DE102012024961A1 - Shaping filter for dose optimization of x-ray device for generation of sectional images of object for computer tomography, has single solid unit comprising two materials, and angle-dependent filters realizing beam parameter correction - Google Patents

Abstract

The filter (59) has a single solid unit that comprises two materials, where composition of the two materials comprises homogenous material mixture and heterogeneous structured material mixture. The filter optimizes image quality relevant parameters such as pixel noise, scattered radiation intensity or attenuation value representation. An angle-dependent filter realizes beam parameter correction for marker making. The angle-dependent filter realizes beam parameter correction for the cut image formation that is dependent on the cone angle.

Description

Technical area

The invention relates to a method for determining a shape filter for computed tomography (CT).

State of the art

In the patent DE 10 2009 053 523 B4 describes a shape filter for CT devices to optimize dose efficiency. Dose efficiency is the relationship between the dose administered to the patient and the dose contributing to the image quality during reconstruction by weighting.

Due to the fact that during reconstruction the outer detector rows are weighted lower compared to the central detector rows, the dose efficiency is lower in this range. The described shape filter, which in positive and negative z-direction (CT rotation axis), d. H. towards the edge of the filter (peripheral area), increasing in thickness, reduces the dose in this area and thus contributes to improving the overall dose efficiency in the reconstruction. The dose efficiency in the peripheral detector area is not directly improved, but the ratio to the central detector lines is reduced, thus increasing the dose efficiency of the entire examination. The shape of the filter depends on the weighting function of the detector lines used in the reconstruction and is independent of the shape and size of the patient.

In the patent DE 10 2009 012 631 B4 a shape filter is described for optimizing the dose in the patient. This filter is shaped in the fan beam direction and constant in the z direction. Its shape is adapted so that in a CT scan of a standard object of predetermined geometry, in this a homogeneous dose distribution is achieved.

description

In the method presented here for producing a shape filter, the goal is to obtain a defined or homogeneous noise and / or a defined or homogeneous scattered radiation intensity and / or attenuation value representation in the reconstructed image.

By way of example, this is illustrated in the description on the basis of the parameter Noise in the reconstructed image, without restricting the actual concept of the invention. The usual approach to achieve this goal is to ensure a homogeneous noise distribution at the detector. However, this does not result in evenly distributed noise in the CT images since different weightings are used due to the Filtered Back Projection (FBP) commonly used in reconstruction. In this case, peripheral detector elements are weighted less than central, which increases the noise in the peripheral regions of the CT image. In this method, starting from a conventional homogeneous noise form filter at the detector, a shape filter is optimized for homogeneous noise in the CT images by means of an FBP correction factor derived from the reconstruction. This beam angle dependent correction factor reduces the thickness of the filter determined in the first step to increase the intensity at the detector and thereby compensate for the nonuniform weighting of the detector elements in the reconstruction.

The filter also consists of one unit, which prevents artefacts caused by edges. This unit can be made of one, two or more materials which consist of a homogeneous and / or heterogeneously structured material mixture. It can be used in the usual materials for X-ray filtering materials such. As aluminum, copper, carbon, polytetrafluoroethylene or breast-equivalent material.

The filter is uniform, d. H. The filter consists of a single fixed assembly and not of several sub-assemblies, which can be moved against each other. The filter assembly itself may consist of a monobloc of a filter material or a combination of different filter materials.

The filter materials may consist of a homogeneous and / or heterogeneously structured material mixture. For the filter suitable materials are for. As aluminum, copper, carbon, polytetrafluoroethylene or breast-equivalent material.

For the design of the shape filter according to the invention, a CT geometry with a focus-isocenter distance (FIA) and an isocenter-detector distance (IDA) is assumed. The filter is placed in a focus-to-filter distance (FFA) and the cylindrical homogeneous phantom (phantom is understood in the description to be synonymous with any type of calibration and examination object, such as the human breast) Radius R in the isocenter. The photon flux density of the X-ray beam at the detector N _detector (θ) is attenuated by the shape _filter thickness d _filter (θ) and by the distance traveled in the phantom d _phantom (θ). N _detector (θ) is determined with equation (1) as a function of the fan angle θ.

The distance traveled by the X-ray in the phantom is calculated as a function of the radius and of the FIA (2).

The spectrum of the X-ray source is calculated taking into account the X-ray tube voltage, the corresponding pre-filtering of the X-ray tube and the anode angle of the X-ray tube anode.

In order to weaken the X-ray accordingly, the filter thickness is calculated as a function of the fan angle θ and the distance traveled in the phantom d _phantom (θ). The aim is to reduce the inhomogeneous noise in the reconstructed CT images, which results from the fan angle-dependent photon flux density at the detector. Typically, without the use of a shape filter, there will be a higher photon density at the edge of the detector compared to the center, which in turn will result in less noise. This noise inhomogeneity is also reflected in the reconstructed images. Therefore, to obtain homogeneous noise in the CT images, the angle-dependent noise is matched at the detector (3). For an ideal detector, the noise is based on a Poisson distribution of the incident number of photons N _detector (θ) (1). With the help of a filter thickness adapted to each angle, it is possible to achieve a homogeneously distributed noise at the detector.

However, due to conical beam reconstruction with ordinary filtered backprojection (FBP), no uniformly distributed noise can be obtained in the reconstructed images. This results from a fixed cutoff frequency of the convolution kernel in the reconstruction, which is used for all volume elements. The high-frequency noise components in the periphery of the CT image are consequently amplified unevenly as compared with the central region. This leads to an increase in the noise to the edge of the image. For this reason, according to the invention, an angle-dependent correction factor W _FBP (θ) is derived from the reconstruction in order to improve the filter design and obtain a homogeneous noise in the reconstructed images.

The following calculations are based on a simple CT geometry, which rotates with the rotation angle α around the origin of a Cartesian coordinate system. The 2 shows an x-ray beam which, starting from the source, interacts with a voxel element v (x, y) and impinges on the detector p '(x, y, α).

V(p) = cos( p / FDA) (6) p'(x, y, α) = FDA·θ(x, y, α) (7)

The FBP algorithm used in the reconstruction is shown in equation (4) where R (p ', α) is the projection data and h (p' - p) is the convolution kernel. U (x, y, α) and V (p) are the spatial weighting factors for each reconstructed volume element as a function of focus-to-detector distance (FDA = FIA + IDA) (5), (6). The position of the X-ray on the curved detector p '(x, y, α) is calculated by the radian of a circle with FIA as radius (7). The fan angle θ can be expressed by means of a coordinate transformation by the angle α with x and y (8).

V (p) = cos (p / FDA) (6) p '(x, y, α) = FDA · θ (x, y, α) (7)

The inhomogeneous noise is caused by the weighting factor U (x, y, α) of the FBP. Especially for small FIA and large fan angles, the factor U (x, y, α) ^{-2 increases} for peripheral volume elements and in turn leads to increased noise in these areas of the CT image. This unevenness should be compensated by the convolution kernel h (p '- p), with the help of a lower cutoff frequency for peripheral or outlying image regions. Instead, in the prior art, a fixed cut-off frequency is used for all volume elements, which leads to an increase in the high-frequency noise components in the periphery and thus also more noise. This irregularity can be approximated by the function F _FBP (θ) depending on the fan angle. In F _FBP (θ), the factor U (x, y, α) ^{-2 is} averaged over all volume elements in the x- and y-directions (N _x , N _y ) and rotation angle α (N _α ) Angle θ lie. F _FBP (θ) is normalized to the central beam and used as the correction factor W _FBP (θ) (10) for the shape filter design. Due to the inventive correction factor, the thickness of the shape filter in the periphery is reduced (11) relative to the original design, which in turn leads to less noise at the detector and thus to a more homogeneous noise distribution in the reconstructed CT image.

For determining the angle-dependent filter thicknesses, N _detector (θ) is calculated for each fan angle (11) and compared with the central beam. The shape _filter thickness d _filter (θ) is iteratively increased to compensate for the lower attenuation in the outer region of the phantom (12) and to meet the condition of the filter design (3). This is repeated for each angle, with a suitable step size, preferably of 0.1 °, until the maximum fan angle for the respective phantom has been reached. This is done for each phantom diameter.

For the final production of the shape filter, the angle-dependent thickness of cylindrical coordinates must be converted into Cartesian coordinates (13), (14). Since the radius of the phantom does not depend on z, it is assumed for simplicity that the filter is constant in the z-direction. x = SFD * tan (θ) + d _filter (θ) * sin (θ) (13) y = d _filter (θ) · cos (θ) (14)

For clinical CT applications, for example, for a breast CT system, it is not possible to use the optimal Fromfilter individually for each phantom diameter due to spatial limitations. One way to solve this problem is to use only a shape filter with a variable FFA, which is determined separately for each phantom diameter.

With a modifiable FFA and a phantom-adapted shape filter, an approximately homogeneous noise distribution is achieved in the reconstructed CT images for all phantom diameters.

The proposed filter design method shows a clear homogenization of the noise in the CT images. It is a new method to adapt the filter based on the noise at the detector from the widely used method to obtain significantly improved CT images using the FBP correction factor according to the invention.

Description of the drawings

1 : Schematic drawing of a CT system with a path of an optical path at an angle θ.

2 : Simple CT geometry for deriving the FBP correction factor

3a : In 3a the detector noise is shown for a phantom diameter of 140 mm, simulated without form filter, with form filter without FBP correction factor and with shape filter with FBP correction factor. The simulation with uncorrected shape filter leads to a homogeneous detector noise as the design condition shows.

3b : In 3b It can be seen that this type of uncorrected shape filter results in increased noise in the periphery as compared to the center of the reconstructed CT image. For the results with shape filter, which was created with FBP correction factor, the noise distribution in the reconstructed images was adjusted.

LIST OF REFERENCE NUMBERS

14

detector

15

X-ray tube

31

chest

57

collimator

59

shaping filter

170

Focus Filter Distance FFA or SFD

171

Focus Isocenter Distance FIA or SID

172

Isocenter detector distance IDA or IDD

173

ISO-Zentrum

180

Radius phantom or object

181

Diameter phantom or object

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• DE 102009053523 B4 [0002]
• DE 102009012631 B4 [0004]

Claims (7)

Form filter for dose optimization of an X-ray machine for generating sectional images of an object based on intensity attenuation of the X-ray, offset projection images, which is designed with respect to the correction of at least one of the object properties such as geometry or intensity attenuation of the object, characterized in that at least one of the image filter for the image quality relevant parameters such as pixel noise, scattered radiation intensity or attenuation value representation for the sectional image and not for the projection is optimized. Mold filter according to claim 1, characterized in that in the form filter at least one beam angle-dependent correction parameter for the sectional image position is realized. Mold filter according to one of the preceding claims, characterized in that at least one beam angle-dependent correction parameter for the sectional image position is realized in the form filter, which is dependent on the fan angle. Mold filter according to one of the preceding claims, characterized in that at least one beam angle-dependent correction parameter for the sectional image position is realized in the form filter, which is dependent on the cone angle. Mold filter according to one of the preceding claims, characterized in that it contains in its composition at least two materials. Mold filter according to claim 5, characterized in that it contains in its composition at least one homogeneous material mixture. Mold filter according to claim 5, characterized in that it contains in its composition at least one heterogeneously structured material mixture.

Images