A False Positive Screening Method for Pulmonary Nodules Based on Convolutional Neural Network

Abstract

The invention discloses a pulmonary nodule false positive screening method based on a convolutional neural network, comprising: (1) obtaining a pulmonary nodule detection result output by a pulmonary nodule detection model; (2) marking the pulmonary nodule detection result Generate a sample; (3) construct a data set according to the sample and the original lung CT image; (4) perform a random offset in 8 directions on the coordinate x of each data pair in the data set, and the offset scale is 0.5X; ( 5) For each data pair in the data set, crop the original lung CT image according to the sample, obtain training samples of different sizes, and augment the training samples; (6) Construct a convolutional neural network; (7) ) Use training samples of 3 sizes to train the convolutional neural network to obtain three pulmonary nodule false positive screening models; (8) use the pulmonary nodule false positive screening model to predict the samples to be tested, and output the prediction results.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a method for screening false positives of pulmonary nodules based on a convolutional neural network. Background technique [0002] Deep learning methods have made great achievements in the field of image processing, which also provides the possibility to apply deep learning technology to identify feature parts in medical image data. At present, the CAD (computer aided diagnosis) system based on deep learning has a wide range of applications in identifying and segmenting organs and feature regions in CT images. [0003] Existing machine learning methods contain a large proportion of false positive results in the pulmonary nodule recognition results of lung CT images, that is, areas that do not have the characteristics of pulmonary nodules are predicted to be positive for pulmonary nodules, resulting in a decrease in the accuracy of the detection results decrease. Con...

AI Technical Summary

Problems solved by technology

[0003] Existing machine learning methods contain a large proportion of false positive results in the pulmonary nodule recognition results of lung CT images, that is, areas that do not have the characteristics of pulmonary nodules are predicted to be positive for pulmonary nodules, resulting in a decrease in the accuracy of the detection results reduction of

Images