Noise and Artifact Removal Method for Medical Spectral CT Image

Abstract

The invention relates to the field of computer tomography, in order to reduce the influence of beam hardening and random noise, output more accurate medical images, and reduce the damage suffered by the human body. To this end, the present invention provides a noise and artifact removal method for medical energy spectrum CT images. The steps are as follows: Step1: construct a virtual phantom; Step2: use a large number of X-ray photons to pass through the phantom to simulate energy spectrum CT imaging; Step3: use The lower dose X-ray photons are used to simulate the spectral CT imaging through the phantom; Step 4: Match the reconstructed images; Step 5: Train the convolutional neural network to complete the training; Step 6: Test the network training effect. The invention is mainly applied to the design and manufacture occasions of medical CT equipment.

Description

technical field [0001] The invention relates to the field of computer tomography (Computed Tomography). For the problem of low signal-to-noise ratio caused by the small number of photons collected in different energy intervals of energy spectrum CT, a convolutional neural network model is used to detect noise and noise in each energy interval. Reconstructed images of artifacts are processed to improve the accuracy of medical images while reducing radiation dose. Specifically, it relates to a noise and artifact removal method for medical energy spectrum CT images. Background technique [0002] The key to medical CT imaging is mainly divided into two points: one is to improve the image accuracy, and the other is to reduce the radiation dose. Multi-energy spectral CT, represented by single-photon counting detectors and energy-integrating detectors, can provide more accurate image information and improve the accuracy of medical imaging, and the more and narrower the energy inte...

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the deficiencies of the prior art and aim at the problems of low-dose noise and artifacts in spectral CT imaging, the present invention aims to propose a method for medical X-ray spectral CT image processing based on convolutional neural network, After a low-dose X-ray scan, the pre-trained convolutional neural network model is used to process the images containing noise and artifacts in each energy range, and restore them into reconstructed images with the same quality as those under high-dose X-ray irradiation conditions. It can reduce the influence of beam hardening and random noise, output more accurate medical images, and reduce the damage to the human body

Images