Sarcoidosis benign and malignant prediction method based on ResNet-Inception model

Abstract

The present invention provides a sarcoidosis benign and malignant prediction method based on a ResNet (Residual Network)-Inception model. The method comprises the steps of: obtaining a sarcoidosis image with a tag of a known sarcoidosis area, performing preprocessing of the sarcoidosis image with the tag, and obtaining a training image data set, a verification image data set and a prediction imagedata set; establishing a sarcoidosis image classification model based on a ResNet-Inception model; and inputting the predication image data set into a final mode of the sarcoidosis image classification model based on the ResNet-Inception model, and obtaining a sarcoidosis benign and malignant prediction result of the sarcoidosis image in the prediction image dataset. The method designs a new network structure model, and can obtain sarcoidosis benign and malignant prediction according to a sarcoidosis CT image.

Description

technical field [0001] The invention belongs to the technical field of medical image processing, and in particular relates to a method for predicting benign and malignant pulmonary nodules based on a ResNet-Inception model. Background technique [0002] The lesions of lung diseases on medical imaging usually appear as pulmonary nodules. According to the continuous research of pattern recognition and other methods, people have proposed the technology of using computers to assist radiologists to detect pulmonary nodules, that is, Computer Aided Diagnosis (Computer Aided Diagnosis). Diagnosis, CAD) system. In CAD, it involves evaluation or classification with some common algorithms, such as algorithms such as neural networks, k-means or support vector machines. However, the main disadvantage of the traditional algorithm is that the extracted features are only local areas, which are not robust enough for common variance and other features. Based on the differences of data sets...

AI Technical Summary

Problems solved by technology

However, the main disadvantage of the traditional algorithm is that the extracted features are only local areas, which are not robust enough for common variance and other features.

Images