DE102009015772A1 - Method for reconstructing data sets of computerized tomography-image of e.g. human, from projection data sets, involves tomographically representing determined partial density of examination object - Google Patents

Abstract

The method involves calculating a local measurement spectrum-based weighting function at a location of an examination object i.e. patient (7), based on recently known material distribution in the examination object with a material-specific energy-based attenuation function. Material analysis is implemented for a basic material e.g. contrast agent, such that material-specific local partial density of the examination object is determined based on the weighting function and the attenuation function. The determined partial density of the examination object is tomographically represented. An independent claim is also included for a computerized tomography system for scanning an examination object.

Description

• – CT-Scan eines Untersuchungsobjektes und Erzeugung der N Projektionsdatensätzen, gemessen mit jeweils unterschiedlicher Spektralwichtung,
• – Rekonstruktion von N CT-Bilddatensätzen, jeweils mit einer Vielzahl von lokalen Bildwerten, die lokale mittlere Schwächungswerte am Ort des Untersuchungsobjektes und eines Messfeldes in Abhängigkeit von den jeweils zum Scan genutzten energieabhängigen Spektralwichtungen wiedergeben.

The invention relates to a method for reconstructing CT image data sets from N ≥ 2 projection data records, recorded with N different spectral weights, comprising the method steps:

• CT scan of an examination object and generation of the N projection data sets, measured in each case with different spectral weighting,
• - Reconstruction of N CT image data sets, each with a plurality of local image values representing local mean attenuation values at the location of the examination subject and a measurement field as a function of the energy-dependent spectral weights used for the scan.

According to the invention, the term spectral weighting is understood to mean the product function S _j (E) .D _j (E) with j = 1 to N, which is energy-specifically the product of the energy-dependent weighting of the radiation spectrum S _j (E) used and the spectral sensitivity of a detector used D _j (E) represents. The index j stands for a different spectral weighting. It is thus possible, with the aid of different spectra and detectors of the same spectral sensitivity or with the aid of the same radiation spectra but using different spectral detector sensitivities or also with simultaneous change of the spectral sensitivity of the detector and with different radiation spectra, to obtain respectively spectrally differently weighted attenuation values.

The above designated method and also CT systems for implementation This method is well known. For example, you can using so-called dual-energy CT systems using of two different X-ray spectra of two tomographic image data sets for the same examination object be reconstructed, each containing the local attenuation values, more precisely the local attenuation coefficients, reproduce by the particular radiation spectrum used and affects the spectral sensitivity of the detector used become.

For the evaluation of the measured data, the so-called base material decomposition is usually used, as described, for example, in US Pat PHYS. MED. BIOL., 1976, VOL. 21, NO. 5, 733-744, "Energyselective Reconstructions in X-ray Computerized Tomography, RE Alvarez and A. Macovski , is described. Here, two image data sets of the concentrations of base materials are generated. As typical base materials, for example, iodine and water or bone and water are used.

Also, on the basis of such tomographic image data sets, a RhoZ projection can be performed, as described in the document DE 101 43 131 B4 is described. After this, the local density and ordinal number are calculated on the basis of the energy-specific attenuation values which were reconstructed by the tomographic image data sets.

at such image-based spectral methods is the beam hardening an important problem. Described descriptively, the move Absorber in the beam path of a projection, the energy weight to higher Energies down. The absorption coefficient μ of it affected data sections are subsequently underestimated. Calculated therefrom sizes such as concentrations or Density and atomic numbers show corresponding systematic results Error on.

Want To avoid these errors, the beam hardening of the the matter penetrating radiation are taken into account. So far, beam hardening correction and spectral algorithm separate procedures. For the normal monoenergetic CT become so-called multiple correction methods for beam hardening correction used. Their goal is to reconstruct local attenuation values with a spatially constant energy weight, where in the picture in medical images a smooth soft tissue impression should arise. This also affects the quality a material decomposition of a tomographic representation.

It is therefore an object of the invention, a method for reconstruction of CT image data sets where the image data sets not by a beam hardening or their corrective measures be influenced and a consequent material decomposition unaffected of spectral influences remains.

These The object is achieved by the features of the independent claims solved. Advantageous developments of the invention are Subject of subordinate claims.

The inventor has recognized that the beam hardening in its influence on the reconstructed image values of tomographic image data sets, their pixels or voxels attenuation coefficients of the examined Represent object or at least indirectly in the form of CT numbers reflect, depending on the used spectrum of the scanning radiation and / or also the spectral sensitivity of the detector used. However, it is possible to use the CT-reconstructed local mean, ie spectrally weighted, attenuation values μ _j (r →) as a function of a local spectral weighting function Ω _j (E, r →) and a non-spectrometer-dependent, ie a local energy-dependent attenuation function μ (E, r →) to describe.

This makes it possible to describe the problem of beam hardening by means of a slowly varying, fault-tolerant and, moreover, easily or quickly calculated function. This results in a direct and clear relationship between the conventionally reconstructed local mean attenuation values μ _j (r →) and the local energy-dependent attenuation function μ (E, r →) to be characterized by the spectral methods:

According to the invention now these local energy-dependent and measurement-spectrum-independent attenuation functions μ (E, r →) used to perform a material decomposition, here no disturbing influences by possible Beam hardening are given.

In accordance with this basic idea, the inventor proposes to improve the method known per se for the reconstruction of CT image data sets from N ≥ 2 projection data records recorded with N different spectral weights S _j (E) * D _j (E) with j = 1 to N, wherein the spectral weighting is defined as the product function of the energy-dependent weighting of the used radiation spectrum S _j (E) and the spectral sensitivity of a detector D _j (E) used.

In this method, it is known to carry out a CT scan of an examination object and to generate the N projection data sets with respectively different spectral weighting S _j (E) .D _j (E) and reconstruction of N CT image data sets each having a multiplicity of local image values, the local ones mean attenuation values μ _j (r →) at the location r → of the examination subject and a measuring field as a function of the energy-dependent spectral weights S _j (E) · D _j (E) used in each case for the scan.

• – eine Ermittlung einer ersten groben Materialverteilung mit M ≥ N vorbestimmten Basismaterialien im Untersuchungsobjekt auf der Basis der zuvor ermittelten lokalen mittleren Schwächungswerte μ _j(r →),
• – Bestimmung erster lokaler energieabhängiger Schwächungsfunktionen μ(E, r →) im Untersuchungsobjekt auf der Basis der groben Materialverteilung und der bekannten Zusammensetzung der Basismaterialien,
• – Berechnung von lokalen messspektrumabhängigen Wichtungsfunktionen Ω_j(E, r →) an den Orten r → des Untersuchungsobjektes auf der Basis der zuletzt bekannten Materialverteilung im Untersuchungsobjekt mit den dazugehörigen materialspezifischen messspektrumunabhängigen Schwächungsfunktionen μ(E, r →),
• – Durchführung einer Materialzerlegung für M Basismaterialien zur Bestimmung von materialspezifischen örtlichen Partialdichten c_i(r →) mit i = 1 bis M des Untersuchungsobjektes unter Verwendung der zuletzt bekannten lokalen energieabhängigen Schwächungsfunktionen μ(E, r →) und den lokalen messspektrumabhängigen Wichtungsfunktionen Ω_j(E, r →), und
• – eine tomographische Darstellung der berechneten örtlichen Partialdichten c_i(r →) des Untersuchungsobjektes.

The improvement of the method according to the invention provides:

• A determination of a first coarse material distribution with M ≥ N predetermined base materials in the examination object on the basis of the previously determined local mean attenuation values μ _j (r →),
• Determination of first local energy-dependent attenuation functions μ (E, r →) in the examination object on the basis of the coarse material distribution and the known composition of the base materials,
• - Calculation of local measurement spectrum dependent weighting functions Ω _j (E, r →) at the locations r → of the examination object on the basis of the last known material distribution in the examination object with the associated material-specific measurement spectrum-independent attenuation functions μ (E, r →),
• Performing a material decomposition for M base materials for determining material-specific local partial densities c _i (r) with i = 1 to M of the object to be examined using the last known local energy-dependent attenuation functions μ (E, r) and the local measurement spectrum-dependent weighting functions Ω _j E, r →), and
• - A tomographic representation of the calculated local partial densities c _i (r →) of the examination object.

An advantageous embodiment of the method according to the invention may be that, after the determination of local partial densities _c.sub.i (r.sup.) Of the base materials of the examination object on the basis of these partial densities, the material distribution is recalculated and the previously described determination of the local energy-dependent attenuation functions is newly determined and the new local partial densities are determined more accurately, so an iterative calculation is performed until a predetermined iteration condition is met. For example, a maximum error between the iteration steps and / or a maximum number of iterations can be used as the termination condition for the iteration.

berechnet werden, wobei E die Energieabhängigkeit und r → die Ortsabhängigkeit der Schwächungsfunktion beschreiben und μ_j(E, r →) die dichtenormierte energieabhängige Schwächungsfunktion des j-ten Materials darstellt.For the iterative calculation of the local measurement spectrum-dependent weighting functions Ω _j (E, r →), the corrected attenuation functions μ (E, r →) can be determined according to

where E describes the energy dependence and r → the spatial dependence of the attenuation function and μ _j (E, r →) represents the density normalized energy-dependent attenuation function of the j-th material.

verwendet werden, mit:

wobei P dem Vorwärtsprojektionsoperator bei einer Rekonstruktion eines CT-Bilddatensatzes und R^–1 inverse Radontransformationsoperator bei einer Rekonstruktion eines CT-Bilddatensatzes entsprechen.Specifically, in this method for calculating the measurement spectrum-dependent local weighting functions Ω (E, r →) at the locations r → of the examination subject, the relationship

to be used with:

where P corresponds to the forward projection operator in a reconstruction of a CT image data set and R ^-1 inverse radon transformation operator in a reconstruction of a CT image data set.

durchgeführt werden, wobei der Index 1 bis N die Anzahl der verwendeten energieabhängigen Spektralwichtungen (S_j(E)·D_j(E)) angibt, der Index 1 bis M die Anzahl der zu bestimmenden Materialien darstellt, während E die Energieabhängigkeit und r → die Ortsabhängigkeit des Schwächungskoeffizienten beschreiben.Material decomposition can be done using the relationship:

where Index 1 to N indicates the number of energy-dependent spectral weights used (S _j (E) * D _j (E)), Index 1 to M represents the number of materials to be determined, while E represents energy dependence and r → describe the location dependence of the attenuation coefficient.

Different energy-dependent spectral weights (S _j (E) .D _j (E)) can be achieved according to the invention by using different radiation spectra. For example, different acceleration voltages and / or different filterings can be used for this purpose.

Alternatively or additionally, different energy-dependent spectral weights S _j (E) .D _j (E) can also be achieved by using different spectral detector sensitivities. For this purpose, different detectors of different spectral sensitivity can be used or energy-specific detectors with different energy sampling limits can be used.

The The method described above can be used in particular in the context of patient examinations used in these studies for the Material decomposition the knowledge about the fundamentally in the human or animal body existing base materials, such as Water or soft tissue, bone, air, implant materials or also applied contrast agent is received.

The However, procedures can also be applied to inanimate objects apply. For example, hereby objects as part of safety investigations or material testing be scanned. For this it is then advantageous as base materials previously known materials, for example from a production process, or as base materials at least one in the article sought after material, for example explosives.

To the The invention also includes a CT system for scanning an examination object, comprising a control and computing unit with a memory for computer program code, being in memory Computer program code is deposited and the computing unit in operation perform the inventive method can.

In the following the invention with the help of the figures will be described in more detail, with only the not-to-understand for the understanding of the invention features are shown. The following reference numerals and abbreviations are used:
1 : CT system; 2 : first X-ray tube; 3 : first detector; 4 : second x-ray tube (optional); 5 : second detector (Optional); 6 : Gantry housing; 6.1 : Swivel arm; 7 : Patient / phantom; 8th : Examination couch; 9 : System axis; 10 : Control and processing unit with optional additional ECG function; 11 : Contrast agent applicator; 12 : ECG lead; 13 : C-arm system; 16.1 to 16.7 : Process steps; A: weakening; D: detector; d: length of the projection path through object area; D (E): spectral sensitivity of the detector at energy E; E: energy; I: radiation intensity; I ₀ : radiation intensity without weakening; k: constant; L: projection path; M: materials; MF: measuring field; O: object; P1-P3: measuring points; Prg ₁ prg _n : computer programs; S: radiation source; S ₁ to S ₇ : method steps; S (E): weighting of the radiation spectrum at the energy E; Z: ordinal number; Δ: error value; γ: X-ray; θ: projection angle; c (r →): local partial density; r →, r → ': location coordinates; μ (E, r →): local energy-dependent and measurement-spectrum-independent attenuation function; μ (r →): local mean attenuation coefficient; Ω (E, r →): local measurement spectrum dependent weighting function. Indexing is consensus-based.

It show in detail:

1 : CT system for carrying out the method according to the invention;

2 : C-arm system for carrying out the method according to the invention;

3 : Schematic representation of the attenuation measurement of a radiation;

4 : Value curve of measurement spectrum-independent specific attenuation functions μ (E, r →) over energy (= basis functions);

5 : Course of two energy-dependent weights of radiation spectra S _j (E) on the energy;

6 : Path of the spectral sensitivity of an exemplary detector D _j (E) via the energy;

7 : Course of two energy-dependent spectral weights S _j (E) · D _j (E) each of a radiation-detector combination on the energy;

8th : Representation of a phantom with actual attenuation values;

9 : CT image of the phantom off 8th with beam hardening artifacts;

10 : Schematic representation of an object O with two penetrating X-rays γ at the projection angle θ and distance t (θ, r);

11 : Flowchart of the LSR method according to the invention;

12 : different local measurement spectrum-dependent weighting functions Ω (E, r → ₀ ) as a function of a penetrated water layer with the initial spectral weighting w (E) of the radiation spectrum used;

13 : CT representation of a phantom with 3 measurement points P1-P3;

14 : exemplary local measurement spectrum-dependent weighting functions Ω (E, r → ₀ ) at the measuring points P1-P3 of the 13 using the radiation detector combination with 140 kVp radiation plotted on the energy;

15 : Flowchart of the method according to the invention with iteration.

The 1 shows an exemplary CT system 1 with which the method according to the invention can be carried out. The CT system 1 shows a first tube / detector system with an x-ray tube 2 and an opposite detector 3 on. Optionally, this CT system 1 via a second x-ray tube 4 with an opposite detector 5 feature. Both tube / detector systems are located on a gantry in a gantry housing 6 is arranged and during the scan around a system axis 9 rotates. The patient 7 is located on a movable examination couch 8th that are either continuous or sequential along the system axis 9 through the gantry housing 6 the measuring field MF is pushed, wherein the weakening of the X-ray emitted by the X-ray tubes by the detectors is measured.

While the measurement can be patient 7 with the help of a contrast agent applicator 11 Also, a contrast agent bolus be injected so that blood vessels can be better recognized or a perfusion measurement can be performed. In cardio recordings may additionally, with the help of an ECG lead 12 , cardiac activity is measured and an ECG gated scan is performed.

The control of the CT system is carried out with the aid of a control and processing unit 10 , in which there are computer programs Prg ₁ to Prg _n , which can also carry out the method according to the invention described above. In addition, via this control and processing unit 10 also the output of image data takes place.

at this CT system with two separate x-ray sources For example, an X-ray tube with an acceleration voltage of 80 kVp and the other x-ray tube with 140 kVp operated. This results in different measurements Spectral weightings, which - with the same spectral sensitivity the detectors used - from the different X-ray spectra of the two x-ray tubes come. Alternatively, however, it is also possible in both x-ray tubes the same radiation spectrum to use, but for different spectral Sensitivity of the detectors. Also a combination Both measures are possible.

The method according to the invention can also be used in conjunction with a C-arm system 13 be used as it is in the 2 is shown. The C-arm system shown here 13 also has an x-ray tube 2 with an opposing planar detector 3 , Both systems are using a swivel arm 6.1 which is attached to a housing 6 is attached, in any position around the patient 7 to pan and scan a measurement field. The patient 7 is on a patient bed 8th additionally using a contrast agent application system 11 to inject contrast media, if necessary, to visualize blood vessels. The system is controlled by a control and processing unit 10 , which has in its memory computer programs Prg ₁ to Prg _n , which can also perform, among other things, the inventive method for image processing. Measurements with different spectral weights can be realized, for example, by the use of an energy-specific detector system.

Both previously described systems are suitable for carrying out of the method according to the invention, the bases thereof will now be described.

In 3 the schematic structure of an X-ray attenuation measurement is shown schematically. An X-ray source S is shown which emits an X-ray beam onto a detector D. At the top, the path of an X-ray γ from the source S to the detector D is free and a radiation intensity I _{0 is} measured. In the lower part of the 3 is an object O in the beam path, whereby the radiation is weakened, so that at the detector D a lower intensity I is measured.

The attenuation A = I / I ₀ is described by the attenuation function μ (E, r →) with respect to the energy E used at the location r →. Typical material examples for μ (E, r →) are in 4 shown by way of example.

The detector D thus registers the quanta coming from the object. In each case, a measurement with object and an object is performed. From the resulting intensities I and I ₀ , the attenuation A = I / I ₀ ε [0; 1]. It thus describes the relative intensity drop through the X-ray attenuation process in the object.

• a) Für monoenergetische Anfangsenergie E₀ der Röntgenquanten ist die Schwächung A^(m)
  
  Hierbei ist der monoenergetische Schwächungskoeffizient μ ^(m)(r →) gleichwertig mit μ(E, r →).
• b) Für polychromatische Röntgenquellen, was Röntgenröhren im Allgemeinen sind, wird mit einem Spektrum, genauer einer spektralen Wichtung S(E) eines Strahlungsspektrums, für die Anfangsenergie gerechnet. Zwei beispielhafte Spektren sind in der 5 für eine Röntgenröhre mit 80 kVp und eine Röntgenröhre mit 140 kVp dargestellt. Die gemessene Schwächung A berechnet sich damit gemäß
  
  mit der spektralen Gewichtungsfunktion
  

Here, a distinction must be made between monoenergetic and polychromatic X-ray sources S.

• a) For monoenergetic initial energy E _{0 of} the X-ray quanta, the weakening A ^(m)
  
  Here is the monoenergetic attenuation coefficient μ ^(m) (r →) is equivalent to μ (E, r →).
• b) For polychromatic X-ray sources, which are X-ray tubes in general, a spectrum, more precisely a spectral weighting S (E) of a radiation spectrum, is calculated for the initial energy. Two exemplary spectra are in the 5 for an 80 kVp x-ray tube and a 140 kVp x-ray tube. The measured attenuation A is thus calculated according to
  
  with the spectral weighting function
  

The detector sensitivity D (E) describes here the spectral weighting of the detector. For an ideally integrating detector one can assume that D (E) = E. However, suitable detectors for these measurements have a very complex structure and must be calculated or measured accurately. A typical sensitivity curve D (E) of a Gd ₂ O ₂ S scintillation detector, as commonly used in conjunction with commercial CT systems, is in 6 shown. The 7 shows the spectral weighting function w (E) calculated according to equation (3) with S (E) and D (E) from the 5 and 6 ,

Medical and industrial CT systems usually use X-ray tubes as the emission source. The 5 shows the typical spectra for acceleration voltages of 80 kVp and 140 kVp of a medical system. The CT measuring process for generating projection data is correctly described by the equation (2). However, in the standard CT measurements, no energy-resolved attenuation values are measured, but energy-integrated attenuation across the entire spectrum is determined. From this, no image data can be calculated that represent μ (E, r →) as image values. For this purpose, spectrally resolved sinogram data sets A _i (E) would be required.

In single-energy CT, this conflict leads to a common approximation. The radon transformation and its inverse start from the physical basics of linear X-rays to the local effective attenuation coefficients μ (r →) of the object to be reconstructed. The errors from the estimation are commonly referred to as "beam hardening artifacts". The underlying model assumes that equation (1) is an estimate of the effective μ (r →) may be used. For the X-ray radiation that passes through the object, a mean detected energy <E> is assumed. When the X-ray quanta pass through thick regions or regions of high atomic number Z, such as bone, this increases the mean energy of the detected X-ray quanta by a few keV as a function of the absorption properties. In the 12 is the hardening of the radiation spectrum and thus also the average energy of a radiation as a function of different thicknesses of water shown. The beam spectrum is thus "hardened" and consequently reconstructed μ (r →) of the mean energy <E> of the primary spectrum S (E) is underestimated.

The error in the estimate can be determined analytically. It can be shown that equation (2) can also be written as follows:

lässt sich der effektive Schwächungskoeffizient wie folgt beschreiben:

With the error term

the effective attenuation coefficient can be described as follows:

The first order error R from equation (4) leads to an overestimation of A and a consequent underestimation of the spectrally weighted attenuation coefficient μ (r →) in the reconstructed image. The 9 FIG. 16 shows a typical example of a CT representation of one in FIG 8th shown phantom in medical CT. The CT image of the phantom shows the well-known typical beam hardening artifacts of a wedge and a "tunneling effect" between the bone phantoms.

Known radiation hardening corrections mitigate these artifacts in practical single-energy CT by correcting the measured attenuation A _i . The length of the attenuating material in the beam path is determined and compensated for the resulting energy shift of the average energy. This algorithm usually gives good results in imaging soft tissue in medical CTs. However, such measures are not aimed at obtaining the quantitative absolute accuracy of reconstructed CT values.

mit dem Vorwärtsprojektionsoperator P{.} und der Abkürzung der räumlichen Pfadintegration

mit l_θ,t(α) dem Projektionspfad durch das Objekt zum Rotationswinkel θ, dem Detektorkanal t und dem Linienparameter α für die Pfadintegration.The 10 shows a standard projection geometry in a CT system. In parallel projection geometry, a ray is described by the projection angle θ and its channel distance t to the center of the CT detector. From this follows the typical projection formula

with the forward projection operator P {.} and the abbreviation of the spatial path integration

with l _{θ, t} (α) the projection path through the object to the rotation angle θ, the detector channel t and the line parameter α for the path integration.

The reconstructed local mean attenuation coefficient is displayed as μ (r →) = R -1 {P {μ (E, r →)}} (9) with the inverse radon transformation operator R ^-1 {.}.

in die Gleichung (9) und unter Berücksichtigung der Bedingung, dass P{μ(E, r →)} unabhängig von E ist, erhält man die Gleichung

By inserting a factor

in equation (9), and taking into account the condition that P {μ (E, r →)} is independent of E, the equation is obtained

führt zuThe expansion of equation (10) by the factor

leads to

mit der lokalen Wichtungsfunktion

This expression can be transformed into

with the local weighting function

Equation (11) thus combines the spectral attenuation function μ (E, r →) with the measured, weighted attenuation coefficient μ (r →). This is followed by an image data based description of the measurement method in the CT and the reconstruction process.

The local weighting function Ω (E, r →) from equation (13) describes the effective spectral weighting at any object position. It depends on the spectral attenuation function of the scanned object μ (E, r →), the image data reconstruction process R ^-1 {.} And the measurement process described by the operator P {.}. The weighting function w (E) results from the weighting function of the system in equation (3).

In the practical application of CT only weighted attenuation coefficients μ (r →) measured. In order to calculate the local weighting function, an estimate of the spectral attenuation functions μ (E, r →) depending on the input data is needed. A dual energy scan results in two attenuation data sets μ ₁ (r →) and μ ₂ (r →). The base material decomposition is applied in an image data set based form to obtain the parameterization of μ (E, r →). This approach is described by

wobei die Elemente der Matrix K ausgedrückt werden durchThis expression separates the energy-dependent basis function f _j (E) from the location-dependent coefficients c _j (r →). The typical values of the basic function in the medical application of CT is a combination of the mass-dependent attenuation functions of water and bone. From equation (12) follows

wherein the elements of the matrix K are expressed by

The equation (15) is easily solvable with the coefficients c _j (r →) by inverting the matrix K. However, the circular dependencies K → _cj (r →) → μ (E, r →) → Ω _i (E, r →) → K must also be resolved. This leads to a two-phase iterative method according to the flowchart in FIG 11 , Here, first in method step S _1, the initial local weighting function is estimated by Ω (K = 0) i (E, r →) = w (E) is set. For example, this may be the result of the knowledge of the initially present CT display with the measured and reconstructed average local attenuation coefficients μ (r →) are estimated in method step S ₂ . From the two results of steps S1 and S2, a first estimation of the local mass attenuation functions μ (E, r →) takes place in step S ₃ . This results in the calculation of the local weighting functions Ω (E, r →) in step S ₄ . In addition, from the knowledge of the local mass attenuation functions μ (E, r →) and a predetermined desired spectrum in step S ₇ according to the invention, the local weighted attenuation coefficients of the object under consideration with respect to the arbitrary assumed spectrum can be calculated and output as a CT representation. If the method for optimization is to be carried out iteratively, these steps can be repeated as often as desired according to the flow diagram shown after a decision in step S ₅ , wherein the last calculated local weighting function is output in step S ₆ . In this way, at every k th iteration step, the estimation of μ (K) j (E, r →) and Ω (K) j (E, r →).

This method is referred to herein as Local Spectral Reconstruction (LSR). It leads to an estimation of both the local weighting function and the attenuation functions μ (E, r →). The LSR method can be transferred to multichannel spectral CT and base material decomposition, for example, with a number of N> 2 spectral channels and a number M <= N base materials. For single-energy CT, μ (E, r →) can be calculated from an energy-weighted effective attenuation coefficient μ (r →) can be estimated.

Out the LSR method according to the equations (12) and (13) gives an estimate of the local weighting function and the actual accuracy of the spectral attenuation functions μ (E, r →) of the Object. The information from the local weighting function gives a deeper understanding of the energy weighting process in CT imaging. For the quantitative application of Spectral CT are the resulting attenuation functions μ (E, r →) the main result. The associated parameters such as base material coefficients can be graphed or for specific Diagnosis questions are analyzed.

It it is important to note that the attenuation function μ (E, r →) theoretically of the weakening effects of the object, the Properties of the reconstruction and the measuring method not affected become. For example, so can the core of the reconstruction of dual-energy measurements be taken into account. This allows pixel registration between the records.

In practice, exact descriptions of the weighting function w (E) and the measurement operator P {.} Are necessary to ensure quantitative results. It should be noted that the reconstruction operator R ^-1 {.} Yields both the image reconstruction and the calculation of the local weighting function. That being said, there are no further effects on the accuracy and precision of μ (E, r →).

Basically, there are several different objectives of the inventive weighting set forth above:
It can be used for beam hardening corrections, for example, to obtain a constant system weighting (SWF) function of an image data set as described in US Pat 5 for w (E) CT measurement for 80 kVp and 140 kVp tube voltage is shown.

Furthermore, monoenergetic calibrations can be performed with a target weighting function given by w _c (E) = δ (E - E ₀ ). One application of these monoenergetic attenuation coefficients is the contrast enhancement in image differences of particular tissue.

by virtue of the circumferential scan in the CT thus results for every point r → in the object a single energy-dependent weighting function. This function Ω (E, r →) is thus independent of traversed path of a beam or beam direction.

The local spectral weighting function Ω (E, r →) thus describes for each location r → the reconstructed image data that contains the attenuation values μ (r →) represent the spectral weighting effective there. It maps the mean of the spectral weights of the tomographic scan paths used in the generation of the associated scan paths μ (r →) data are available. Ω (E, r →) can be calculated exactly from the energy-dependent attenuation functions μ (E, r →).

Equation (12) combines the reconstructed image data representing the effective attenuation coefficients μ (r →), with the energy-dependent attenuation function μ (E, r →). The structure of equation (12) is similar to the result of the linear estimation. The global weighting function w (E) is replaced by the local weighting function Ω (E, r →). Equation (12) is itself an exact formula. However, it has intrinsic properties, for example, the local weighting function Ω (E, r →) requires the knowledge of μ (E, r →) of the entire object.

The local weighting function Ω (E, r →) can be calculated from a space integral over μ (E, r →) be calculated in polar or Cartesian coordinates.

If, for example, the local weighting function Ω (E, r →) for the center r _{0 of} an aqueous object with a specific diameter is calculated, the radiation spectrum S ₂ (E) being the 140 kVp spectrum of an x-ray tube according to 5 and the spectral sensitivity D (E) of an integrating GdOS detector according to 6 is assumed, the result is a resulting Ω (E, r → ₀ ) as in 12 shown for different diameter of the aqueous object. With increasing diameter, here 10 cm, 20 cm, 30 cm and 40 cm, the energy weighting Ω (E, r → ₀ ) shifts to higher energies.

A spectral reconstruction is performed as follows.

1. Spectral measurement

J CT measurements with different spectral weights w _j (E), j = 1, 2, ... J are performed. For example, these can be dual energy scans (J = 2) with an x-ray tube setting of 80 kVp and 140 kVp for two separate scans. The resulting sinograms can be reconstructed into image data sets with the aid of generally known reconstruction algorithms, for example a filtered backprojection. One thus obtains J different image data sets from the local spectrally weighted and measurement spectrum-dependent attenuation values μ _j (r →). The reconstructed datasets contain beam hardening effects.

2. Calculation of the local weakening

) kann aus den Bilddatensätzen aus Schritt 1 geschätzt werden. In medizinischen CTs kann zum Beispiel durch eine Segmentierung durch Grenzwertbildung von μ _j(r →) oder eine Basismaterialzerlegung eine Separation in zwei Hauptschwächungen, beispielsweise Wasser und Knochen, erzeugen. Die resultierende Anfangsschwächung gibt eine erste Abschätzung der messspektrumabhängigen Wichtungsfunktionen Ω⁽⁰⁾(E, r →) entsprechend den Gleichungen (11) und (12).The calculation of the spectrometry-dependent weighting functions Ω (E, r →) requires the X-ray tube spectrum S (E), the detector response function D (E) and the initial estimation of the energy-dependent attenuation functions μ ⁽⁰⁾ (E, r →). Both the X-ray tube spectrum S (E) and the detector response function D (E) can be obtained from simulations or direct measurements with a calibrated detector. The initial estimation of the measurement spectrum-independent attenuation functions μ ⁽⁰⁾ (E,

) can be estimated from the image data sets from step 1. In medical CTs, for example, segmentation can be done by thresholding μ _j (r →) or a base material decomposition produce a separation into two major weakenings, for example, water and bone. The resulting initial attenuation gives a first estimate of the measurement spectrum dependent weighting functions Ω ⁽⁰⁾ (E, r →) according to equations (11) and (12).

3. Reconstruction of the local energy-dependent weakening

Along with μ _j (r →) from step 1 and Ω ⁽⁰⁾ (E, r →) from step 2, the parameterization of the energy-dependent attenuation functions μ ⁽¹⁾ (E, r →) can be calculated from equation (13). For this purpose, a base material decomposition known per se is used, but on the basis of the attenuation function μ ⁽¹⁾ (E, r →) independent of the measuring radiation and its spectrum and spectral weighting in the detector, and thus independent of the spectra. For example, a dual energy basal material decomposition generates M = 2 concentrations of water and bone.

4th iteration

) berechnen. Somit ergibt sich ein stetiger iterativer Prozess. Aufgrund der langsam variierenden und fehlertoleranten mathematischen Struktur von Ω(E, r →) ist die Anzahl der Iterationsschritte gering. In der Praxis werden ein bis zwei Iterationen ausreichen.With the, over μ ⁽⁰⁾ (E, r →), more precise information about the local energy-dependent attenuation function μ ⁽¹⁾ (E, r →) from step 3, one can return to step 2 and a more recent value Ω ⁽¹⁾ (e,

) to calculate. This results in a steady iterative process. Due to the slowly varying and fault-tolerant mathematical structure of Ω (E, r →), the number of iteration steps is small. In practice, one to two iterations will suffice.

In the 14 are three exemplarily determined local measurement spectrum dependent weighting functions Ω _j (E, r →) at the locations P1, P2 and P3 of a in the 13 shown measuring field with a phantom 7 shown. This was based on the spectrum of a 140 kVp X-ray of a CT system. The effect of the beam hardening can be seen in the course of the weighting functions at the locations P1, P2 and P3, which experience an even stronger shift to the harder spectrum the more central the measuring point is selected.

A schematic representation of this method according to the invention is illustrated with an embodiment using iteration in the 15 with the process steps 16.1 to 16.7 shown.

In the first process step 16.1 an object to be examined is scanned with the help of a CT system, based on at least two different scanning spectra, and it becomes the local average attenuation values μ _j (r →) reconstructed. The index j defines the different spectral weighting functions of the radiation detector combinations used here. In step 16.2 is based on the determined data sets of local mean attenuation values μ _j (r →) a coarse material distribution in the examination subject and also measuring field - this includes, for example, a patient bed in the measurement field - and in knowledge of the resulting and based on the base materials recognized here previously known energy-dependent attenuation functions μ ⁽¹⁾ (E, r →) determined. This will be in the process step 16.3 the local measurement spectrum-dependent weighting functions Ω _j (E, r →) according to the equation 11 - or according to the Equation 12 - determined.

On the basis of the thus determined local measurement spectrum-independent attenuation functions μ (E, r →) and the local measurement spectrum-dependent weighting functions Ω _j (E, r →), the method step now becomes 16.4 a detailed base material decomposition take place. This will be the previously existing knowledge 16.7 on the fundamentally possible material composition of the examination object, so possible existing base materials M ₁ ... M _N , utilized. In a patient who is examined by means of a contrast agent, for example, water or soft tissue, bone, air (in the lung and other measuring field) and one or more contrast agents can be used as base materials. The energy-dependent attenuation functions μ ₁ (E, r →) ... μ _N (E, r →) of these N materials are either previously known or can be determined via detailed preliminary examinations exactly outside the current CT method. Basically, however, this knowledge is already in the process step 16.2 used, since already here rialverteilung due to an initial Mate energy-dependent attenuation functions μ (E, r →) are used.

If the local partial densities c _i (r →), also called mass concentrations, are calculated in this way, then these can be determined in the method step 16.6 in a tomographic representation. At the in 15 However, the process shown is in step 16.5 first decided whether an abort criterion for the iteration is met. For this purpose, for example, a determined error Δ between two iteration results can be calculated and an iteration termination can be initiated if an error value Δ is undershot below a constant k or if a predetermined number of iterations n is exceeded. Otherwise, becomes the process step 16.3 branched.

When The result is the user absorption data or image data no longer available for examination used radiation and one of the used Measuring system allow neutral representation of the object to be examined.

In addition, it is pointed out that under reconstructed CT image data sets, the spectrally weighted and therefore measurement spectrum-dependent attenuation values μ themselves are understood, although in the presentation of these reconstructed CT image data sets, so-called CT numbers or CT values X in HU (FIG. Hounsfield units) are shown in a gray scale. The conversion between CT value X and attenuation value μ normalized to water is carried out according to the equation

In summary is therefore proposed in this invention, from at least two different spectral weights determined CT data sets with local energy-averaged attenuation values an initial one To determine material distribution of an object to be examined, under Application of this material distribution local measurement spectrum dependent Determine weighting functions which are local, measurement spectrum independent Calculate attenuation values, using a Base material decomposition using local energy-averaged Attenuation values, the local measurement spectrum dependent Weighting functions and the local measurement spectrum independent Attenuation functions the local partial densities of the examination object be calculated. In addition, an iteration with out the calculation now more accurately known material distribution and / or local measurement spectrum-dependent weighting functions and / or local measurement spectrum independent attenuation functions be performed. Of course this can Method be combined with all the features mentioned above.

Especially it is understood that the above features of the invention not only in the specified combination, but also in other combinations or alone, without to leave the scope of the invention.

QUOTES INCLUDE IN THE DESCRIPTION

This list The documents listed by the applicant have been automated generated and is solely for better information recorded by the reader. The list is not part of the German Patent or utility model application. The DPMA takes over no liability for any errors or omissions.

Cited patent literature

• - DE 10143131 B4 [0005]

Cited non-patent literature

• - PHYS. MED. BIOL., 1976, VOL. 21, NO. 5, 733-744, "Energyselective Reconstructions in X-ray Computerized Tomography, RE Alvarez and A. Macovski [0004]

Claims (18)

Method for reconstructing CT image data sets from N ≥ 2 projection data sets recorded with N different spectral weights (S _j (E) * D _j (E) with j = 1 to N), defined as product function from the energy-dependent weighting of a used radiation spectrum (S _j (E)) and the spectral sensitivity of a detector (D _j (E)) used, comprising the following method steps: 1.1. CT scan of an examination subject ( 7 ) and generation of the N projection data sets, measured in each case with different spectral weighting (S _j (E) .D _j (E)), 1.2. Reconstruction of N CT image data sets, each with a plurality of local image values, the local mean attenuation values ( μ _i (r →)) at the location (r →) of the examination subject ( 7 ) and a measuring field (MF) as a function of the energy-dependent spectral weights (S _j (E) .D _j (E)) used for the scan, characterized in that the following method steps are carried out: 1.3. Determination of a first coarse material distribution with M ≥ N predetermined base materials in the examination subject ( 7 ) on the basis of the previously determined local mean attenuation values ( μ _i (r →)), 1.4. Determination of first local energy-dependent attenuation functions (μ (E, r →)) in the examination subject ( 7 ) based on the coarse material distribution and the known composition of the base materials, 1.5. Calculation of local measurement spectrum-dependent weighting functions (Ω _j (E, r →)) at the locations (r →) of the examination subject ( 7 ) on the basis of the last known material distribution in the examination object ( 7 ) with the associated material-specific energy-dependent attenuation functions (μ (E, r →)), 1.6. Performing a material decomposition for M base materials for the determination of material-specific local partial densities (c _i (r →) with i = 1 to M) of the examination object ( 7 ) using the last known local energy-dependent attenuation functions (μ (E, r →)) and the local measurement spectrum-dependent weighting functions (Ω _j (E, r →)), and 1.7. Tomographic representation of the calculated local partial densities (c _i (r →) of the examination subject ( 7 ). Method according to the preceding claim 1, characterized in that after the determination of local partial densities (c _i (r →) of the base materials of the examination object ( 7 ) the so-known material distribution of the base materials, the steps 1.5 to 1.6 are performed until a predetermined iteration condition is met. Method according to the preceding claim 2, characterized in that in the iterative calculation of the local measurement spectrum-dependent weighting functions (Ω _j (E, r →)) corrected energy-dependent attenuation functions (μ (E, r →)) according to

where E describes the energy dependence and r → the spatial dependence of the attenuation functions and μ _j (E, r →) represents the density normalized energy-dependent attenuation functions of the j-th material. Method according to one of the preceding claims 1 to 3, characterized in that for the calculation of the local measurement spectrum-dependent weighting functions (Ω (E, r →)) at the locations (r →) of the examination subject ( 7 ) the relationship

is used with:

where P corresponds to the forward projection operator in a reconstruction of a CT image data set and R ^-1 radon transformation operator in a reconstruction of a CT image data set. Method according to one of the preceding claims 1 to 4, characterized in that for material decomposition the relationship:

where Index 1 to N indicates the number of energy-dependent spectral weights used (S _j (E) * D _j (E)), Index 1 to M represents the number of materials to be determined, while E represents energy dependence and r → describe the location dependence of the attenuation coefficient. Method according to one of the preceding claims 1 to 5, characterized in that different energy-dependent spectral weights (S _j (E) * D _j (E)) are achieved by the use of different radiation spectra. Method according to one of the preceding claims 1 to 6, characterized in that different energy-dependent spectral weights (S _j (E) * D _j (E)) are achieved by using different spectral detector sensitivities. Method according to one of the preceding claims 1 to 7, characterized in that a patient ( 7 ) is used. Method according to the preceding Claim 8, characterized in that at least one of Base materials is water. Method according to one of the preceding Claims 8 to 9, characterized in that at least one of the basic materials is bone. Method according to one of the preceding Claims 8 to 10, characterized in that at least one of the base materials is air. Method according to one of the preceding Claims 8 to 11, characterized in that at least one of the base materials is a contrast agent. Method according to one of the preceding Claims 1 to 7, characterized in that as Object of investigation an inanimate object is used. Method according to the preceding Claim 13, characterized in that as base materials previously known materials are used. Method according to one of the preceding Claims 13 to 14, characterized in that as base materials, at least one material sought in the article is used. Method according to one of the preceding claims 1 to 15, characterized in that the CT scan of the examination object ( 7 ) is performed over a plurality of equally distributed scanning angles. Method, wherein: - from at least two different spectral weights (S _j (E) * D _j (E)) determined CT data sets with local mean attenuation values ( μ _i (r →)) an initial material distribution of an examination object ( 7 ) is determined, using this material distribution local measurement spectrum-dependent weighting functions (Ω _j (E,

)), which can be used to calculate local energy-dependent attenuation functions (μ (E, r →)), using a base material decomposition using the local energy-averaged attenuation values (FIG. μ _j (r →)), the local measurement spectrum-dependent weighting functions (Ω _j (E, r →)) and the local energy-dependent attenuation functions (μ (E, r →)) the local partial densities (c _i (r →)) of the examination object are calculated , CT system for scanning an examination subject comprising a control and computing unit ( 10 ) with a memory for computer program code (Prg ₁ prg _n ), characterized in that computer program code (Prg ₁ -Prg _n ) is stored in the memory, which can perform the method steps of one of the preceding method claims in operation.