RU2575405C2 - Low-dose computed tomography (ct) - Google Patents

Abstract

FIELD: medicine.

SUBSTANCE: invention refers to computed tomography. A method involves the stages of: eliminating a projection records noise from low-dose scanning of a projection area; the low-dose scanning matches with that one forming image records having the noise level, which is higher than a pre-set target noise in the image records; the projection records are reconstructed with the eliminated noise to form three-dimensional image records; the noise level of the image records is determined; the noise level is evaluated for the high-dose scanning image records; the high-dose scanning noise is related to the low-dose scanning noise, and the noise is eliminated in the three-dimensional image records within the image on the basis of the relation.

EFFECT: improvement of the quality of the three-dimensional images.

11 cl, 4 dwg

Description

DETAILED DESCRIPTION OF THE INVENTION

The following generally relates to computed tomography (CT) and finds particular application in the detection and reconstruction of CT data, and more particularly in CT at low doses.

CT scanners emit ionizing radiation that can cause damage to living tissue, leading to an increased risk of cancer, tumors and genetic damage at typical doses, and can cause skin burns and / or hair loss at high doses. For example, the literature indicates that a CT study with an effective dose of ten (10) millisievert (mSv) may be associated with an increase in the likelihood of cancer with a fatal outcome by approximately one (1) chance out of two thousand (2000).

Various approaches have been proposed in the literature in order to reduce the time of exposure of a patient to ionizing radiation (i.e., to reduce the dose for a patient) during a CT study. One approach suggested in the literature is to use ultra-low dose technology for CT studies of less than mSv or fractions of mSv. Unfortunately, technology at ultra-low doses in fractions of mSv can lead to CT images of non-diagnostic quality with high noise levels (ie low signal-to-noise ratio) and artifacts in the form of noise in the figure.

To turn such images into clinically acceptable images, a significant noise reduction is required, equivalent to a dose increase of the order of one thousand percent (1000%) or more. Unfortunately, the modern algorithms for non-iterative noise elimination described in the literature are developed with the ability to reduce the noise level equivalent to increasing the dose to about two hundred percent (200%). Thus, the resulting CT images have lower image quality (more noise) than images obtained using traditional scanning technology with an effective dose of 10 mSv. Iterative methods require a long reconstruction time and for the active elimination of noise can lead to unnatural patterns of noise in images.

In view of the foregoing, there is an unresolved need for other technologies at ultra-low doses for CT studies.

Aspects of this application resolve the above and other issues.

According to one aspect, the method includes eliminating noise in the projection data from scanning at low doses in the projection region, reconstructing the projection data with the noise eliminated to generate volumetric image data, and eliminating noise in the volumetric image data in the image region.

According to another aspect, the system includes a projection region processor that eliminates noise in projection data from scanning at low doses, a reconstruction module that reconstructs the projection data with no noise and generates image data, and an image region processor that eliminates noise in the image data, so that the noise level of the image data with the eliminated noise is essentially identical to the estimated noise level in the image data for image data when scanning at a large dose x

According to another aspect, a computer-readable storage medium encoded using instructions that, when executed by a computer processor, instruct the processor to: use noise elimination in the projection area and image area to generate image data from scanning at low doses that have a noise level substantially identical scanning noise at high doses.

The invention may take the form of various components and arrangements of components, as well as various steps and arrangements of steps. The drawings are for the purpose of illustrating preferred embodiments only and should not be construed as limiting the invention.

FIG. 1 illustrates a visualization system in connection with a projection region processor and an image region processor.

FIG. 2 illustrates an exemplary projection area processor.

FIG. 3 illustrates an example image area processor.

FIG. 4 illustrates an example method.

FIG. 1 illustrates an imaging system 100, for example, a computed tomography (CT) scanner. The visualization system 100 includes a stationary gantry 102 and a rotating gantry 104, which is rotatably mounted on the stationary gantry 102. The rotating gantry 104 rotates around the study area 106 around a longitudinal axis or Z axis.

A radiation source 108, such as an x-ray tube, is supported by a rotating gantry 104 and rotates with the rotating gantry 104 and emits radiation. The radiation source 108 is configured to perform scans at large and small doses.

A radiation-sensitive detector grating 110, located opposite the source 108, determines the radiation that passes through the study area 106, and generates projection data indicating it. The radiation-sensitive detector array 110 may include one or more rows of radiation-sensitive pixel elements.

A support 112, such as a couch, supports the object in the study area 106. The support 112 can be used to position the object differently relative to the X, Y, and / or Z axes before, during, and / or after scanning.

A general purpose computing system serves as an operator console 114, which includes human-readable output devices, such as a display and / or printer, and input devices, such as a keyboard and / or mouse. The software, which is permanently located in console 114, enables the operator to control the operation of system 100, for example, by allowing the operator to select scanning technology at low doses (e.g., fractions of mSv), a processing algorithm in the projection area, a reconstruction algorithm, a processing algorithm in the image area etc.

The projection region processor 116 processes the projection data generated by the radiation-sensitive detector array 110, the reconstruction module 118 reconstructs the projection data processed by the projection region processor 116, and generates volumetric image data indicating the region of interest 106, and the image region processor 120 processes the volumetric data images formed by reconstruction module 118.

As described in more detail below, with respect to scans at low doses, proper processing in the projection area includes adaptively eliminating noise in the projection data, and proper processing in the image area includes eliminating noise in the image data, so that the noise level of the image with the eliminated noise matches the level of noise that can be obtained by performing targeted scanning at high doses instead of scanning at low doses.

This elimination of noise in the projection and image areas makes it possible to scan at dose levels in the fraction of mSv (ultra-low) when generating images with noise levels comparable to scans at high doses (for example, in mSv and higher). Such scans are ideal for procedures such as brain scans, screening scans for quantifying calcium, pediatric scans and / or other scans. Such scans can also shorten, reduce and / or replace preliminary scans.

It will be appreciated that the projection region processor 116 and / or the image region processor 120 may be part of the reconstruction module 118. In addition, one or more of the projection region processor 116, reconstruction module 118, or image region processor 120 may be excluded from system 100, for example, in a computing system such as a workstation or the like. In addition, one or more processors may execute machine-readable instructions encoded and / or implemented on a local or remote computer-readable storage medium, such as a storage device, to implement one or more of a projection area processor 116, reconstruction module 118, or image area processor 120 .

FIG. 2 illustrates an exemplary projection area processor.

Input processor 116 is a projection reading describing the linear integrals of the attenuation coefficients of the x-ray radiation through the scanned patient. For detections that include a dynamic focal spot, the projection region processor 116 may individually process the readings of each different focal spot or, alternatively, process them together after interleaving with each other.

The first parameter determination unit 202 determines a first parameter of the projection data. In the illustrated embodiment, the first parameter includes the noise variance of the projection data. Second parameter determination module 204 determines a second parameter of projection data. In the illustrated embodiment, the second parameter includes a change in projection data, which can be estimated according to the scalar product of the gradients or determined differently.

Weighting generator 206 generates noise elimination weights based at least on the basis of the first and second parameters and one or more algorithms 208. As an example, if the first and second parameters indicate that the projection data includes lower local noise variance and higher local variation, weight generator 206 uses an algorithm to generate a weight coefficient that decreases faster with the distance between adjacent projection readings and Kazan, which is to be interpolated. In another example, if the first and second parameters indicate that the projection data includes higher local noise dispersion and lower local variation, the weight generator 206 uses an algorithm to generate a weight coefficient that decreases more slowly with the distance between adjacent readings according to the projection and indication to be interpolated.

The projection noise elimination module 210 applies weights to eliminate noise in the projection data. In one case, this actually leads to the replacement of each projection by the weighted sum of the initial readings in its three-dimensional environment. In general, this leads to the active elimination of projection noise with higher noise, which does not include the structure of interest, the intermediate exclusion of projection noise with higher noise, which include the structure of interest, and a small exclusion of projection noise with lower noise, which include the structure of interest. By adaptively eliminating noise in the projection as such, blurring of image details for a given noise elimination in images can be reduced.

In one embodiment, the noise elimination module 210 in the projection area applies weights, as shown in equation 1:

equation 1

представляет интерполированное показание с исключенными шумами,

представляет зашумленные данные проекции в окружении показания i, а

представляет адаптивные весовые коэффициенты. Это, по существу, заменяет каждое исходное показание по проекции взвешенной суммой исходных показаний по проекции в предварительно определенном окружении относительно исходного показания по проекции.Where

represents an interpolated reading with no noise,

represents noisy projection data surrounded by readings i, and

represents adaptive weights. This essentially replaces each projection original reading with the weighted sum of the projection original readings in a predetermined environment relative to the original projection reading.

могут быть определены на основе уравнения 2:Adaptive Weights

can be determined based on equation 2:

equation 2

представляет ненормализованные весовые коэффициенты взвешенной суммы.Where

represents abnormalized weighted weighting factors.

может быть определен на основе уравнения 3:In a non-limiting embodiment, the parameter

can be determined based on equation 3:

equation 3

представляет расстояние между соседом

и показанием

, а

является параметром, который определяет активность сглаживания.Where

represents the distance between a neighbor

and indication

, but

is a parameter that determines smoothing activity.

может быть определен на основе уравнения 4:In a non-limiting embodiment, the parameter

can be determined based on equation 4:

equation 4

задает минимальное снижение весовых коэффициентов с расстоянием,

представляет оцененное локальное изменение,

представляет среднее значение локальных изменений, вычисленных в различных областях в области проекции,

представляет дисперсию шума,

представляет среднее значение из значений локальной дисперсии шума, вычисленных в различных областях в области проекции.Where

sets the minimum reduction in weights with distance,

represents the estimated local change,

represents the average value of local changes calculated in various areas in the projection area,

represents noise variance

represents the average of local noise variance values calculated in various areas in the projection area.

и

назначаются значения, которые максимизируют изменение

между различными местоположениями в области проекции. В отличие от

, параметр

является независимым от

и

. Это дает возможность управления активностью исключения шумов и одновременного поддержания высокой зависимости весовых коэффициентов от

и

.To minimize blurring of diagnostic parts,

and

assigned values that maximize change

between different locations in the projection area. Unlike

, parameter

is independent of

and

. This makes it possible to control the activity of eliminating noise and at the same time maintaining a high dependence of weight coefficients on

and

.

FIG. 3 illustrates an example image area processor 120.

The noise variance determination module 304 in the projection area determines the noise variance for low projection scan projection data. The noise variance estimation module 306 in the projection data estimates the noise variance for the projection data that should be generated if a targeted scan is performed at high doses.

Shaper 302 noise exclusion coefficients generates a noise exclusion coefficient. The illustrated noise elimination factor generator 302 allows for crosstalk of the detector.

The projection variance estimation module 308 estimates the projection variance for scanning at low doses based on the noise variance estimated by 304 and the noise elimination coefficient.

The image noise estimation module 310 estimates noise in images for scanning at low doses based on the corresponding projection variance and noise in images for scanning at high doses based on the corresponding projection variance.

The image area noise elimination unit 312 eliminates noise in image data based on noise in images for scanning at low doses and noise in images for a larger dose.

In one embodiment, the above noise reduction coefficient may be generated by the noise reduction coefficient generator 302, as shown in equation 5.

equation 5

представляет коэффициент исключения шумов.Where

represents the noise reduction factor.

Projection variances for scanning at low doses and scanning at high doses, respectively, can be estimated using the estimated variance of the projection readings and noise elimination factors through equations 6 and 7:

equation 6

и

and

equation 7

представляет дисперсию проекции для сканирования при малых дозах,

представляет локальную дисперсию шума сканирования при малых дозах,

представляет дисперсию проекции для сканирования при больших дозах и

представляет локальную дисперсию шума для сканирования при больших дозах.Where

represents the variance of the projection for scanning at low doses,

represents the local dispersion of scanning noise at low doses,

represents the variance of the projection for scanning at high doses and

represents local noise variance for scanning at high doses.

Projection-dependent scalar dispersion values can be generated for each projection by summing over their subsets, for example, for a predefined group (e.g. 150, 200, 300, etc.) of the central readings belonging to the center line of the projection.

For an aperture-weighted filtered rear projection, the noise variance values in the center of the portal rotation obtained for scanning image data at low doses with excluded noise in the projection and for scanning image data at high doses can be determined using equations 8 and 9, respectively:

equation 8

и

and

equation 9

является константой для данного набора параметров реконструкции, а I обозначает различные осевые изображения.Where

is a constant for a given set of reconstruction parameters, and I denotes various axial images.

The noise elimination in the image area used by the noise elimination module 312 in the image region can be determined as shown in equation 10:

equation 10

представляет исключение шумов в области изображений и передается в комплементарный алгоритм исключения шумов в области изображений.Where

represents the elimination of noise in the image area and is transmitted to the complementary algorithm for eliminating noise in the image area.

в качестве ввода модуль 312 исключения шумов в области изображений исключает шумы в данных изображений, так что уровень шума каждого осевого изображения, реконструированного из проекций с исключенными шумами, совпадает с уровнем шума, который должен быть получен посредством выполнения сканирования при больших дозах вместо сканирования при сверхмалых дозах.Using

as input, the noise elimination module 312 in the image area eliminates noise in the image data, so that the noise level of each axial image reconstructed from the projections with the noise eliminated coincides with the noise level that should be obtained by scanning at high doses instead of scanning at ultra-low doses.

FIG. 4 illustrates a method for forming images with noise levels of scans at high doses from scans at low doses.

At 402, scanning at low doses (for example, 10 mCl) is performed for the object. As described herein, such a scan may correspond to a scan with an effective dose of a fraction of mSv.

At 403, the linear integrals of the x-ray attenuation coefficients are calculated in the projection area.

At 404, the projection data of the linear integrals are eliminated by noise. As described herein, the proper elimination of noise in the projection area includes adaptively eliminating noise in the projection data such that projection data having lower local noise dispersion and higher local variation are less excluded from noise than are eliminated noise projection data having a higher local noise variance and lower local variation.

At 406, noise-eliminated projection data is reconstructed to form image data.

At 408, image data is excluded from noise. As described herein, the proper elimination of noise in the image area includes the elimination of noise in the data such that the noise level of the image data with the noise canceled is essentially the same as the noise level of the targeted scan at high doses (for example, 100 μl).

The above steps can be implemented by machine-readable instructions, which, when executed by the computer processor (s), instruct the processor to perform the steps described herein. In such a case, instructions are stored on a computer-readable storage medium, such as a storage device, associated and / or otherwise accessible to a relevant computer.

The invention has been described with reference to preferred embodiments. Obviously, after reading and understanding the above detailed description, modifications and changes may be made. The invention should be construed as including all such modifications and changes to the extent that they fall within the scope of the attached claims or their equivalents.

Claims (11)

1. A method of reconstructing image data of a computed tomography, comprising the steps of:
exclude noise in projection data from scanning at low doses in the projection area, while scanning at low doses corresponds to scanning that generates image data with a noise level that is higher than a predetermined target noise level in the image data;
reconstructing the projection data with the eliminated noise to form volumetric image data;
determine the noise level of the image data;
Estimating noise levels for high-dose scan image data
determine the ratio of the scanning noise level at high doses to the noise level of the scanning image at low doses and
eliminate noise in the data of volumetric images in the image area based on the ratio. 2. The method of claim 1, wherein the noise-free image data has a noise level substantially identical to a predetermined target noise level in the image data. 3. The method according to any one of paragraphs. 1-2, in which scanning at low doses has an effective dose of a fraction of mSv, and scanning at high doses has an effective dose of ten mSv or higher. 4. The method according to any one of paragraphs. 1-2, additionally containing phase, in which:
- adaptively eliminate noise in projection data based on local noise variance and local variation. 5. The method according to claim 4, in which projection data having a lower local noise dispersion and a higher local variation are less excluded from noise than projection data having a higher local noise variance and a lower local variation . 6. The method according to any one of paragraphs. 1-2, in which the exclusion of noise in the projection area is an iterative exclusion of noise. 7. The method according to any one of paragraphs. 1-2, in which eliminating noise in the projection area reduces blurring of image details. 8. A system for reconstructing computed tomography image data, comprising:
a projection region processor (116) that eliminates noise in projection data from scanning at low doses;
reconstruction module (118), which reconstructs the projection data with the eliminated noise and generates image data; and
an image region processor (120) that eliminates noise in the image data so that the noise level of the image data with the eliminated noise is substantially identical to the estimated noise level in the image data for image data when scanning at high doses, the image region processor (120) configured to determine a noise level of image data; Estimate noise levels for high-dose scan image data determine the ratio of the scanning noise level at high doses to the noise level of the scanning image at low doses and exclude noise in the volumetric image data in the image area based on the ratio. 9. The system of claim 8, further comprising:
- shaper (302) of noise exclusion coefficients, which determines the reduction in noise variance of projection data for scanning at low doses due to noise exclusion;
a module (304) for estimating the noise variance in the projection area for low-dose scans, which estimates the noise variance of the projection data for low-dose scans before eliminating noise;
- a projection variance estimation module (308) that estimates projection variance for scanning at low doses based on a noise variance determined by a noise variance estimation module and a noise elimination coefficient determined by a noise exclusion coefficient determination module;
a module (306) for estimating the variance of projection noise in the projection area for scans at high doses, which estimates the variance of the projection for scanning at high doses;
a module (310) for estimating noise in images, which estimates the variance of noise in images for scanning at low doses with excluded noise in the projection and for scanning at high doses; and
a noise elimination unit (312) in the image area eliminates noise in image data based on noise variance in images for scans at lower and higher doses. 10. The system according to claim 8, in which the noise exclusion coefficient takes into account the crosstalk of the detector. 11. The system according to any one of paragraphs. 8-9, in which the noise exclusion coefficient is formed for a subset of projections.

Images