilizing a plurality of preprocessed magnetic resonance three-dimensional images, and outputting pixel position information of four end points of a rectangular framecontaining a tumor region, the probability that the rectangular frame is the tumor region and the probability that each pixel in the rectangular frame belongs to a primary tumor region; and inputtinga to-be-recognized magnetic resonance three-dimensional image into the deep semantic segmentation network to obtain the probability that each pixel belongs to a primary tumor region. According to themethod, through a deep learning method, a Mask RCNN network architecture is utilized and improved, so that the prediction accuracy and the generalization ability of the model can be effectively improved.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method and device for automatically identifying primary tumors of nasopharyngeal carcinoma. Background technique [0002] In the medical field, IMRT technology can greatly improve the survival rate and quality of life of cancer patients. However, this advanced treatment method requires accurate judgment of the contour of the target tumor, and the development of radiation treatment plan can take hours of radiologists. [0003] Because rhinitis is adjacent to the skull base, 60%-70% of patients experience skull base bone destruction at the time of treatment, and 18% of patients even have intracranial and / or cavernous sinus violations, making surgery difficult. At the same time, 85% of patients are accompanied by metastasis of neck or throat lymph nodes, which are not suitable for surgery. The World Health Organization divides nasopharyngeal carcinoma into three types: ...

AI Technical Summary

Problems solved by technology

[0005] Embodiments of the present invention provide a method and device for automatically identifying primary tumors of nasopharyngeal carcinoma, to solve the problem of low accuracy in identifying primary tumors in the prior art

Images