IMPROVED MOTION ARTIFACT COMPENSATION THROUGH METAL ARTIFACT REDUCTION

Description

TECHNICAL FIELD OF THE INVENTION

[0001] The present invention relates to methods for digital volume tomography (DVT) in the dental field. The present invention particularly relates to the reduction of motion artifacts in a DVT image. BACKGROUND OF THE INVENTION

[0002] During a CBCT scan, the imaging components—X-ray tube and X-ray detector—face each other and rotate around the patient. This creates a sequence of X-ray projection images that form a sinogram. Using the projection geometry, a volume is reconstructed from the sinogram. The projection geometry describes the geometric properties of the CBCT device and its trajectory during the scan. It can be expressed using projection matrices.

[0003] In a CBCT scan, X-ray-opaque structures such as metals lead to image artifacts in the reconstructed volume. These image artifacts arise when the sensitivity of the X-ray detector is insufficient to physically accurately represent the X-ray attenuation. This causes problems, especially behind highly absorbent structures such as metals, where noise predominates. This leads to inconsistent values ​​in the reconstruction process, which often manifest as stripe artifacts around the metal areas, incorrect absorption values ​​in the metal areas, and incorrectly imaged metal contours in the volume.

[0004] If the patient moves during the CBCT acquisition or if the device calibration is outdated, the incorrect projection geometry is applied in the reconstruction process. This leads to motion artifacts in the reconstructed volume due to the offsetting of inconsistent data. A motion artifact compensation (MAC) procedure, or geometric calibration of the acquisition, can estimate the projection geometry from a given patient image. Metal artifacts interfere with the convergence and / or accuracy of the MAC, which estimates the projection geometry based on a similarity measure between a simulated sinogram of the reconstructed volume and the measured sinogram. The simulated sinogram is generated by projecting the artifact-affected volume. Metal artifacts lead to significant data inconsistencies.

[0005] Reference is also made to the following documents: Hahn Andreas et al, “Two methods for reducing moving metal artifacts in cone-beam CT,” Medical Physics., US, (201808), vol. 45, no. 8, doi:10.1002 / mp.13060, ISSN 0094-2405, pages 3671 - 3680, XP055872245A. Brehm M. et al, "Artifact-resistant motion estimation with a patient-specific artifact model for motion-compensated cone-beam CT", Medical Physics 2013 John Wiley and Sons Ltd USA, (2013), vol. 40, no. 10, doi:10.1118 / 1.4820537, XP012178423A. Maur Susanne, "Geometric autocalibration for dental volume tomography", Heidelberg, Germany, (20200622), pages 1 - 174, URL: http: / / www.ub.uniheidelberg.de / archiv / 28433, (20211213), XP055872274A. DISCLOSURE OF THE INVENTION

[0006] The aim of the present invention is to provide a method for the geometric calibration of a digital volume tomography (DVT) image in the dental field, which can enable improved motion artifact compensation (MAC) through metal artifact reduction (MAR).

[0007] This object is achieved by the method according to claim 1. The subject-matter of the dependent claims relates to further developments and preferred embodiments.

[0008] The method according to the invention is used for the geometric calibration of a DVT image in the dental field. It comprises the following steps: (S1) reconstruction of a first volume from a sinogram with an initial projection geometry; (S2) detection of the metal regions in the sinogram; (S3) correction of the metal regions in the sinogram; (S4) reconstruction of a second volume from the corrected sinogram from step (S3) with the initial or a varied projection geometry; (S5) geometric calibration by varying the projection geometry and evaluation based on a similarity measure between a simulated sinogram of the reconstructed second volume and the sinogram orthe corrected sinogram, wherein the simulated sinogram is calculated from the reconstructed second volume using the varied projection geometry; wherein at least a sub-area of ​​the following data: a) sinogram from step (S1); b) corrected sinogram from step (S3); c) simulated sinogram; d) intermediate result for calculating the similarity measure derived from one of said sinograms a)-c) is evaluated differently than the remaining areas of the data during the calculation of the similarity measure, wherein said sub-area includes the metal areas from step (S2).

[0009] A key feature of the present invention is the use of metal artifact correction to reconstruct the volume used in the MAC and the improvement of the MAC through special treatment, preferably filtering / weighting, of the metal regions during the calculation of the similarity measure. Another key beneficial effect of the present invention is the increased accuracy and faster convergence of the MAC. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In the following description, the present invention is explained in more detail using exemplary embodiments and with reference to the drawings, wherein Fig.1 - shows an overview of the method for geometric calibration of a CBCT image in the dental field according to one embodiment; Fig.2 - shows a flowchart according to another embodiment; Fig.3 - shows a flowchart according to another embodiment; Fig.4- shows a flowchart according to another embodiment; Fig.5 - shows a flowchart according to another embodiment; Fig.6 - shows a computer-assisted DVT system on which the method according to the invention can be carried out.

[0011] The reference numbers shown in the drawings indicate the elements listed below, which will be referred to in the following description of the exemplary embodiments. 1. DVT system 2. X-ray machine 3. X-ray tube 4. X-ray detector 5. Control unit 6. Head fixation 7. Bite block 8. Computer 9. Display Definitions

[0012] I 1 , I 2 :Projection images I :Mean of the image I N :Number of pixels in the projection image I ( u,v ):Pixel value of the projection image I at the point ( u,v ) s ( I 1 , I2 ):Similarity measure for projection images MSE ( I 1 , I 2 ):Mean Square Error MAD ( I 1 , I 2 ):Mean Absolute Difference NCC 1( I 1 , I 2 ):Normalized Cross Correlation NCC 2( I 1 , I 2 ):Pixel-based Normalized Cross Correlation GC 1( I 1 , I 2 ):Gradient Correlation 1 GC 2( I 1 , I 2 ):Gradient Correlation 2 GI 1( I 1 , I 2 ):Gradient Information GI 2( I 1 , I 2 ):Gradient Information with linear Scaling GO ( I 1 , I 2 ):Gradient Orientation Log :Laplace of Gaussian MI ( A , B ):Mutual Information

[0013] The method according to the invention is used for the geometric calibration of a DVT image in the dental field. As described in Fig. 2As shown, it comprises the following steps: (S1) reconstruction of a first volume (1) from a sinogram with an initial projection geometry; (S2) detection of the metal regions in the sinogram; (S3) correction of the metal regions in the sinogram; (S4) reconstruction of a second volume (2) from the corrected sinogram from step (3) with the initial or a varied projection geometry; (S5) geometric calibration by varying the projection geometry and evaluation based on a similarity measure between a simulated sinogram of the reconstructed second volume (2) and the sinogram orthe corrected sinogram, wherein the simulated sinogram is calculated from the reconstructed second volume (2) using the varied projection geometry, wherein at least a partial area of ​​the following data: a) sinogram from step (S1); b) corrected sinogram from step (S3); c) simulated sinogram; d) intermediate result for calculating the similarity measure derived from one of said sinograms a)-c) is evaluated differently than the remaining areas of the data during the calculation of the similarity measure, wherein said partial area includes the metal areas from step (S2).

[0014] Metals or artificial, X-ray-opaque structures form distinct contours in the sinogram. In the simulated sinogram, these contours are usually unclear or distorted due to the inaccurate projection geometry in the reconstruction process, the missing physical information in the metal regions, and the metal artifacts in the reconstructed volume. By using metal artifact reduction (MAR) in steps (S2) - (S4), the metal artifacts in the reconstructed volume can be reduced, but usually not completely or physically correctly restored. These inconsistent metal regions hinder the convergence of the geometric calibration, or motion artifact correction (MAC). By evaluating the metal regions differently during the calculation of the similarity measure, the convergence of the geometric calibration can be improved and, in many cases, accelerated, as well as its accuracy increased.

[0015] The partial areas in step (S5) are preferably larger than the detected metal areas from step (S2), so that artifacts at the edges of the metal areas can also be detected. The correction of the metal areas in the sinogram in step (S3) can also be performed on larger areas to compensate for inaccuracies in the detection of the metal areas in step (S2).

[0016] Figure 3shows a preferred embodiment in which the method comprises the following step after step (S5): (S6) repetition of steps (S4)-(S5). This serves to iteratively improve the MAC. The iterative repetition is terminated when a termination criterion, or convergence criterion, is reached. Possible termination criteria are: a) whether the change in the projection geometry is smaller than a threshold; b) whether the change in the final volume is smaller than a threshold; c) whether the number of iteration steps is greater than a threshold; d) whether the computing time is greater than a threshold.

[0017] In a further preferred embodiment, the method comprises the following step after step (S5): (S7) repetition of steps (S2)-(S5) or (S2)-(S6). This serves to iteratively improve the MAR by using the intermediate result of the MAC. This also requires an iterative improvement of the MAC. The iterative repetition is terminated when a second termination criterion, or convergence criterion, is reached. Possible second termination criteria are: a) whether the change in the projection geometry is smaller than a threshold; b) whether the change in the final volume is smaller than a threshold; c) whether the number of iteration steps is greater than a threshold; d) whether the computing time is greater than a threshold; e) whether the amount of metal in the volume or sinogram is greater than a threshold. Figure 4 shows this embodiment combined with previous preferred embodiments.

[0018] In a further preferred embodiment, the method comprises the following step after step (5): (S8) Reconstructing a third volume (3) taking into account the estimated projection geometry and applying a metal artifact correction. The third volume takes into account the results of the MAR and MAC correction methods. Figure 5 shows this embodiment combined with other preferred embodiments.

[0019] In a further preferred embodiment, the detection of the metal regions in the sinogram in step (S2) consists of: (a) detecting the metal regions in the first volume (1); and (b) projecting the detected metal regions into the sinogram. The detection of the metal regions in the volume has the advantage that it is usually more robust or easier to implement than direct detection of the metal regions in the sinogram. By applying the projection geometry, or the estimated projection geometry, the detected metal regions can be transferred to the sinogram. The metal regions detected in the volume can be enlarged to compensate for an inaccurate projection geometry and / or an inaccurate detection of the metal regions.

[0020] In a further preferred embodiment, the detection of the metal regions in the sinogram in step (2) consists of: (a') generating a simulated sinogram of the first volume (1); and (b') detecting the metal regions in the simulated sinogram from step (S2)(a) and transferring the detected metal regions to the sinogram. The detection of the metal regions is easier in the simulated sinogram than in the sinogram because a simplified representation of the superimposed structures is possible when generating the simulated sinogram. The metal regions detected in the simulated sinogram can be transferred to the sinogram because the projection geometries estimated for the sinogram were used when generating the simulated sinogram.

[0021] In a further preferred embodiment, the method comprises the following step after step (S4): (a) Determining the metal regions in the second volume (2) using the detected metal regions in the sinogram. The metal regions in the second volume are determined by backprojection or reconstruction of the detected metal regions in the sinogram using the projection geometry or the estimated projection geometry. This step is necessary if the metal regions have been detected in the simulated sinogram.

[0022] In a further preferred embodiment, the method comprises the following step after step (S4): (b) correcting the metal regions in the second volume (2), which in turn consists of one or more of the following steps: (b1) filling the metal regions in the second volume (2) with values ​​from the reconstructed first volume (1) or artificial values ​​indicative of the metal regions; (b2) weighted blending of the values ​​from step (S4)(b1) with the values ​​from the second volume (2). This has the advantage that the metal regions in the simulated sinogram in step (S5) are better suited for comparison with the sinogram and for calculating the similarity measure. Figure 5 shows this embodiment combined with other preferred embodiments.

[0023] In a further preferred embodiment, the correction of the metal regions in the sinogram in step (S3) consists of one or more of the following steps: (a) filling the metal regions in the corrected sinogram with new pixel values, which are either calculated from the neighboring pixels or correspond to artificial pixel values; (b) weighted blending of the new pixel values ​​from step (S3)(a) with the pixel values ​​of the sinogram. The metal regions in the sinogram cannot be physically measured correctly due to the strong absorption. By replacing them with plausible pixel values, the creation of metal artifacts during reconstruction can be avoided. The new, plausible pixel values ​​can be artificial values ​​or calculated from neighboring pixels in the sinogram. By weighted blending of the new pixel values ​​with the original pixel values, part of the uncertain physical information is preserved.

[0024] In a further preferred embodiment, the other evaluation of the subregions in at least one of the said data during the calculation of the similarity measure in step (S5) comprises a pixel-wise weighting, which preferably lies in the value range 0 to 1. This has the advantage that the metal regions and their surroundings are less important in the calculation of the similarity measure and thus do not hinder the convergence of the MAC or do so less strongly.

[0025] In a further preferred embodiment, the other evaluation of the subregions in at least one of the aforementioned data during the calculation of the similarity measure in step (S5) comprises local filtering. This has the advantage that the metal regions and their surroundings are less heavily influenced in the calculation of the similarity measure and thus do not impede the convergence of the MAC, or do so to a lesser extent.

[0026] In a further preferred embodiment, the projection geometry or the estimated projection geometry of the DVT image relative to the patient's head is described by geometric parameters. The projection geometry is preferably described with intrinsic parameters and extrinsic parameters, wherein the intrinsic parameters include the relative position between the X-ray source and the X-ray detector as well as their resolution, and the extrinsic parameters include a transformation consisting of rotation and translation for each projection image and selected sub-region of the volume.

[0027] In a further preferred embodiment, in addition to the image data, an initial projection geometry, e.g. in the form of device calibration data, can also be transferred as input data to the method for geometric calibration (step S5).

[0028] The method according to the invention is a computer-implementable method and can be carried out on a computer-assisted DVT system (1). Fig. 6shows an exemplary embodiment of a DVT system (1). For this purpose, the present invention also comprises a computer program with computer-readable code. The computer program can be provided on a data storage device. The computer-assisted DVT system (1) comprises an X-ray device (2) for carrying out the patient image, with which the sinogram is generated. The X-ray device (2) has an X-ray emitter (3) and X-ray detector (4) which are rotated around the patient's head during the image acquisition. The trajectory of the X-ray emitter (3) and the X-ray detector (4) during the image acquisition can describe a circular path. Alternatively, it can take on a different shape. By simultaneously controlling several actuators, a device trajectory around the patient's head which deviates from a purely circular path can be achieved. The patient's head is positioned in the X-ray device using the bite block (7) and the head fixation (6).The computer-assisted DVT system (1) comprises an operating unit (5), preferably a computer (8) or a processing unit that can be connected to the X-ray device (2), and preferably a display (9), among other things for visualizing the data sets. The computer (8) can be connected to the X-ray device (2) via a local network (not shown) or alternatively via the Internet. The computer (8) can be part of a cloud. Alternatively, the computer (8) can be integrated into the X-ray device (2). The calculations can alternatively take place in the cloud. The computer (8) executes the computer program and supplies the data sets, among other things for visualization on the display (9). The display (9) can be spatially separated from the X-ray device (2). The computer (8) can preferably also control the X-ray device (2). Alternatively, separate computers can be used for control and reconstruction.

[0029] According to the present invention, the data sets generated by the above-mentioned embodiments can be presented to a physician for visualization, in particular for diagnostic purposes, preferably by means of a display (9) or a printout Similarity measure

[0030] In the following, the similarity measures s ( I 1 , I 2 ) in detail. The similarity measure is a scalar quantity that measures the similarity between two projection images I 1 , I 2. Instead of similarity measures, negated difference measures can also be used. Similarity measures can be divided into two classes: pixel-based similarity measures and histogram-based similarity measures. Examples of pixel-based similarity measures:

[0031] a) Mean Square Error MSE I 1 I 2 = 1 N ∑ u , v ∈ I I 1 u v − I 2 u v 2 b) Mean Absolute Difference MAD I 1 I 2 = 1 N ∑ u , v ∈ I I 1 u v − I 2 u v c) Normalized Cross Correlation NCC 1 I 1 I 2 = ∑ u , v ∈ I I 1 u v − I 1 ¯ I 2 u v − I 2 ¯ ∑ u , v ∈ I I 1 u v − I 1 ¯ 2 ∑ u , v ∈ I I 2 u v − I 2 ¯ 2 d) Pixel-based Normalized Cross Correlation NCC 2 I 1 I 2 = ∑ u , v ∈ I I 1 u v − I 1 ¯ I 2 u v − I 2 ¯ I 1 u v − I 1 ¯ I 2 u v − I 2 ¯ e) Gradient Correlation 1 GC 1 I 1 I 2 = 1 2 NCC 1 ∂ I 1 ∂ u ∂ I 2 ∂ u + NCC 1 ∂ I 1 ∂ v ∂ I 2 ∂ v , with NNC1 from c). f) Gradient Correlation 2 GC 2 I 1 I 2 = 1 2 NCC 2 ∂ I 1 ∂ u ∂ I 2 ∂ u + NCC 2 ∂ I 1 ∂ v ∂ I 2 ∂ v , with NNC2 from d). g) Gradient Information GI 1 I 1 I 2 = 1 N ∑ u , v ∈ I g u v min ∂ I 1 2 ∂ u + ∂ I 1 2 ∂ v ∂ I 2 2 ∂ u + ∂ I 2 2 ∂ v mit g u v = θ u v + 1 2 und θ u v = ∂ I 1 ∂ u ∂ I 2 ∂ u + ∂ I 1 ∂ v ∂ I 2 ∂ v ∂ I 1 2 ∂ u + ∂ I 1 2 ∂ v ∂ I 2 2 ∂ u + ∂ I 2 2 ∂ v . h) Gradient Information with linear scaling GI 2 I 1 I 2 = 1 N ∑ u , v ∈ I g u v min ∂ I 1 2 ∂ u + ∂ I 1 2 ∂ v , α ∂ I 2 2 ∂ u + ∂ I 2 2 ∂ v with g u v = θ u v + 1 2 and θ u v = ∂ I 1 ∂ u ∂ I 2 ∂ u + ∂ I 1 ∂ v ∂ I 2 ∂ v ∂ I 1 ∂ u 2 + ∂ I 1 ∂ v 2 ∂ I 2 ∂ u 2 + ∂ I 2 ∂ v 2 and α as a scaling factor.

[0032] Here are g ( u, v ) and θ ( u , v ) Examples of intermediate results for calculating the similarity measure derived from at least one of the sinograms a)-c). Where a) denotes the sinogram from step (S1); b) the corrected sinogram from step (S3); and c) the simulated sinogram. i) Gradient Orientation GO I 1 I 2 = 1 max N , C 1 ∑ u , v ∈ I : ∂ I 1 ∂ u 2 + ∂ I 1 ∂ v 2 > t 1 ∩ ∂ I 2 ∂ u 2 + ∂ I 2 ∂ v 2 > t 2 2 − ln cos − 1 θ u v + 1 2 with θ u v = ∂ I 1 ∂ u ∂ I 2 ∂ u + ∂ I 1 ∂ v ∂ I 2 ∂ v ∂ I 1 ∂ u 2 + ∂ I 1 ∂ v 2 ∂ I 2 ∂ u 2 + ∂ I 2 ∂ v 2 and scalar constants C1, t1, t2. j) Laplace of Gaussian LoG ≜ Δ G σ u v = ∂ 2 ∂ u 2 G σ u v + ∂ 2 ∂ v 2 G σ u v = u 2 + v 2 − 2 σ 2 4 σ 2 e − u 2 + v 2 / 2 σ 2 with G σ u v = 1 2 πσ 2 exp − u 2 + v 2 2 σ 2 and the width of the Gaussian kernel σ. Examples of histogram-based similarity measures: k) Mutual Information

[0033] MI A B = ∑ a ∈ A , b ∈ B p a b log p a b p a p b with p ( a ) as the probability of the value occurring a in the introductory image I 1 , p ( b ) as the probability of the value occurring b in the introductory image I 2 , and p ( away ) as conditional probability for the occurrence of the values away in the introductory images I 1 and I 2 .

[0034] The other evaluation of the sub-areas in step (S5) during the calculation of the similarity measure can be realized, for example, as follows: a') MSE ( w I 1 , w I 2 ), where w is a pixel-wise weighting and w <=1 in the sub-areas and w=1 in the remaining areas. a") MSE w 1 f 11 I 1 + w 2 f 12 I 1 , w 1 f 21 I 2 + w 2 f 22 I 2 , where w 1 and w 2 pixel-wise weights and for each pixel w 1 + w 2 = 1 and f 11 ,f 12 , f 21 , f 22 local filters are. g') GI I 1 I 2 = 1 N ∑ u , v ∈ I w u v g u v min ∂ I 1 ∂ u 2 + ∂ I 1 ∂ v 2 ∂ I 2 ∂ u 2 + ∂ I 2 ∂ v 2 , where w is a pixel-wise weighting and w<=1 in the partial areas and w=1 in the remaining areas. k') for the mutual information calculation is p(a) calculated from (w I 1 ) and p(b) is calculated from ( w I 2 ), where w is a pixel-wise weighting [0,1].

Claims

1. Computer-implemented method for the geometric calibration of a CBCT image in the dental field, which comprises the following steps: (S1) reconstructing a first volume (1) from a sinogram with an initial projection geometry; (S2) detecting the metal areas in the sinogram; (S3) correcting the metal areas in the sinogram; (S4) reconstructing a second volume (2) from the corrected sinogram from step (S3) with the initial or a varied projection geometry; (S5) geometric calibration by varying the projection geometry and evaluation based on a similarity measure between a simulated sinogram of the reconstructed second volume (2) and the sinogram or the corrected sinogram, wherein the simulated sinogram is calculated from the reconstructed second volume (2) using the varied projection geometry, wherein at least a partial area of the following data: a) sinogram from step (S1); b) corrected sinogram from step (S3); c) simulated sinogram; d) the intermediate result for calculating the similarity measure derived from at least one of said sinograms a)-c) is evaluated differently than the remaining areas of the data during the calculation of the similarity measure, wherein said partial area includes the metal areas from step (S2).

2. Method according to Claim 1, characterized in that it comprises the following step after step (S5): (S6) repeating steps (S4)-(S5).

3. Method according to one of the preceding claims, characterized in that it comprises the following step after step (S5): (S7) repeating steps (S2)-(S5) or (S2)-(S6).

4. Method according to one of the preceding claims, characterized in that it comprises the following step after step (S5): (S8) reconstructing a third volume (3) taking into account the estimated projection geometry and applying a metal artifact correction.

5. Method according to one of the preceding claims, characterized in that the detection of the metal areas in the sinogram in step (S2) consists of: (a) detecting the metal areas in the first volume (1); (b) projecting the detected metal areas into the sinogram.

6. Method according to one of Claims 1-4, characterized in that the detection of the metal areas in the sinogram in step (S2) consists of: (a') generating a simulated sinogram of the first volume (1); (b') detecting the metal areas in the simulated sinogram from step (S2)(a) and transferring the detected metal areas to the sinogram.

7. Method according to Claim 6, characterized in that it comprises the following step after step (S4): (a) determining the metal areas in the second volume (2) using the detected metal areas in the sinogram.

8. Method according to Claim 5 or 7, characterized in that it comprises the following step after step (S4): (b) correcting the metal areas in the second volume (2) consists of one or more of the following steps: (b1) filling the metal areas in the second volume (2) with values from the reconstructed first volume (1) or artificial values indicative of the metal areas; (b2) weighted blending of the values from step (S4)(b1) with the values from the second volume (2).

9. Method according to one of the preceding claims, characterized in that correcting the metal areas in the sinogram in step (S3) consists of one or more of the following steps: (a) filling the metal areas in the corrected sinogram with new pixel values which are either calculated from the neighbouring pixels or correspond to artificial pixel values; (b) weighted blending of the new pixel values from step (S3)(a) with the pixel values of the sinogram.

10. Method according to one of the preceding claims, characterized in that the other evaluation of at least a partial area of said data during the calculation of the similarity measure in step (S5) comprises a pixel-wise weighting, which preferably lies in the value range 0 to 1.

11. Method according to one of the preceding claims, characterized in that the other evaluation of at least a partial area of said data during the calculation of the similarity measure in step (S5) comprises local filtering.

12. Computer program comprising computer-readable code which, when it is executed by a computerized CBCT system (1), prompts said system to execute the method steps of one of the preceding method claims.

13. Computerized CBCT system comprising an X-ray device (2) and a computing unit (8) which is configured to execute the computer program according to Claim 12.