FIELD OF THE INVENTION

The present invention is related to a method and a device for providing X- ray images.

BACKGROUND OF THE INVENTION

It is known from X-ray images that there exists a compromise between image spatial resolution and noise level of the X-ray image. For example, in medical X- ray images spatial resolution depends on the detector pitch size as well as the X-rays' effective beam size. Either a large focal spot is used to get good signal to noise ratio images but their spatial resolution is poor, or sharp images are obtained through small focal spot acquisitions but at the expense of the noise level. In current systems, noise reduction is achieved through the use of spatial and / or temporal filtering techniques applied to the images or image sequences produced by those systems. However these image processing techniques do have some disadvantages such as that the noise reduction is often achieved at the expense of artifacts or signal deterioration. A combination of different focus acquisitions is known from the domain of optics, where the goal is for example to provide a photograph where all pixels are in-focus, as described in "All-in-focus photo image creation by wavelet transforms" by Shirai K, Nomura K and Ikehara M in: Electronics and Communications in Japan, part III, Fundamental Electronic Science, vol. 90, issue 3, pages 57-66, published 2007. A combination of two different focal spots by two electron sources in one device is known from US 6,104,781. Further, it is known from Ultrasound, to combine several images corresponding to different probe angulations in order to reduce the speckle noise and improve the image quality. This technique is described in US 6,464,638. But there may still be a need for improvement of image quality. SUMMARY OF THE INVENTION

The present invention aims at image quality improvement, in particular for noise reduction in X-ray images.

The object is reached with a method of providing images with an X-ray imaging system comprising an X-ray tube according to claim 1 and a system according to the independent device claim.

In an exemplary embodiment the method comprises the steps of acquiring at least a first and a second image of the same region with different focal spot sizes of the X-ray tube, wherein one of the first and the second images is acquired with a small focal spot size and the other image is acquired with a large focal spot size, and combining the two images, wherein the combination step includes decomposing the image data of the first image and the second image, determining the noise content of the decomposed image data, determining at least one weighting factor, depending on the signal content and noise content of the decomposed data of at least one of the two images, modulating the decomposed image data of at least one of the two images using the weighting factor and recomposing the decomposed image data of the at least two images to form a single new image.

One of the advantages of the invention is that the method provided allows to keep the good spatial resolution of the small focal spot image, i.e. to keep the sharpness even though the image is noisy, and the signal to noise ratio of the large focal spot image, which is less sharp but has a good signal to noise ratio. Contrary to this, known noise reduction methods only make use of the images' content for image quality improvement without resorting to specific acquisition techniques. For example, the so- called Laplacian multi-resolution pyramid has been used to reduce noise and to enhance contrast. Further, more sophisticated techniques like wavelet, ridglet, or curvelet transforms are used to achieve the same purpose. It is also known to use temporal filters, i.e. several images taken under the same acquisition conditions are combined so as to reduce the noise through temporal averaging. Artifacts or signal deterioration are often observed on the filtered images. According to the present invention, the good spatial resolution of the small focal spot image is used and denoised through a smart combination with the less noisy acquisition, i.e. the image acquired with a larger focal spot size.

The present invention relies on altering the acquisition procedure as well as combining this with an image processing, preferably a digital image processing. Thereby both opposite aspects in X-ray imaging, i.e. noise versus sharpness, are being considered. As a background one can say that the smaller the anode angle, the smaller the effective focal spot size, so the better the spatial resolution of the image, but also the smaller the field of view (FOV). For example, in cardiac applications the required maximum FOV is generally smaller than in vascular procedures. Hence, a smaller anode angle can be used, leading to a sharper image. Further, also reduction of the electron beam width leads to a smaller focal spot, but this goes at the cost of reduced loadability of the anode so a reduction of the generated X-ray intensity and of the signal to noise ratio of the detected X-ray image of the object. By combining at least two images with different focal spot sizes the final output image is improved as it provides both sharpness, as well as a good signal to noise ratio. It is noted that the effective focal spot size may also be called "focal size". But unfortunately this can lead to a misunderstanding since in X-ray images there are no different foci in the sense of the focus point or focus length of an optical lens system. It is rather a geometric principle that leads to blurr: In theory, if the X-ray generating source was punctual, the resulting image of an object would be infinitely sharp. In practice, the surface of the anode that produces the X-ray beam has a non-zero area. Thus, a point in-between the X-ray tube and the X-ray detector, and within the FOV, will be imaged at the detector level as a non-punctual shape, whose size depends on the X-ray source focal spot size, the anode angle with respect to point being imaged, and the distance ratio from the X-ray tube to the imaged point and from the imaged point to the X-ray detector.

Combining two or more images acquired with different focal spot sizes has the effect that using a phantom containing geometrical patterns as an object to be displayed, the image combination is not performed in a point-wise manner, i.e. independently of neighboring pixels. Rather, the image combination depends on the image content, which can easily be detected in the final output image displayed on a display. Furthermore, the final combination also depends on the noise level of the small focus image. Hence, according to the present invention, varying the noise level of the input small focal spot image consequently has the effect that a change in the final output image occurs.

To reduce the noise level without affecting the interesting signal, it is necessary to know the input noise level to be able to compare it with the signal level. Hence, the noise statistics estimation of the decomposed image data is an important step. The weighting factor used to combine both acquisitions depends on the signal and the noise level (Sigma) of the decomposed data. If the decomposed data > T * Sigma (T being the threshold fixed by the user) the pixel is determined to be signal. Hence, the weighting factor is high. That means the data of the small beam size image is kept in the combined data. If the decomposed data < T * Sigma, the pixel is considered to be noise. Instead of attenuating this pixel, the coefficient of the large beam size image is kept, which has a low noise level, so the weighting factor is set to a small value. In this exemplary embodiment, the sigma estimation is needed to denoise at the same level whatever the input noise level is. This is especially necessary in X-ray images for which the noise level is intensity dependent. That means the noise level varies with the intensity of the pixel in the image.

In an exemplary embodiment of the method the determining of the at least one weighting factor is depending on the image data of the small focal spot image. That is, the weighting factor, i.e. a number or an image, is computed from the decomposed image data of the small focal spot image and used to weight the combination of at least one of the two acquisitions. In another exemplary embodiment of the method, the determining of the at least one weighting factor is depending on the image data of the large focal spot image. This results in an image processing adaptable to the respective image content. In other words, the combining procedure is adaptable to the actual case depending on the image content shown.

In another exemplary embodiment of the method the weighting factor is evaluated or computed using a spatial analysis as well as a noise statistic estimation of the decomposed small focal spot image. The evaluation can be adjusted to the detailed image content of the first image, i.e. the image with small focal spot. In another exemplary embodiment of the method the spatial analysis includes the steps of detecting features within the decomposed image data of the small focal spot image, defining separate regions with the detected features using a predetermined threshold and using only the image data of the defined regions of the small focal spot image, for recomposing the new image.

In another exemplary embodiment of the method the input images are decomposed into sub-bands, wherein the weighting factor for the evaluation is determined by analyzing sub-band parameters of the image with small focal spot, to provide a better determination of the weighting factor considering the images' content in an improved way.

In another exemplary embodiment of the method external information is provided for determining the at least one weighting factor, to enable, for example, a manual or automated adjustment or setting of the weighting factor.

According to the invention, in an exemplary embodiment the object is also reached with an X-ray image system comprising an X-ray image generating device with an X-ray tube, a processing unit and a display. The image generating device is arranged to acquire a first and a second image with different focal spot sizes of the X-ray tube, wherein one of the first and the second images is acquired with a small focal spot and the other image is acquired with a large focal spot. The processing unit is arranged to combine the two images by a decomposition of the image data of the first image and the second image, to determine at least one weighting factor depending on the signal content and noise content of the decomposed data of at least one of the two images and a modulation of the decomposed image data of at least one of the two images using the weighting factor, and to recompose the decomposed image data of the at least two images, to form a single new image. Further, the display is arranged to display the new image.

The determining of the weighting factor comprises an estimation of the noise statistics of the decomposed image data, an analysis of the decomposed data of at least one image.

As mentioned before, combining two or more images acquired with different focal spot sizes has the effect that using a phantom containing geometrical patterns as an object to be displayed, the image combination is not performed in a point- wise manner, i.e. independently of neighboring pixels. Rather, the image combination depends on the image content, which can easily be detected in the final output image displayed on a display. Furthermore, the final combination also depends on the noise level of the small focal spot image. The use of the invention can thus easily be detected as well in the final output image by varying the noise level of the input small focal spot image.

In another exemplary embodiment of the present invention, a computer program element is provided that is characterized by being adapted to perform the steps of the method according to one of the preceding embodiments. This computer program element might therefore be stored on a computing unit, which might also be part of an embodiment of the present invention. This computing unit may be adapted to perform or induce the performing of the steps of the method described above. Moreover, it may be adapted to operate the components of the above-described system. The computing unit can be adapted to operate automatically and/or to execute the orders of a user.

This exemplary embodiment of the invention covers both a computer program that right from the beginning uses the invention and a computer program that by means of an update turns an existing program into a program that uses the invention.

Further on, the computer program element might be able to provide all necessary steps to fulfill the procedure of the method as described above.

According to a further exemplary embodiment of the present invention, a computer-readable medium is presented wherein the computer-readable medium has a computer program element stored on it which computer program element is described by the preceding section. According to a further exemplary embodiment of the present invention, a medium for making a computer program element available for downloading is provided, which computer program element is arranged to perform the method according to one of the previously described embodiments of the invention.

These and other aspects of the invention will be apparent from the exemplary embodiments described hereinafter with reference to the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 schematically describes an X-ray imaging device according to the invention;

Fig. 2 schematically shows the relation between generated beam size and focal spot size for a large effective focal spot and a small effective focal spot; and

Fig. 3 schematically shows the method steps according to the invention.

DETAILED DESCRIPTION OF EMBODIMENTS

Figure 1 schematically shows an X-ray imaging system 12 for providing X-ray images with an improved quality, particular regarding noise reduction in X-ray images. Note that the example shown is a so-called C-type system. Nevertheless, the invention also relates to other types of X-ray imaging systems. A source of X-ray radiation 14, i.e. an X-ray tube, is provided to generate X-ray radiation. Further, a table 16 is provided to receive a subject to be examined, and an X-ray image detection module 18 is located opposite the source of X-ray radiation 14, i.e. during the radiation procedure, the subject is located between the source of X-ray radiation 14 and the detection module 18. The latter is sending image data to a data processing unit 20, which is connected to both the detection module 18 and the source 14. Furthermore a display 22 is arranged in the vicinity of the table 16 to display information to the person operating the X-ray imaging system, i.e. a clinician. Preferably the display 22 is movably mounted in order for an individual adjustment depending on the examination situation. Also, an interface unit 24 is arranged to input information by the user.

Basically the image detection module 18 acquires images, i.e. generates images that are further processed in the data processing unit 20, said procedure being described more detailed in the following. As a result of the process in a preferred embodiment of the invention a single final output image is displayed on the display 22 to the clinician. For generating said final output image, the following steps are provided. Figure 3 presents the general processing scheme corresponding to these steps, whereas figure 2 schematically shows the relation between generated beam size and focal spot size.

In figure 2 one aspect of the present invention is shown in principle. In X- ray image acquisition, an electron beam 52 with a direction D and a beam size B is established between a cathode 54 and an anode 56. The anode 56 is arranged in an angle to the direction D of the electron beam 52. The electron beam 52 hits the anode 56 such that an X-ray beam 58 is emitted which is characterized by the focal spot size defining its focal spot size F. A large effective focal spot of the beam 58 results in a better signal to noise ratio. For more sharpness of the generated image a small effective focal spot is required. To reduce a large beam size F _L the generated electron beam can be reduced from a large electron beam size B _L to a small electron beam size B _s (see right part of figure 2) resulting in a small focal spot size F _s . Of course, it is also possible to vary the effective focal spot size by changing the angle of the anode 56 to a smaller anode angle (shown in middle part fig. X).

When having a large anode angle and a large size of the electron beam, which is also described as a long filament length, this results in a good field coverage and a large effective focal spot. It also results in a good power loading of the detector (left part in figure T). A small anode angle and a long filament length leads to a poor field coverage and a small effective focal spot. The power loading is good, too (middle part in figure 2)

Further, a large anode angle with a small filament length leads to a good field coverage and a small effective focal spot. The power loading is poor (see figure 2, right part).

Furthermore, in medical X-ray images spatial resolution depends on the detector pitch size as well as the X-rays' effective beam size, as mentioned above. As illustrated in figure 2, the effective beam size F is dictated by the angle between the anode 56 and the cathode 54 and by the electron beam size B from the cathode 54. The anode angle is mostly system specific and determines the maximum field of view in the object to be imaged. The smaller the angle, the smaller the effective beam size (so the better the resolution of the image) but also the smaller the FOV. For example, in cardiac applications the required maximum FOV is generally smaller than in vascular procedures, so a smaller anode angle can be used, leading to a sharper image.

Also, reduction of the electron beam width B leads to a smaller focal spot size F, which is illustrated in right part of figure 2 in comparison to left part in figure 2. But this reduction also means a reduced loadability of the anode. This means that the generated X-ray intensity and the signal to noise ratio of the detected X-ray image of the object are reduced. The electron beam width can be controlled electronically inside the X-ray tube depending on the best compromise between resolution and signal to noise ratio of the image.

One can say, that a compromise exists between the image spatial resolution and the noise level, i.e. noise versus sharpness. Either a large focal spot is used to get good signal to noise ratio images but their spatial resolution is poor, or sharp images are obtained through small focal spot acquisitions but at the expense of the noise level. It has shown, that reducing the noise in the small focal spot images, or conversely, increasing the sharpness of large focal spot images through image processing, is indeed a challenge, and the performance achievable is rather limited.

The present invention provides a new method for X-ray imaging, combining both small and large focal spot images, which is illustrated in figure 3 giving a general overview of the implementation of the invention. In a first acquisition step 112 at least two different input images are acquired by the image generating device (not shown). Further, the two images are processed in a combination procedure 114 to provide a single output image 116 as a final image, which shows improved final image quality. According to the invention, in the acquisition step 112 a first image 118 with a small focal spot F and a second image 120 with a large focal spot F are acquired. The combination procedure 114 itself can be subdivided into four different blocks 122, 128, 136 and 142, which are surrounded by a dotted line in figure 3 each.

In the first block 122 the input images 118, 120 are decomposed thanks to individual decomposition steps 124, 126 into sub-bands, such as multi-resolution sub- bands. These sub-bands can for example be the outputs of a Laplacian pyramid or a wavelet decomposition. This block 122 can also output only one sub-band equal to the input image (identity filter), because the following processing can be applied on the original image as well.

In the second block 128 the image data of the small focal spot image 118 is evaluated, to compute at least one weighting factor. The weighting factor is then used in the block 136 to combine the sub-bands of the small and large focal spot images 118, 120.

These weighting factors may depend on the sub-band content. Therefore, a first step 130 of this second block 128 comprises an analysis of the sub-band. This comprises a spatial analysis of the sub-band as well as a noise statistics estimation. This is applied in order to differentiate features from homogeneous noisy regions and in order to be able to adapt the sub-band combination as a function of their content and noise level.

A second step 132 comprises the calculation of the weighting factor, which is depicted with Fac. The weighting factor may by different for each pixel of the sub-band. For example, it will be near to "1" on highly contrasted features and near to "0" for homogeneous regions with no features.

In a typical application of the method, the analysis phase can be provided with a threshold on the sub-band versus noise ratio values, or with a linear/non-linear function of the sub-band and noise values. These operations can also be applied on a normalized gradient or Hessian of the sub-band.

Of course, other implementations are possible. The computed weighting factors may be independent of sub-band content. For example, considering the Laplacian pyramid decomposition in the first block, it is possible to exchange the high frequencies of the small and large focal spot image, independently of their content. Therefore, one implementation consists in using a weighting factor Fac equal to "1" for the first sub- band of the small focal spot image and to "0" for the one of the large focal spot image and the opposite for the other sub-bands. This method is simpler than the signal dependent combination, but obtains an output image with a larger high frequency noise. Of course, one different weighting factor is computed for each level of the decomposition. Once the weighting factors are computed or determined, the small {BandS _l ) and large (BandL _r ) focal spot sub-bands are combined:

Output ! = Fac i x BandS _r + (1 - Faci ) x BandL _r where Output _x is the combined sub-band _r> and Fac _l is the weighting factor for the sub-band _r .

Of course other combinations of the method steps (non- linear w.r.t the analysis output) can be involved. From a general point of view

Output, = F(BandS _1? BandL,, FaC ₁ ). Therefore the at least one weighting factor is provided to a modulation procedure 134 which is part of the third block 136. The modulation procedure 134 comprises applying the weighting factor to the decomposed image of the small focal spot image. Additionally a reciprocal 138 of the weighting factor is applied to the decomposed set of data of the large focal spot image in a second modulation procedure 140. Then the two new sets of decomposed image data are combined in an aggregation step 141.

Still further, the fourth block 142 comprises a recomposing step 144 of the image from the combined sub-bands. To a certain extend, this can be seen as an inverse step of the decomposition processes 124, 126. That means, if no weighting factors are applied between the decomposition and the reconstruction, this fourth block 142 returns the original image.

In the exemplary embodiment of the method shown, the weighting factor of each level of the decomposition are computed using a spatial analysis of at least the decomposed image data of the small focal spot image. The spatial analysis includes the steps of detecting features within the decomposed image data, defining separate regions with the detected features using a predetermined threshold, using the decomposed image data of the small focal spot image in the regions where features have been detected and the decomposed image data of the large focal spot image in other regions to construct a new set of decomposed image data and then recomposing the new image. Hence, with the method according to the invention it is possible to keep the good spatial resolution of the small focal spot image 118 only in regions where features are detected, and to preserve the good signal to noise ratio of the large focal spot image 120 in homogeneous areas or in areas with less features. It enables a noise reduction in the small focal spot image without the drawbacks of known noise reduction image processing techniques (artifacts, signal deteriorations). So far, the spatial combination of images with large focal spot 120 and small focal spot 118 has been described. According to the invention, temporal methods can be used in the same way, together with their combination with spatial methods.

Further, the invention has been described to be used on static images, i.e. only a pair of images with different focal spot sizes is needed. According to the invention, the method can also be applied on image sequences. For instance it is possible to pair small and large focal spot images of a sequence and continuously process them in pairs. For example, so-called "focal spot size swapping" can occur continuously. The images are then paired using an image with small focal spot and an image with and large focal spot. Further, they are processed with the method steps described-above per pair. Finally, the two images are composed to one resulting new image per pair, which is then deployed for a continuous display of the sequence.

With this embodiment it is also possible to compensate for body or organ motion occurring between the two different focal spot size images of one pair, or variously exploited during temporal filtering. Of course, more than two foci can be considered to extent the invention easily to multi-foci compounding.

The method of the invention can be applied to any X-ray systems where alternating the tube's focal spot size can be provided in a practical manner. As mentioned before, figure 1 shows an X-ray imaging device according to the invention. The source of X-ray radiation 14 is provided such that the image acquisition is possible with different focal spots sizes, for example with a small and a large focal spot. Hence, the detection module 18 provides respective image information, which is then further processed within the data processing unit 20 according to the method as describe above. Combining two or more images 118, 120 acquired with different focal spot sizes has the effect that using a phantom containing geometrical patterns as an object to be displayed, the image combination 114 is not performed in a point-wise manner. Rather, the image combination depends on the image content, which can easily be detected in the final output image 116 displayed on the display 22. Furthermore, the final combination also depends on the noise level of the small focal spot image. The use of the invention can thus easily be detected as well in the final output image by varying the noise level of the input small focal spot image.

While the invention has been illustrated and described in details in the drawings and forgoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive; the invention is not limited to the disclosed embodiments.