Method and device for decomposing base material in dual-energy spectrum CT projection domain based on deep learning

Abstract

The invention discloses a method and a device for decomposing a base material in a multi-energy spectrum CT projection domain based on deep learning. The method includes: a network training process, using the multi-energy spectrum CT to collect the multi-energy spectrum and multi-color projection of a known calibration phantom; The volume CT image is directly reconstructed with a multi-color projection under one energy spectrum, and divided into multiple base material images, and the line integrals of these base material images along each ray are obtained respectively; designed for multi-color projection decomposition The deep neural network uses the multi-spectral multi-color projection of the calibration phantom as the network input, and the line integral of the base material image of the calibration phantom as the output label to complete the training; in the network application process, the multi-spectrum multi-color of the measured object is The projection is input into the neural network, and the line integral of the multi-base material image is decomposed; then the multi-base material density image of the measured object is reconstructed from these line integrals. The invention has good anti-noise performance, has no strict requirements on the morphological and structural correlation between the calibration phantom and the measured object, and does not need to measure energy spectrum information in advance.

Description

technical field [0001] The invention relates to the technical field of X-ray multi-spectrum CT imaging, in particular, to a multi-spectrum CT-based material image decomposition method based on deep learning. Background technique [0002] X-ray CT imaging technology is a technology that uses the principle of interaction between X-rays and matter to image the internal information of the scanned object. X-rays emitted by conventional CT ray sources have different wavelengths and frequencies, and obey a certain continuous polychromatic energy spectrum distribution like visible light. Conventional CT acquires the polychromatic projection of the X-ray energy spectrum through the object to be measured, rather than the ideal single-energy X-ray projection. The images reconstructed by the traditional single-energy CT reconstruction algorithm not only have significant hardening artifacts, but also different substances may have the same or similar CT values, which are difficult to dis...

AI Technical Summary

Problems solved by technology

When the convolutional neural network processes each pixel of the image, it uses a large amount of neighborhood information, which has the advantage of good denoising performance, but at the same time it causes a disadvantage, that is, the difference between the samples for training the network and the image of the object to be processed by the network must be very close in structure

Since it is difficult to obtain a large amount of information of the image to be processed in advance to train the convolutional neural network in many cases, the scope of application of this type of method is severely limited.

Images