MVCT Image Texture Enhancement Method Based on Double Regular Constraints

Abstract

The invention discloses an MVCT image texture enhancement method based on double canonical constraints, which mainly solves the problem that the MVCT image cannot be enhanced in the prior art. The scheme is: 1) Obtain multiple KVCT and MVCT images from the same part of the human body; 2) Normalize the obtained CT image data sets, and then perform block extraction on each pair of CT images to obtain a CT image block data set; 3) Establish a 13-layer MVCT image texture enhancement network, use the CT image block dataset as training data, use the gradient descent algorithm to optimize the network, and obtain a trained network; 4) Input a complete MVCT image into the trained network In the network, the enhanced MVCT image can be output. The invention can better maintain the edge and details of the image while enhancing the image texture, improve the image quality, facilitate doctors to read and diagnose the MVCT image, correct the error of the lesion position, and ensure the accuracy of radiotherapy.

Description

technical field [0001] The invention belongs to the technical field of medical image processing, in particular to a method for MVCT image texture enhancement, which can be used to improve the CT image quality and the visual effects of imaging organs and tissues. Background technique [0002] Megavoltage computed tomography (MVCT) and kilovolt computed tomography (KVCT) are two common forms of X-ray CT. Compared with KVCT, the voltage of the imaging tube of MVCT equipment is higher, which can find cancerous areas in tissues and organs, and is widely used in preoperative radiation therapy of tumors and cancers. However, MVCT is noisy and not suitable for tracking treatment. The current mainstream method is to use MVCT imaging before treatment, register with the KVCT image of the radiotherapy plan, and correct the error of the lesion position to ensure the accuracy of radiotherapy. As people pay more attention to CT radiation, the utilization rate of MVCT, which is less harmf...

AI Technical Summary

Problems solved by technology

[0003] At present, MVCT image enhancement mainly focuses on the direction of image denoising. There are two types of methods, one is based on projection domain, which includes bilateral filtering, static wavelet transform, maximum a posteriori probability estimation, etc. The signal dependence is strong, and the resolution of the image after denoising will be reduced to a certain extent; the other is the method based on neural network, which includes RED-CNN, denoising autoencoder, DnCNN, etc. After denoising, these methods There is no significant improvement in the visual effect and contrast of the image, and there will be soft tissue edge blurring and other phenomena

[0004] In addition, the biggest shortcoming of these two types of methods is: since the enhancement task of the MVCT image is only simplified to the denoising task, this simple denoising can neither greatly improve the image quality, but also lose the contrast of the MVCT image and details

Images