DE102013214361A1 - Method for reducing quantization artifacts - Google Patents

Abstract

The invention relates to a method for reducing image aberrations in a digitized image signal of an image detector with pixels arranged in a matrix, in particular for reducing quantization artifacts, comprising the following steps: S1) generating an image data record by digitizing analog image signals of the pixels of the image detector (4) Formation of an at least first projection, S2) creation of a first reconstruction from the image data of the image data set for generating reconstructed image data in the reconstruction space, S3) suppression of image defects in the first reconstruction by filtering the reconstructed data in the reconstruction space to produce filtered image data, S4) filtered data into the projection space to produce at least one filtered projection, S5) determining at least one threshold value indicative of a maximum possible difference of the image signals of the pixels due to the image defect, S6) Erm calculating the size of differences of the image signals of the pixels in the first and the filtered projection, S7) calculating a new improved projection data set from the fusion of the image signals of the pixels in the first and the filtered projection, wherein a removal of the filtered projection data is performed determined according to step S6 differences greater than the threshold. S8) creation of a second reconstruction with the improved projection data set and S9) reproduction of the second reconstruction with improved image quality.

Description

The invention relates to a method for reducing image errors in a digitized image signal of an image detector with pixels arranged in a matrix, in particular for reducing quantization artifacts.

Such a method for the reduction of aberrations in a digitized image signal of an image detector can be used, for example, in an angiography system for the angiographic examination of a patient who, for example, from the US 7,500,784 B2 is known and based on the 1 is explained.

The 1 shows an exemplified monoplanes X-ray system with one of a stand 1 in the form of a six-axis industrial or articulated robot held C-arm 2 at the ends of which is an X-ray source, for example an X-ray source 3 with X-ray tube and collimator, and an X-ray image detector 4 are mounted as an image recording unit.

By means of the example of the US 7,500,784 B2 Known articulated robot, which preferably has six axes of rotation and thus six degrees of freedom, the C-arm 2 be spatially adjusted, for example, by turning around a center of rotation between the X-ray source 3 and the X-ray image detector 4 is turned. The angiographic X-ray system according to the invention 1 to 4 is in particular about centers of rotation and axes of rotation in the C-arm plane of the X-ray image detector 4 rotatable, preferably around the center of the X-ray image detector 4 and around the center of the X-ray image detector 4 cutting axes of rotation.

The known articulated robot has a base frame which is fixedly mounted, for example, on a floor. It is rotatably mounted about a first axis of rotation a carousel. On the carousel is pivotally mounted about a second axis of rotation a rocker arm, on which is rotatably mounted about a third axis of rotation, a robot arm. At the end of the robot arm, a robot hand is rotatably mounted about a fourth axis of rotation. The robot hand has a fastener for the C-arm 2 which is pivotable about a fifth axis of rotation and about a perpendicular thereto extending sixth axis of rotation rotatable.

The realization of the X-ray diagnostic device is not dependent on the industrial robot. It can also find common C-arm devices use.

The X-ray image detector 4 may be a rectangular or square semiconductor flat detector, preferably made of amorphous silicon (a-Si). However, integrating and possibly counting CMOS detectors may also be used.

In the beam path of the X-ray source 3 is on a tabletop 5 a patient table a patient to be examined 6 as a research object. At the X-ray diagnostic facility is a system control unit 7 with an image system 8th connected to the image signals of the X-ray image detector 4 receives and processes (controls are not shown, for example). The X-ray images can then be displayed on displays by means of a ceiling-mounted, longitudinally movable, pivoting, rotating and height-adjustable carrier system 9 held monitor light 10 to be viewed as. In the system control unit 7 is still a device 11 provided whose function will be described below.

Instead of in 1 For example, illustrated X-ray system with the stand 1 In the form of the six-axis industrial or articulated robot, the angiographic X-ray system can also be a normal ceiling-mounted or floor-mounted bracket for the C-arm 2 exhibit.

Instead of the example illustrated C-arm 2 For example, the angiographic x-ray system may also include separate ceiling and / or floor mount brackets for the x-ray source 3 and the X-ray image detector 4 have, for example, are electronically rigidly coupled.

In radiography or fluoroscopy by means of such an X-ray diagnostic device, the medical 2-D data of the X-ray image detector 4 in the picture system 8th if necessary cached and then on a monitor the monitor 10 played.

In this so-called flat-detector computed tomography, the X-ray image detector is used 4 incident X-ray quanta recorded in an analogous manner and implemented digitally. To be digitally stored, the signal must be quantized. In contrast to normal computed tomography, this is done on a rather rough scale, which usually has 16 or 14 bit steps. Unfortunately, this quantization of the detector signal can result in reduced image quality.

There are many approaches to reducing noise, especially image noise, in the literature. For example, structure-controlled filters such as Andreas Maier et al. [1], or used bilateral filters, such as Anna Gabiger et al. [2] or C. Tomasi et al. [3]. There are also filters that try to do that To estimate and minimize noise locally like this Jing Wang et al. [4] is described.

The invention is based on the object to form a method for the reduction of quantization artifacts of the type mentioned in such a way that a good noise reduction is achieved in a simple manner.

The object is achieved for a method of the type mentioned by the features specified in claim 1. Advantageous embodiments are specified in the dependent claims.

• S1) Erzeugung eines Bilddatensatzes durch Digitalisierung von analogen Bildsignalen der Pixel des Bilddetektors (4) zur Bildung einer wenigstens ersten Projektion,
• S2) Erstellung einer ersten Rekonstruktion aus den Bilddaten des Bilddatensatzes zur Erzeugung rekonstruierter Bilddaten im Rekonstruktionsraum,
• S3) Unterdrückung von Bildfehlern in der ersten Rekonstruktion durch Filterung der rekonstruierten Bilddaten im Rekonstruktionsraum zur Erzeugung von gefilterten Bilddaten,
• S4) Abbildung der gefilterten Bilddaten in den Projektionsraum zur Erzeugung wenigstens einer gefilterten Projektion,
• S5) Ermittlung wenigstens eines Schwellenwerts, der einen maximal möglichen Unterschied der Bildsignale der Pixel aufgrund des Bildfehlers kennzeichnet,
• S6) Ermittlung der Größe von Differenzen der Bildsignale der Pixel in der ersten und der gefilterten Projektion,
• S7) Berechnung eines neuen verbesserten Projektionsdatensatzes aus der Fusion der Bildsignale der Pixel in der ersten und der gefilterten Projektion, wobei eine Entfernung derjenigen gefilterten Projektionsdaten durchgeführt wird, wenn die gemäß Verfahrensschritt S6 ermittelten Differenzen einen größeren Wert als den Schwellenwert ergeben. Es werden also alle Bildsignale der Pixel mit Differenzen größer als der Schwellenwert aus der ursprünglichen Projektion ermittelt, während alle Bildsignale der Pixel, die sich weniger unterscheiden, aus der gefilterten Projektion bestimmt werden können.
• S8) Erstellung einer zweiten Rekonstruktion mit dem verbesserten Projektionsdatensatz,
• S9) Wiedergabe der zweiten Rekonstruktion mit verbesserter Bildqualität.

The problem is solved for a method for the reduction of quantization artifacts according to the invention by the following method steps:

• S1) generating an image data set by digitizing analog image signals of the pixels of the image detector ( 4 ) for forming at least a first projection,
• S2) creation of a first reconstruction from the image data of the image data set for generating reconstructed image data in the reconstruction space,
• S3) suppression of image defects in the first reconstruction by filtering the reconstructed image data in the reconstruction space to produce filtered image data,
• S4) mapping the filtered image data into the projection space to produce at least one filtered projection,
• S5) determining at least one threshold characterizing a maximum possible difference of the image signals of the pixels due to the image defect,
• S6) determining the size of differences of the image signals of the pixels in the first and the filtered projection,
• S7) calculating a new improved projection data set from the fusion of the image signals of the pixels in the first and the filtered projection, wherein a removal of those filtered projection data is performed if the differences determined according to method step S6 result in a value greater than the threshold value. Thus, all the image signals of the pixels having differences greater than the threshold value from the original projection are determined, while all the image signals of the pixels which differ less can be determined from the filtered projection.
• S8) creating a second reconstruction with the improved projection data set,
• S9) Reproduction of the second reconstruction with improved picture quality.

According to the invention, the image errors due to the digitization of the analog image signals can be caused quantization artifacts.

It has proved to be advantageous if at least one denoising method from the group of median filters, Gaussian filters, bilateral filters, minimization of the TV standard and / or structure-controlled filters is used to suppress image errors according to method step S3.

In an advantageous manner, to determine at least one threshold value according to method step S5, the quantization stages used in the digitization of analog image signals can be used.

According to the invention, a weighted sum of the two images can be used to calculate a new improved projection data set according to method step S7, where the weight of the weighted sum is at least one criterion from the group of magnitude of the difference, strength of the local noise at the current pixel and strength of the difference in a local Neighborhood may contain.

Advantageously, the method for digitizing the detector signals according to method step S1 in a non-linear quantization can be performed.

In order to achieve a further image improvement, the method steps S3 to S8 can be repeated several times.

Advantageously, the creation of a second reconstruction according to method step S7 can be combined with other two-step or multi-step methods, such as metal artifact reduction or iterative reconstruction methods.

According to the invention, the image data set according to method step S1 can be a 3-D image data set and / or a temporal sequence of successive image data.

It has proven to be advantageous if the method for reducing image aberrations in a digitized image signal of an image detector with pixels arranged in a matrix of an angiography system for angiographic examination or treatment of an organ, vascular system or other body regions is used as the examination subject of a patient, the angiography system an X-ray source, the X-ray image detector, attached to the ends of a C-arm, a patient support table having a table top for supporting the patient, a system control unit, an imaging system, and a monitor.

The invention is explained in more detail with reference to embodiments shown in the drawing. Show it:

1 a known C-arm angiography system with an industrial robot as a carrying device,

2 a projection with equidistant quantization errors,

3 a reconstruction of the projection according to 2 with a clear picture noise,

4 a line profile of the reconstruction according to 3 with clearly visible quantization levels,

5 a reconstruction according to 3 with suppression of image noise by filtering,

6 a projection according to 2 with suppression of image noise by filtering,

7 a line profile with indistinct quantization levels,

8th a method sequence according to the invention,

9 an initial flawed reconstruction,

10 a filtered reconstruction with improvements regarding quantization errors,

11 an inventively improved reconstruction and

12 a flowchart.

In the 2 is a projection or an X-ray image 12 represented, which is afflicted with equidistant quantization errors. The corresponding in 3 illustrated reconstruction 14 shows clear image noise caused by poor quantization. In the picture of the reconstruction 14 is an image noise in homogeneous areas of about 10 Hounsfield Units (HU) measurable.

The 4 shows the associated line profile 15 along the line 13 in the x-ray 12 according to 2 , in which the gray values GW are plotted over pixels or the distance. The quantization levels are clearly visible.

A suppression of image noise in the image space or reconstruction space is possible. However, even edges are blurred. The result of such filtering 16 is in the 5 to 7 to see. The image noise in the reconstruction space with 2 HU is significantly reduced, as is the 5 shows. However, the picture gives a very artificial appearance. A of the reconstructed image in the reconstruction space in the projection space 18 according to 6 shows that the quantization artifacts are reduced. This is also the associated line profile 20 along the line 19 in the X-ray image of the projection room 18 according to 7 in which again the gray values GW are plotted over pixels or the distance. The quantization levels are not so clearly visible.

• S1) Der beispielsweise durch die Röntgendiagnostikeinrichtung gemäß 1 erzeugte Bilddatensatz einer ersten Projektion bzw. eines ersten Röntgenbildes 12 durch Digitalisierung von analogen Bildsignalen des Bilddetektors 4 wird abgespeichert.
• S2) Aus diesem ersten Bilddatensatz wird eine erste oder initiale Rekonstruktion 14 im Bildraum oder Rekonstruktionsraum erstellt, die jedoch mit den erwähnten, oben genannten Bildfehlern behaftet ist (siehe 3).
• S3) Die Bildfehler, insbesondere das Bildrauschen, beispielsweise hervorgerufen durch äquidistante Quantisierungsfehler, werden durch Filterung 16 der Rekonstruktion im Bildraum oder Rekonstruktionsraum stark unterdrückt. Im dargestellten Beispiel wurde ein bilateraler Filter verwendet. Es kann aber auch jedes andere 2-D/3-D-Entrauschungsverfahren Anwendung finden.
• S4) Das gefilterte bzw. entrauschte Bild im Rekonstruktionsraum wird dann in den Projektionsraum 18 abgebildet.
• S5) Da die Höhen der bei der Digitalisierung verwendeten Quantisierungsstufen bekannt sind, kann ein maximal bedingt durch die Quantisierung möglicher Unterschied der Pixelsignale als Schwellenwert ermittelt werden.
• S6) Alle Pixelwerte der gefilterten Projektion, die sich mehr als diesen Unterschied von der ursprünglichen Projektion, bzw. dem Röntgenbild 12, unterscheiden, d. h. die einen Unterschied aufweisen, der größer als der Schwellenwert ist, werden aus der ursprünglichen Projektion, dem Röntgenbild 12, ermittelt bzw. übernommen. Es kann in diesem Falle davon ausgegangen werden, dass bei diesen Pixelwerten beispielsweise eine relevante Kante durch die Filterung verwischt wurde. Alle Pixelwerte der gefilterten Projektion, die sich weniger unterscheiden, d. h. die einen kleineren Unterschied als der Schwellenwert aufweisen, werden aus der neuen rauschreduzierten gefilterten Projektion ermittelt. Es liegt dabei die Überlegung zugrunde, dass dieser Unterschied bzw. die Differenz auf einem von einer Quantisierungsstufe stammenden Bildfehler beruht.
• S7) Aus der Fusion der beiden Datensätze gemäß der im Verfahrensschritt S6 festgelegten und beschriebenen Regel, nach der gefilterte Projektionsdaten mit Differenzen größer als der Schwellenwert entfernt wurden, wird ein neuer verbesserter Projektionsdatensatz im Projektionsraum 18 bestimmt.
• S8) Der verbesserte Projektionsdatensatz wird erneut rekonstruiert, so dass man als Ergebnis eine zweite Rekonstruktion mit verbesserter Bildqualität erhält.
• S9) Diese zweite Rekonstruktion mit verbesserter Bildqualität, wie sie beispielsweise anhand der 9 bis 11 demonstriert wird, wird wiedergegeben. Die Kanten bleiben erhalten und das Bildrauschen ist mit 5 HU halb so groß, wie das in der originalen ersten Rekonstruktion.

A reduction according to the invention of the quantization artifacts in the reconstruction space can now be determined according to the following, with reference to FIG 8th explained method steps, for example in the device 11 respectively:

• S1) The example by the X-ray diagnostic device according to 1 generated image data set of a first projection or a first x-ray image 12 by digitizing analog image signals of the image detector 4 is saved.
• S2) This first image data set becomes a first or initial reconstruction 14 created in the image space or reconstruction space, but which is subject to the above-mentioned image errors (see 3 ).
• S3) The image errors, in particular the image noise, caused, for example, by equidistant quantization errors, are produced by filtering 16 Reconstruction in the image space or reconstruction space strongly suppressed. In the example shown a bilateral filter was used. However, it can also be any other 2-D / 3-D Entrauschungsverfahren apply.
• S4) The filtered or noisy image in the reconstruction space is then placed in the projection space 18 displayed.
• S5) Since the heights of the quantization steps used in the digitization are known, a maximum possible difference of the pixel signals due to the quantization can be determined as a threshold value.
• S6) All pixel values of the filtered projection, which is more than this difference from the original projection, or the X-ray image 12 , ie, which have a difference greater than the threshold, become the original projection, the X-ray image 12 , determined or accepted. In this case, it can be assumed that, for example, a relevant edge was blurred by the filtering in the case of these pixel values. All pixel values of the filtered projection which differ less, ie which have a smaller difference than the threshold, are determined from the new noise reduced filtered projection. It lies thereby the Consideration that this difference or difference is based on an image error originating from a quantization step.
• S7) From the fusion of the two data records according to the rule defined and described in method step S6, after which filtered projection data having differences greater than the threshold value has been removed, a new improved projection data set is obtained in the projection space 18 certainly.
• S8) The improved projection data set is reconstructed again, resulting in a second reconstruction with improved image quality.
• S9) This second reconstruction with improved image quality, as shown for example by the 9 to 11 is demonstrated is played back. The edges are preserved and the image noise is half the size of 5 HU, as in the original first reconstruction.

• – Im Verfahrensschritt S3 können verschiedene Entrauschungsverfahren einsetzt werden: Median-Filter, Gauß-Filter, Bilateral-Filter, Minimierung der TV−Norm und/oder Struktur−gesteuerte Filter.
• – Im Verfahrensschritt S6 kann eine "weiche" Entscheidung getroffen werden. Statt einer "harten" Entweder−Oder−Wahl, kann beispielsweise auch eine gewichtete Summe beider Bilder im Verfahrensschritt S6 erfolgen. Das Gewicht der Summe kann dabei verschiedene Kriterien enthalten: – Stärke des Unterschieds, – Stärke des lokalen Bildrauschens an dem aktuellen Pixel und/oder – Stärke des Unterschieds in einer lokalen Nachbarschaft.
• – Alternativ zu einer äquidistanten Quantisierung bei der Erzeugung eines ersten Projektionsdatensatzes durch Digitalisierung von analogen Bildsignalen des Bilddetektors 4 kann auch eine nicht−lineare, beispielsweise eine logarithmische Quantisierung angenommen werden. Üblicherweise werden die Bilddaten in einer Domäne aufgenommen, die erst mit einem negativen Logarithmus verarbeitet werden muss. Dadurch ergeben sich lokal unterschiedliche Quantisierungsstufen, die von der Stärke des Signals abhängig sind. Die Verfahrensschritte S5 bis S7 müssen dann entsprechend angepasst werden.
• – Die Verfahrensschritte S3 bis S8 können mehrfach wiederholt werden, um eine weitere Bildverbesserung zu erreichen.
• – Eine Kombination mit anderen zwei− oder mehrschritt Verfahren, wie Metallartefakt-Reduktion oder iterativen Rekonstruktionsverfahren ist möglich und sinnvoll, da somit Gesamtrechenzeit eingespart werden kann.

There are a multiplicity of alternative embodiments for these method steps according to the invention. Some are listed below, the list being exemplary only, not exclusive:

• In method step S3, various denoising methods can be used: median filter, Gaussian filter, bilateral filter, minimization of the TV standard and / or structure-controlled filters.
• In method step S6, a "soft" decision can be made. Instead of a "hard" either-or-choice, for example, a weighted sum of both images in step S6 can take place. The weight of the sum may include several criteria: strength of the difference, strength of the local noise at the current pixel, and / or strength of the difference in a local neighborhood.
• - As an alternative to an equidistant quantization in the generation of a first projection data set by digitizing analog image signals of the image detector 4 It is also possible to assume a non-linear, for example a logarithmic, quantization. Usually, the image data is recorded in a domain that first needs to be processed with a negative logarithm. This results in locally different quantization levels, which are dependent on the strength of the signal. The method steps S5 to S7 must then be adapted accordingly.
• - The process steps S3 to S8 can be repeated several times to achieve a further image enhancement.
• - A combination with other two- or multi-step method, such as metal artifact reduction or iterative reconstruction method is possible and useful, since thus total computing time can be saved.

Based on 9 to 11 the improvement according to the invention will now be explained and visualized in more detail. The 9 gives an initial or first reconstruction 21 again, which is still associated with the known, above-mentioned errors (quantization errors). The filtered reconstruction 22 according to 10 Although already shows improvements in quantization errors. With the filtering, however, image-relevant edges were suppressed. These are again improved by the above-mentioned rule improved reconstruction 23 visualized that 11 can be seen.

The 12 shows a flowchart of the inventive method according to 8th , Starting from a through the angiographic X-ray system 1 to 4 captured analog image data set 24 a digitization takes place 25 and storing the first digital projection data 26 , The initial reconstruction according to method step S2 yields reconstruction data 27 in the picture or reconstruction room. A filtering 28 Smoothes the noisy reconstruction, so you get filtered reconstruction data 29 from which, by mapping the reconstructed image in the reconstruction space into the projection space, filtered projection data 30 determined.

A thresholding 31 derives from digitization 25 a threshold SW. A differential stage 32 determines the size of the deviations of the pixel values of the filtered projection 30 from those of the original projection 26 , Using a pixel-wise query 33 it is checked if differences in the pixel values of the projections are below or above the threshold.

If the difference values of the projections exceed the threshold, the pixel values become 34 from the original projection data 26 taken over, since with high probability at these pixel values a relevant edge was blurred by the filtering. If the differences of the pixel values of the projection are smaller than the threshold, the pixel values become 35 from the new noise-reduced filtered projection 30 assuming that this difference is due to an aberration of a quantization level.

These pixel values determined in this way 34 and 35 the projections 26 and 30 be in a merger 36 merged and then a second reconstruction 37 Computed with improved picture quality, then in a playback 38 for example, on the monitor of the monitor 10 is shown.

In the pixel-wise query 33 with subsequent merger 36 For example, instead of a "hard" either-or-choice, a "soft" decision can be made by a weighted sum of both images.

So far, quantization errors in reconstruction methods are not explicitly modeled. The quantization error caused by the digitization of the output signal of the X-ray image detector 4 is formed, is well known and can be determined exactly. Therefore, it is also possible to distinguish exactly which errors may have arisen during the quantization and which not. As a result, the method according to the invention can explicitly model and correct this. In the reconstruction, the error is expressed by image noise. A filtered projection value is, to a first approximation, better than the errored value, as is a comparison of the 9 and 10 clarified.

By the basis of the 8th As discussed above, a measurable improvement in image quality is achieved without requiring changes to the product hardware. There is only a better modeling of the imaging process. The success shows the reproduction of the improved reconstruction in 11 ,

Unfortunately, the method requires at least a second reconstruction run. However, this is also used in other prototypes, in other executable software or concrete modeling of a subcomponent of the target system, such as the reduction of metal artefacts. Therefore, a combination with these makes sense.

literature

• [1] Andreas Maier, Lars Wigström, Hannes G. Hofmann, Joachim Hornegger, Lei Zhu, Norbert Strobel, and Rebecca Fahrig, "Three-dimensional Anisotropic Adaptive Filtering of Projection Data for Noise Reduction in Cone Beam CT," Med. Phys., To appear 2011 ,
• [2] Anna Gabiger, Robert Weigel, Steven Oeckl, Peter Schmitt, "Enhancement of CT Image Quality via Bilateral Filtering of Projections," The First International Conference on Image Formation in X-Ray Computed Tomography, Salt Lake City, United States, 2010, p. 140-143 ,
• [3] C. Tomasi, R. Maduchi, "Bilateral Filtering for Gray and Color Images," ICCV '98: Proceedings of the Sixth International Conference on Computer Vision, Washington, DC, United States, 1998, p. 839-846 ,
• [4] Jing Wang, Tianfang Li, Hongbing Lu, and Zhengrong Liang, "Penalized weighted least-squares approach to sinogram noise reduction and image reconstruction for low-dose x-ray computed tomography," IEEE Trans Med Imaging, vol. 25, pp. 1272-1283, 2006 ,

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

• US 7500784 B2 [0002, 0004]

Cited non-patent literature

• Andreas Maier et al. [0013]
• Anna Gabiger et al. [0013]
• C. Tomasi et al. [0013]
• Jing Wang et al. [0013]

Claims (12)

Method for reducing image aberrations in a digitized image signal of an image detector ( 4 ) with pixels arranged in a matrix, characterized by the following steps: S1) Generation of an image data record by digitizing analogue image signals of the pixels of the image detector ( 4 S2) creation of a first reconstruction from the image data of the image data set for generating reconstructed image data in the reconstruction space, S3) suppression of image defects in the first reconstruction by filtering the reconstructed image data in the reconstruction space to produce filtered image data, S4) S5) determining at least one threshold characterizing a maximum possible difference of the image signals of the pixels due to the image defect, S6) determining the size of differences of the image signals of the pixels in the first and the second image the filtered projection, S7) calculating a new improved projection data set from the fusion of the image signals of the pixels in the first and the filtered projection, wherein a removal of those filtered projection data is performed when the gem According to method step S6, differences result in a value greater than the threshold value. S8) creation of a second reconstruction with the improved projection data set and S9) reproduction of the second reconstruction with improved image quality. A method according to claim 1, characterized in that the image errors due to the digitization of the analog image signals are caused quantization artifacts. A method according to claim 1 or 2, characterized in that for suppressing aberrations according to method step S3 at least one Entrauschungsverfahren from the group of median filter, Gaussian filter, bilateral filter, minimization of the TV standard and / or structure-controlled filters uses becomes. Method according to one of Claims 1 to 3, characterized in that the quantization stages used in the digitization of analog image signals are used to determine at least one threshold value in accordance with method step S5. Method according to one of claims 1 to 4, characterized in that for the calculation of a new improved projection data set in accordance with method step S7, a weighted sum of both images takes place. The method of claim 5, characterized in that the weight of the weighted sum includes at least one of the strength of difference, the strength of the local noise at the current pixel, and the strength of the difference in a local neighborhood. Method according to one of claims 1 to 6, characterized in that for digitizing the detector signals according to method step S1, a non-linear quantization is performed. Method according to one of claims 1 to 7, characterized in that the method steps S3 to S8 are repeated several times. Method according to one of claims 1 to 8, characterized in that the creation of a second reconstruction according to method step S7 is combined with other two- or multi-step method. Method according to one of claims 1 to 9, characterized in that the image data set according to method step S1 is a 3-D image data set. Method according to one of claims 1 to 10, characterized in that the image data set according to method step S1 is a temporal sequence of successive image data. Method according to one of Claims 1 to 11, characterized in that it is used to reduce image aberrations in a digitized image signal of an image detector ( 4 ) with pixels of an angiography system arranged in a matrix for the angiographic examination or treatment of an organ, vascular system or other body regions as a subject of examination of a patient ( 6 ), wherein the angiography system uses an X-ray source ( 3 ), the X-ray image detector ( 4 ) at the ends of a C-arm ( 2 ), a patient table with a table top ( 5 ) for storage of the patient ( 6 ), a system control unit ( 7 ), an image system ( 8th ) and a monitor ( 10 ) having.

Images