A chest X-ray film denoising method based on a deep convolutional neural network

Abstract

The invention discloses a chest X-ray film denoising method based on a deep convolutional neural network. The method comprises the steps of collecting chest X-ray film data, performing data format conversion, performing preprocessing to obtain an original image block for training, adding Gaussian noise to generate a noisy image block, and taking the paired original image block and noisy image block as a training data set; Constructing a convolutional neural network model for removing chest X-ray film noise, wherein the convolutional neural network model comprises a deep convolutional neural network and a residual error network; Taking the paired original image blocks and noisy image blocks as input, and performing training to obtain a trained convolutional neural network model X-ReCNN; taking the chest X-ray data of the noise to be removed as an input characteristic map of X-ReCNN, removing noise, and outputting the predicted denoised chest X-ray film.. According to the method, the noise in the chest X-ray film can be removed with light weight, high speed and high precision, the parameters of a network structure are greatly reduced, and the network training time is shortened.

Description

technical field [0001] The invention belongs to the technical field of image denoising, and in particular relates to a method for denoising a chest X-ray film based on a deep convolutional neural network. Background technique [0002] Chest X-ray is an inexpensive, fast and easy-to-obtain medical imaging technique. Compared with computed tomography (CT), traditional chest X-rays are not only cheaper, but also have lower radiation dose. Now medical research shows that 1.5% to 2% of tumors may be due to high radiation from CT Dosage causes. Especially in the early diagnosis of diseases such as pneumonia, pneumothorax, interstitial lung disease, heart failure, fracture, hiatal hernia, etc., X-rays are widely used. In addition, chest X-ray examination is the standard examination method for more than 300 million people in the country in 2014. This number is still increasing, resulting in hundreds of millions of chest X-rays being produced every year. [0003] Although chest X...

AI Technical Summary

Problems solved by technology

[0005] Denoising autoencoders and convolutional denoising autoencoders (CNN DAE) are extensions of classic autoencoders. Both models try to learn clean denoised pictures directly by building models, but the robustness of the models And the generalization ability is not good enough, the actual performance effect is not good when training

Images