DE102009017437A1 - Method for determining set of mapping rules for mapping points in three-dimensional space to points in two-dimensional x-ray image, involves deriving respective mapping rules and partially mapping calibration object in x-ray images - Google Patents

Abstract

The method involves placing a calibration object with predetermined structures in an x-ray imaging system. Two-dimensional x-ray images of the calibration object are obtained with a set of relative positions of imaging components in a position of an x-ray C-arm. A sequence is assigned to each of the x-ray images. Respective mapping rules are derived based on assignment of the sequences. The calibration object is partially mapped in the x-ray images. A combined x-ray image is produced using the x-ray images.

Description

The The invention relates to a method for determining a plurality of Mapping rules for the mapping of points in the three-dimensional space on points in a two-dimensional x-ray image through an X-ray imaging system. The dots in the three-dimensional Spaces are contained in volume elements of a real object, the by the mapping rule on a surface element of a X-ray detector is imaged, wherein the surface element corresponds to a point in a two-dimensional x-ray image.

Such Mapping requirements, often as so-called projection rules are required, if made from a plurality of two-dimensional X-ray images recovered volume information should be, namely in the context of the so-called 3D reconstruction, the one back projection, thus a reversal of the mapping rules, includes. To win a 3D reconstruction, need x-rays from different perspectives be recorded. Classically, this is an X-ray C-arm, an X-ray source and an X-ray detector contributes to a rotation axis in different positions rotated, and at each position an X-ray image is taken.

It may be that an image object is too big, than that the X-ray detector could detect the whole object. For this purpose one has gone over to such Object only partially record. This is one of the imaging Components X-ray source and X-ray detector, typically the X-ray detector, re of the X-ray C-arm. In one and the same position of the X-ray C-arc there are then two different relative positions from X-ray source and X-ray detector. As a rule, it suffices to assign two such relative positions choose a slightly larger object with the then recorded two-dimensional X-ray images completely grasp. In a robot-guided C-arm is opposite to a traversability of the X-ray detector The C-arm is not required, here can by tilting of the X-ray detector different relative positions be obtained.

By the mobility of the imaging component, namely in particular of the X-ray detector, there is an additional Degree of freedom, which in the definition of the mapping rule, the when 3D reconstruction is needed, to be considered is.

A Abbildungsvorschrift of the type mentioned wins you conventionally in that a calibration object, a so-called calibration phantom, accompanies exactly the same settings will be shown later for obtaining the two-dimensional X-ray images for the 3D reconstruction should be taken. You make that Calibration object and can predetermined structures of the calibration object recognize in the pictures. Due to a recognition algorithm let then the structures in the individual 2D X-ray images Assign structures to the calibration object.

When Such a calibration object has proved to be particularly efficient in that, for. B. in a cylindrical Plexiglas body, a Plurality of (small) bodies, preferably as spheres are formed, are arranged. When using balls is It is known to use balls of different sizes use. It suffices here if in the 2D X-ray images can be detected, whether a ball of the first size or a ball of the second size is shown. The sequence of balls is chosen in the calibration object so that each subsequence of eight bullets clearly one particular Job is to assign. Because the balls are two different Sizes can be, can be treat this as a binary code. The code represents then the location in the calibration object. The consequence of bullets is typically arranged spirally in the calibration object.

Should now for a later 3D reconstruction of such Object that is too big to be completely in one single 2D X-ray image, a calibration, Thus, a determination of mapping rules, done with the help of the calibration object, so when choosing different relative positions of X-ray source and X-ray detector also the calibration object incomplete in the individual X-ray images displayed. To the larger object of interest to fully map selects namely extreme positions of the X-ray detector.

In the DE 10 2006 041 033 A1 A method is described in which the individual 2D X-ray images of an object of interest are combined to form a virtual image, on the basis of which a backprojection is carried out as part of the 3D reconstruction. The DE 10 2007 026 115 A1 deals with the fact that such a virtual X-ray image is only used for a filtering, but then a back mapping to the individual X-ray images takes place and then these are used individually to produce the 3D reconstruction. The present method is similar to that in the DE 10 2007 026 115 A1 described basic situation that mapping rules for at least two relative positions of the X-ray source and X-ray detector at one and same position of the X-ray C-arm must be obtained.

A method for determining a mapping rule also deals with the DE 10 2007 042 333 A1 , Here, the mapping rule is determined at a position of the imaging components, and when moving to another position parameters are determined that determine the nature of the process. With the aid of these traversing parameters, a conversion is then carried out in order to obtain a mapping rule also for the further position.

In In practice, one has not been able to derive the mapping rules without further two-dimensional X-ray images. When using the calibration object with spiral Sequence of balls in binary coding was gone so far from that at least eight consecutive balls at the same time have to be imaged in the same x-ray image to clearly assign sequences. Because at extreme Positions of the X-ray detector not or only in small proportion such episodes are pictured with eight balls So far in addition to the actual for the Calibration of necessary X-ray images at extreme positions X-ray detector also X-ray images at central positions of the X-ray detector on. At a central position of the X-ray detector is a large part of the calibration object, if not the full calibration object, see so that the individual ball sequences clearly and with conventional Algorithms can be assigned.

It is very elaborate if an additional consequence 2D X-ray images, namely in the middle position of the X-ray detector to different positions of the X-ray C-arm, must be recorded. Especially is the x-ray imaging system and the one operating this Staff for a certain time because of the need the extraction of these additional images not for other tasks free.

It is the object of the invention, in the above-described starting situation, that a plurality of mapping rules for different relative positions from X-ray source and X-ray detector be obtained at the same position of the X-ray C-arm should, in comparison to the prior art for a time saving to care.

The Task is achieved by a method with the features according to claim 1 solved.

While so far an assignment to the individual X-ray images with respect the question of which of the predetermined structures where in these X-ray images on the basis of a separate X-ray image takes place, takes place according to the invention the Step of allocating exclusively using the Plurality of X-ray images in which the calibration object each incomplete is shown. This assignment is sufficient to derive the respective mapping rules.

The Invention is based on the finding that it is suitable when using Analytical methods is possible, even with imaging of less as eight spheres in a sequence in the individual X-ray images make reliable assignments or that with others Calibration objects such assignments exclusively on Basis for the derivation of the mapping rules needed X-rays is possible.

In particular, it is possible for the assignment to each of the plurality of X-ray images to take place separately. In the past, in particular, methods have been described which reduce errors in such assignments. An example is that published after the filing date of the present application DE 10 2008 035 656 described method. In this method, error quantities are determined and, on the basis of the value of these error variables, a decision is made as to whether a specific assignment is to be regarded as valid or not. Another method that could be successfully used is that in the article "Geometry Calibration for Arbitrary Trajectories" by Dr. Ing. Florian Vogt described method. This article was published by Siemens AG in 2008.

However, a particularly reliable assignment of structures of the calibration object to structures in the X-ray images is obtained in any case if a combined X-ray image is composed of the plurality of X-ray images, wherein the plurality of X-ray images are typically defined and / or the calibration object is selected in that the combined X-ray image then completely shows the calibration object. As part of the assignment, the combined X-ray image can thus be subjected to automatic structure recognition. For example, when using the calibration object described above, one sequence of four spheres may be imaged in the one x-ray image and in the other x-ray image a sequence of another four spheres which, when the combined x-ray image is formed, become a closed series of eight spheres. Then it can be detected better with the combined X-ray image than with the individual X-ray images, which ball sequence of the calibration object is so depicted. The structures of the calibration object thus imaged and recognized in the combined X-ray image can then Subsequently, in turn, the individual X-ray images are assigned, which were used to form the combined X-ray image.

in principle is to be expected that it overlaps with the individual X-ray images gives: The same structure is at the one edge of the x-ray image and at the opposite edge of the other X-ray image shown. The assembly should then ideally be done that does not accidentally display this same structure twice is, because then would have in the combined X-ray twice the same structure of the object to be calibrated as shown Structure to be assigned. The assembly of the X-ray images to a combined x-ray is then relieved if previously performed a so-called 2D 2D registration becomes. Upon registration, pattern recognition occurs, and the patterns from an image data set become the patterns from a assigned to another image data record. It then becomes a mapping rule determines the one imaginary image from the one image data set on the other image data set describes. When performing a 2D-2D registration for each pair of X-ray images (in particular at adjacent positions of the X-ray detector recorded X-ray images) is obtained by the 2D-2D registration an imaginary mapping rule, the the image from one of the x-ray images to the other describes the X-ray images of a couple. This thought Mapping instructions can then be used in the composition become. (The mapping rule found in the 2D-2D registration has nothing to do with the mapping rule of three-dimensional space to do on the two-dimensional X-ray image to which it is present ultimately works.)

alternative or in addition to the 2D 2D registration can structurally recognized structures then checked for a predetermined criterion whether they correspond to structures of the calibration object. Becomes for example, a series of large circles and small circles as a suspected illustration of big balls and small balls detected, so by a mistake in assembling a sequence to have emerged in the combined picture, which is not at all a real one Structure corresponds. Instead of the most appropriate structure to use, it is advantageous if in case of non-compliance of the tentatively recognized structure is not assigned, a "wrong" structure so not used. Only with complete correspondence becomes a preliminary recognized structure an X-ray image assigned.

As already executed, the calibration object comprises as a predetermined Structures a preferably spiral-shaped sequence of small ones Bodies, preferably balls, of different sizes. In order to recognize error structures of the calibration object which, during assembly, be it with or without 2D 2D registration, may occur then in a pictured sequence the small body of the distance adjacent imaged body and / or one of several imaged body defined angle can be used. exceeds the distance from the bodies to corresponding patterns in the combined X-ray image, for example, a maximum distance to another pattern that is part of a previously identified sequence is, the two patterns are not considered as belonging together considered. For example, lie the midpoints of several patterns on a line, and is another pattern off the line with a certain minimum distance or with certain Angle based on the last pattern from the previously recognized sequence, so this further, outlying pattern is not used, because this could be an effect of composition the X-ray pictures act. By using a distance or angle criterion or a similar criterion effectively avoiding that erroneous sequences be recognized, which are actually not shown.

following preferred embodiments of the invention below Referring to the drawing described in the

1 schematically illustrates the steps of a method for determining two mapping rules, as it is traversed in the prior art,

2 schematically illustrates the steps of a method for determining two mapping rules, as performed according to a first embodiment of the invention, and

3 schematically illustrates the steps of a method for determining two mapping rules, as carried out according to a second embodiment of the invention.

starting point is that an X-ray imaging system available in which the X-ray source and the X-ray detector are common can be moved, for example, when you at the X-ray C-arm are attached, and in addition one of these imaging Components, typically the X-ray detector, in each position can be moved again. In the case of an X-ray C-arc So would be the X-ray detector opposite the X-ray C-arm movable. In a robot-guided C-arm would be a procedure of the X-ray detector not necessary as long as the relative positions can be changed by tilting the detector are.

It Now, by the X-ray image recording system, a 3D reconstruction an object can not be obtained by individual X-ray images is completely detectable. Man manages to to take several pictures in a position of the X-ray C-arm, where in each case the position of the X-ray detector is different. In the totality of images is then each Part of the object to be imaged shown somewhere. Present will be assumed that two X-ray images are sufficient, which are recorded at a detector position A and a detector position B.

Around To create a 3D reconstruction, it is necessary to follow the imaging rules from the volume space of the object to the X-ray detector to know. Such a mapping rule, often referred to as a projection rule and by a projection matrix is obtained as part of a calibration.

you for this purpose places a calibration phantom in the x-ray image recording system as the object to be imaged will be placed later. In the present case, the calibration phantom should be a cylindrical plexiglas body in which a series of balls with two different sizes embedded by a binary number with "I" for a big ball and "0" for a small sphere specifiable sequence for each sequence of at least eight balls clearly assignable to a location in the calibration phantom is.

Such a calibration phantom is in the article of N. Strobel, B. Heigl, T. Brunner, O. Schütz, M. Mitschke, K. Wiesent, T. Mertelmeier: "Improving 3D Image Quality of X-Ray C-Arm Imaging Systems by Using Properly Designed Pose Deteration Systems for Calibrating the Projection Geometry ", Medical Imaging 2003; Physics of Medical Imaging, edited by Paffe, Martin J .; Antonuk, Larry E. Proceedings of the SPIE, Vol. 5030, pp. 943-954, 2003 described.

It Now exactly those positions are taken, which also later in obtaining 2D X-ray images of the subject of interest Image object should be taken. Then be in these positions Calibration images obtained, namely in step S10 a calibration image A at the detector position A and in step S10 'a calibration image B at the detector position B. In the prior art, it is for required, an additional calibration C at an intermediate position C according to step S10 "record.

The Calibration images A, B and C are subsequently in the steps S12, S12 'and S12' 'subjected to a pattern recognition, which is why is going to detect the balls of the calibration phantom, which are typically in the form of circles in the pictures show.

Now these balls are to be assigned unique sequences so that to determine the mapping requirement is clearly stated, which structures in space on which places in the respective pictures are shown.

in the In the prior art, the image C is used in a step S14 to to recognize such sequences. The following is based on the Steps of detecting the balls in steps S12 and S12 ' and recognizing the sequences in image C in step S14 - in particular using information about the respective displacement or tilting of the X-ray detector - a Assignment to sequences in the individual images A and B according to the Steps S16 and S16 'made. After having assigned such sequences are, in steps S18 and S18 ', the respective mapping rule derived to the image.

It is laborious if, as in step S10 '' is provided, always a further calibration image are recorded got to. If one wants the detector position A and B respectively at a plurality from positions of the X-ray C-arc, so must one complete passage of the X-ray C-arm through all his positions take place at intermediate position C. This Passing takes time that you would like to save.

The inventive method in all Embodiments is based on that no additional Calibration picture C is recorded, so the step S10 'and accordingly the Step S12 'omitted.

In a first embodiment of the method according to the invention, as in 2 1, sequence detection is performed in steps S14a and S14'a based on the individual calibration images A and B, respectively. The sequence recognition takes place independently of the respective other calibration image.

Novel methods for sequence detection make it possible to carry out steps S14a and S14'a in such a way that sequences are detected with high reliability. The novel methods are so-called "robust" methods, ie they take into account that errors can also occur. For example, in the course of steps S14a and S14'a, the method may be published from the filed after the filing date of the present application DE 10 2008 035 656 be performed. The method is based on the fact that not every initially assigned structure is finally adopted. Rather, a selection of a part of previously made assignments. This can be a mistake be taken into account. For example, a projection matrix is calculated on the basis of a selected part of assignments, and then this projection matrix is again applied to the calibration phantom. It is thus possible to detect a deviation error between an X-ray image generated by application of the projection matrix and the actual X-ray image on the basis of which an assignment is to be made. With regard to this error, a criterion can be set and, if appropriate, the assignment and / or its selection can be made anew each time until the predetermined criterion is met.

Just like that The method can also be used (additionally or alternatively) will be included in the Siemens AG publication entitled "Geometry Calibration for Arbitrary Trajectories" the author Dr. Florian Vogt is described. In this process a recognition of sequences is performed using minimum distances and minimum angles, thus ensuring that artifacts not mistakenly assigned to a sphere in the calibration images and then a sequence is considered recognized, even not shown. In the same way mistakes can be eliminated arising from the fact that the calibration phantom from a unfavorable perspective has been screened, so that several Subareas of the calibration phantom overlapping each other imaged on the same image area or sub-image area are.

In the second embodiment of the method according to the invention, as in 3 1, a combined image of the two calibration images A and B is generated in a step S22. Since the calibration images A and B typically have a small overlap area in the representation, they can be combined to form an image. One can imagine this as amateur photographers take a panorama: they gradually turn into different positions and take one picture at a time. The images can then be simply superimposed, and it creates approximately the panorama. And in order to perfect the composition of the calibration images A and B, a step S20 of the 2D-2D registration of the images A and B may precede, so that, due to the registration, a mapping rule from image A to B is known, which may be considered at step S22 , In step S14b, the sequences in the combined image are then recognized. This step can be carried out as in the prior art step S14: The combined image then replaces the specially recorded calibration image C. In particular, step S14b can also be the method of the DE 10 2008 035 656 and alternatively or additionally, the method of the article "Geometry Calibration for Arbitrary Trajectories" by Dr. Ing. Florian Vogt, published by Siemens AG , are used.

After this in step S14b, the sequences have been recognized, as in the state the technique, namely steps S16 and S16 'and subsequently the steps S18 and S18 '.

With the two by means of 2 or 3 described method is dispensed with a recording of an additional calibration image C, which does not correspond to an image, which is to be made later by an object to be imaged. This waiver saves time during calibration.

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 102006041033 A1 [0008]
• - DE 102007026115 A1 [0008, 0008]
• - DE 102007042333 A1 [0009]
• - DE 102008035656 [0016, 0037, 0039]

Cited non-patent literature

• - "Geometry Calibration for Arbitrary Trajectories" by Dr. Ing. Florian Vogt [0016]
• - N. Strobel, B. Heigl, T. Brunner, O. Schütz, M. Mitschke, K. Wiesent, T. Mertelmeier: "Improving 3D Image Quality of X-Ray C-arm Imaging Systems by Using Properly Designed Pose Determination Systems for Calibrating the Projection Geometry ", Medical Imaging 2003; Physics of Medical Imaging, edited by Paffe, Martin J .; Antonuk, Larry E. Proceedings of the SPIE, Vol. 5030, pp. 943-954, 2003 [0029]
• - "Geometry Calibration for Arbitrary Trajectories" by Dr. Ing. Florian Vogt, published by Siemens AG [0039]

Claims (7)

A method for determining a plurality of imaging specifications for the imaging of points in three-dimensional space on points in a two-dimensional X-ray image by an X-ray image recording system, wherein the plurality of imaging regulations are assigned the same position as an X-ray source and X-ray detector carrying X-ray C-arm as imaging components and wherein the method comprises the following steps: placing a calibration object with predetermined structures in the x-ray imaging system, at each of the plurality of relative positions of the imaging components in the same position of the x-ray C-arc acquisition (S10 , S10 ') in each case of a two-dimensional X-ray image of the calibration object, wherein the calibration object in the plurality of X-ray images is in each case incompletely displayed, for each X-ray image Zuo (S16, S16 '), which are shown in the predetermined structures in it, - due to the assignment Derive (S18, S18') of the respective mapping rule, characterized in that, for the purpose of assigning exclusively the plurality of X-ray images is used, in which the calibration object is incompletely displayed. The method of claim 1, wherein among the plurality X-ray images a combined X-ray image is composed (S22), and in the for the purpose of assigning the combined X-ray image of an automatic structure recognition (S14b) being thus recognized, in the combined X-ray image depicted structures of the calibration object the individual X-ray images be assigned (S16, S16 '), the formation of the combined X-ray image served. The method of claim 2, wherein prior to assembly (S22) of the combined X-ray image to X-ray images in pairs a 2D-2D-registration (S20) for the determination of an imaginary one Mapping rule showing the picture of one of the X-ray images on the other X-ray image of a couple describes performed is, wherein the imaginary mapping rule in the composition (S22) is used. A method according to claim 2 or 3, wherein in the Structure recognition preliminary recognized structures after a be checked for a predetermined criterion whether they correspond to structures of the calibration object and only in such Corresponding to an X-ray image to be assigned. Method according to one of the preceding claims, in which the calibration object as a predetermined structures one, preferably spiral, sequence of bodies, preferably Has balls, different size. Method according to claim 5 in its reference back to one of claims 2 to 4, wherein the recognition an imaged sequence of the bodies of the distance of adjacent imaged body and / or a mapped by several Body defined angle is used. The method of claim 1, wherein the assigning to each of the plurality of X-ray images is done separately.