WO2013076662A1

WO2013076662A1 - Spectral image processing in x-ray imaging

Info

Publication number: WO2013076662A1
Application number: PCT/IB2012/056594
Authority: WO
Inventors: Hanns-Ingo Maack; Jon Erik FREDENBERG; Björn NORELL; Carl Magnus ASLUND
Original assignee: Koninklijke Philips Electronics N.V.; Philips Intellectual Property & Standards Gmbh
Priority date: 2011-11-25
Filing date: 2012-11-21
Publication date: 2013-05-30

Abstract

The present invention relates to X-ray imaging providing spectral image information. In order to provide enhanced spectral image information, a method (100) for enhancing spectral X-ray imaging is described, comprising the following steps: An X-ray image comprising spectral image data is provided (102). Then, the spectral image data is transformed (103) into frequency representations. Next, channels representing different tissue structures in the frequency representations are identified (104). Then, a tissue specific modification of the X-ray image is applied (106), generating a tissue enhanced image (108). The modification comprises a weighting in the frequency representations. Further, the tissue enhanced image is provided (110).

Description

SPECTRAL IMAGE PROCESSING IN X-RAY IMAGING

FIELD OF THE INVENTION

The present invention relates to an X-ray image processing arrangement for enhanced spectral image information, an X-ray imaging system for providing enhanced spectral image information, a method for enhancing spectral X-ray imaging, as well as to a computer program element and a computer readable medium.

BACKGROUND OF THE INVENTION

X-ray imaging is used, for example, in mammography, e.g. for detecting breast cancer, or for assessing the risk of breast cancer development. Besides simple attenuation X-ray images, or fluoroscopy images, also spectral X-ray imaging, for example spectral mammography, is provided, for example for a material decomposition to separate different tissue types. As an example, a so-called glandularity map is provided. US 7,342,233 B2 describes producing an image containing spectral information, wherein an array of photon conversion channels is provided counting pulses in different ranges of strength. However, deriving tissue information from the X-ray data also has an effect on the signal-to-noise ratio and thus on other valuable image contents, for example. Thus, the achievable image information may distract from other also valuable information.

SUMMARY OF THE INVENTION

Thus, there is a need to provide enhanced spectral image information in spectral X-ray imaging.

The object of the present invention is solved by the subject-matter of the independent claims, wherein further embodiments are incorporated in the dependent claims.

It should be noted that the following described aspects of the invention apply also for the X-ray image processing arrangement for enhanced spectral image information, the X-ray imaging system for providing enhanced spectral image information, the method for enhancing spectral X-ray imaging, as well as to the computer program element and the computer readable medium.

According to a first aspect of the present invention, an X-ray image processing arrangement for enhanced spectral image information is provided, comprising an interface unit, a processing unit, and a display unit. The interface unit is configured to provide an X- ray image comprising spectral image data. The processing unit is configured to transform the spectral image data into frequency representations, and to identify channels representing different tissue structures in the frequency representations, and to apply tissue specific modification of the X-ray image, to generate a tissue enhanced image. The modification comprises a weighting in the frequency representations. The display unit is configured to provide the tissue enhanced image.

The term "spectral image" relates to X-ray image data relating to X-ray radiation of at least two different energies, for example in form of a high-energy (spectral-) image and a low-energy (spectral-) image.

For example, X-ray image processing arrangement is a mammography X-ray image processing arrangement, and the provided X-ray image is a mammography, or mammogram.

According, to a second aspect of the present invention, an X-ray imaging system is provided for providing enhanced spectral image information, wherein an X-ray source; an X-ray detector arrangement and an X-ray image processing arrangement according to the above described examples are provided. The X-ray source is configured to provide X- ray radiation towards the X-ray detector arrangement. The X-ray detector arrangement is configured to detect radiation with at least two different energies in order to provide the spectral image data.

According to an exemplary embodiment, the X-ray system is a mammography X-ray system and the X-ray image is a mammography image. The processing unit is configured to generate a glandularity image from the mammography image, and to provide different image contrast enhancements to different channels in the glandularity image. The processing unit is further configured to apply a larger contrast enhancement for the tissue enhanced image to image locations with more dense tissue than to image locations with less dense tissue.

According to a third aspect of the present invention, a method for enhancing spectral X-ray imaging is provided, comprising the following steps:

a) providing an X-ray image comprising spectral image data;

b) - transforming the spectral image data into frequency representations;

- identifying channels representing different tissue structures in the frequency representations; applying tissue specific modification of the X-ray image, generating a tissue enhanced image; wherein the modification comprises a weighting in the frequency representations; and

c) providing the tissue enhanced image.

For example, the identification of channels comprises the identification of frequencies relating to predetermined object sizes, e.g. for the handling of different structural sizes. Thus, structures relate to sizes.

The channels are also referred to as bands.

For example, the method for enhancing spectral X-ray imaging is a method for enhancing spectral mammography X-ray imaging.

According to an exemplary embodiment, the X-ray image is a mammography image and a glandularity image is generated from the mammography image. Different image contrast enhancement is provided to different channels in the glandularity image. For the tissue enhanced image, a larger contrast enhancement is applied to image portions with more dense tissue than to image portions with less dense tissue.

According to a further exemplary embodiment, channels of the X-ray image are scaled and de-noised, generating a control image. For the modification in step b), a multiplication of image data of the X-ray image with the control image is provided.

The term "channel" is also referred to as "sub-band", wherein the term "sub- bands" refers to splitting the image into different subsets in the domain of spatial frequency. For example, one option for this is the use of a Laplacian image decomposition that leads to "sub-bands" which differ in the range of spatial frequency they represent even though they are defined in the spatial domain. In this sense, adding the spectral-images at the end of the procedure to generate a full image is corresponding to a re-transformation to the spatial domain.

The channel may be provided as a sub-band of a Laplace pyramid derived by

Laplacian image decomposition.

According to a further exemplary embodiment, the following steps are provided:

bl) - for the transforming, a band pass filter is applied to the image, generating a band pass image;

- combining the band pass image with the control image by a multiplication of the band pass image with the control image, generating a combined image; and

- adding the image and the combined image, generating the tissue enhanced image. For example, the band pass filtering is selection of channels.

The addition step is a re-transformation by deregulating / reversing the channels to achieve the image, i.e. the tissue enhanced image.

According to a further exemplary embodiment, the following steps are provided:

b2) - for the transforming, a transformation generating a plurality of image bands is applied;

- combining the image bands with the control image by a multiplication of the control image with the respective image band, generating a plurality of outgoing image bands; and

- reconstructing the outgoing image bands with a retransformation from the frequency representations into an image, generating the tissue enhanced image.

For example, the transformation into the frequency representations comprises a transformation into spatial frequency domain, e.g. a Fourier transformation to the spatial frequency domain, and the reconstructing comprises a retransformation into the spatial domain.

As another example, the transformation into the frequency representations comprises Laplacian image decomposition leading to frequency representations in the spatial domain, and the reconstructing comprises a retransformation by reversing the frequency representations into the image.

According to a further exemplary embodiment, the following steps are provided:

b3) - dividing the image in at least the following spectral- images: a low energy spectral-image and a high energy spectral-image;

- for the transforming, the low energy spectral- image and the high energy spectral-image are each transformed into spectral-image channels relating to at least a highpass band and a lowpass band;

- combining the respective spectral-image bands of different spectral-images, wherein the combining comprises an application of a weighting and enhancement procedure, generating a plurality of outgoing enhanced image channels; and

- reconstructing the enhanced image bands with a retransformation from the frequency representations into an image, generating the tissue enhanced image.

As another example, the transformation into the frequency representations comprises a Laplacian image decomposition leading to frequency representations in the spatial domain, and the reconstructing comprises a retransformation by reversing the frequency representations into the image.

According to a further exemplary embodiment, in step b3), a predetermined division is applied to the control image, generating a plurality of control channels.

According to a further exemplary embodiment, for step b), different weighting factors are applied to the different frequency representations, wherein different weighting factors are assigned to different determined frequency channels of the image data, and wherein energy subtraction and/or energy weighting is applied to the different frequency channels according to the assigned weighting factors.

According to a further exemplary embodiment, the spectrally processed image is displayed in combination with a regular X-ray image, for example a regular mammogram. For example, the user can toggle between the spectrally processed image and the regular X- ray image.

According to a an aspect of the present invention, spectral image information contained in the X-ray image, e.g. the mammography image, is used for an assigned enhancement, wherein determined parts of the spectral image data are treated in a

predetermined manner, for example by enhancing the respective image content. Thus, the further modification, such as an enhancement, only affects the particular chosen portions of the spectral image data, whereas the respective rest of the spectral image data is not affected by the particular modification. Following, the image content provided in the selected portions is enhanced, without having negative effects on the other image content. For example, energy weighting can be applied for small structures, whereas for large structures, energy subtraction may be provided. According to the present invention, only the target tissue, i.e. the tissue of particular interest, is enhanced in contrast, the remaining tissue is unaffected. For example, this provides an improved visibility of glandular tissue. Further, also tumor conspicuity is improved without losing high frequency structures. For example, high frequency anatomical noise is suppressed. As a further example, an improved micro-calcification signal-to-noise ratio is provided without affecting low frequency structures. Thus, an intuitive visualization of the spectral information is provided to the user. For example, tissue spectral contrast enhancement is provided, for example based on a glandularity map, or on other kinds of material decomposition map. As a further example, feature specific image combination is provided. The selective image modification, for example enhancement, is achieved in the frequency representations, for example in the spatial frequency domain, to be able to determine and select particular frequencies relating to particular tissue parameters, for example.

These and other aspects of the invention will become apparent from and will be elucidated with reference to the embodiments described hereinafter.

BRIEF DESCRIPTION OF THE DRAWINGS

Exemplary embodiments of the invention will be described in the following with reference to the following drawings.

Fig. 1 schematically shows an X-ray image processing arrangement according to an exemplary embodiment of the present invention.

Fig. 2 shows an X-ray imaging system according to an exemplary embodiment of the present invention.

Fig. 3 shows basic steps of an exemplary method for enhancing spectral X-ray imaging according to the present invention.

Fig. 4 shows a further example of a method according to the present invention.

Fig. 5 shows a still further example of a method according to the present invention.

Fig. 6 shows a block diagram as an example of the method shown in Fig. 5. Fig. 7 shows a further exemplary embodiment of the method according to the present invention.

Fig. 8 shows a block diagram as an example of the method shown in Fig. 7. Fig. 9 shows a further example of a method according to the present invention. Fig. 10 shows a block diagram as an example of the method shown in Fig. 9. Fig. 11 shows a further exemplary embodiment of a method according to the present invention.

Fig. 12 shows a block diagram as an example of the method shown in Fig. 11. Fig. 13 shows a detectability graph.

Figs. 14A to 14H show examples of tissue enhancement according to the present invention.

Fig. 15 shows a further example of an algorithm layout according to an exemplary embodiment of the present invention.

Figs. 16A and 16B show examples of spectral image data.

DETAILED DESCRIPTION OF EMBODIMENTS

Fig. 1 shows an X-ray image processing arrangement 10 with an interface unit 12, a processing unit 14, and a display unit 16. The interface unit 12 is configured to provide an X-ray image comprising spectral image data. The processing unit 14 is configured to transform the spectral image data into frequency representations, and to identify channels representing different tissue structures in the frequency representations, and to apply tissue specific modification of the X-ray image to generate a tissue enhanced image. The modification comprises a weighting in the frequency representations. The display unit 16 is further configured to provide the tissue enhanced image.

The X-ray image processing arrangement may be built into an X-ray imaging system 11, for example provided as a mammography stand-up apparatus 18, shown in Fig. 2.

The system 11 comprises an X-ray source 20, and an X-ray detector arrangement 22. The X-ray source 20 is configured to provide X-ray radiation towards the X- ray detector arrangement 22. The latter is configured to detect radiation with at least two different energies in order to provide the spectral image data.

For example, the X-ray system 11 is a mammography system using multi-slit- technology and a photon counting detector.

In Fig. 2, the X-ray imaging system 11 is shown with a vertical support structure 24, to which the X-ray source 20 and the X-ray detector arrangement 22 are movably mounted such that they can be adjusted in height, as indicated with a first double arrow 26. The X-ray detector arrangement 22 comprises a first support surface 28, on which, for example, a patient can arrange a breast onto. Further, an adjustable compression plate 30, or compression paddle, is arranged above the first support surface 28, such that a breast can be arranged inbetween. The compression plate 30 is adjustable to adapt the distance between the first support surface 28 and the breast touching surface of the compression plate 30, as indicated with a second double arrow 32.

Further, a housing structure 34 is arranged in the vicinity, for example for receiving, i.e. housing, the interface unit 12 and the processing unit 14 of Fig. 1. Further, on top of the housing structure 34, a further box 36 very schematically indicates, for example, the display unit 16, shown in Fig. 1. It is noted that the (data and supply) connections between the different system parts are not further shown.

It is further noted, that the present invention is also related to other types of (mammography) X-ray imaging systems, for example imaging systems in which the patient is arranged lying on a patient support with the face looking downwards. It is further noted that a mammography X-ray imaging system may also comprise a movable structure to which the X-ray source and the X-ray detector arrangement are mounted.

The detector arrangement 22 detects photons which are weighted according to information content with higher weights for low energy photons, in order to optimize the signal-to-quantum-noise ratio. Thus, the X-ray system according to the present invention employs a photon counting detector that separates photons into two bins according to their energy. For example, this is provided with no additional dose to the patient, and also no risk of motion blur between the bins, and also with relatively good energy resolution. For example, the different energies provided by the X-ray radiation from the X-ray source are provided as a low energy image and a high energy image. This is achieved by providing the detector arrangement 22 as a photon counting detector separating photons into at least two energy categories. Of course, further energy categories than just low and high can be provided.

According to an example, the processing unit 14 is configured to generate a glandularity image from the mammography image, i.e. the X-ray image. The processing unit 14 is further configured to provide different image contrast enhancements to different channels in the glandularity image and to apply a larger contrast enhancement for the tissue enhanced image to image locations with more dense tissue than to image locations with less dense tissue. It must be noted that this will be described further in the following with reference to the figures showing examples of methods according to the present invention.

According to a further example, also described in further detail below, the processing unit is configured to divide the image in at least the following spectral- images: a low energy spectral-image and a high energy spectral-image. For the transformation, the processing unit 14 is further configured to apply a transformation of the low energy spectral- image and the high energy spectral- image each into channels relating to at least a highpass band and a lowpass band, and to combine the respective spectral-image bands of the different spectral-images, wherein the combining comprises the application of a weighting and enhancement procedure, generating a plurality of outgoing enhanced image channels. The processing unit is further configured to reconstruct the enhanced image bands with a retransformation from the frequency representations into an image, generating the tissue enhanced image.

In the following, basic steps of a method 100 for enhancing spectral X-ray imaging are explained with relation to Fig. 3. The method 100 comprises the following steps: In a first step 102, an X-ay image, e.g. a mammography image, is provided, comprising spectral image data. Then, in a transformation step 103, the spectral image data is

transformed into frequency representations. Next, in an identification step 104, an

identification of channels representing different tissue structures in the frequency

representations is provided. It is also provided a further step 106 of applying tissue specific modification of the X-ray image, such as the mammography image, generating a tissue enhanced image 108, wherein the modification comprises a weighting in the frequency representations. Further, in a second step 110, the tissue enhanced image is provided. For example, the tissue enhanced image 108 is shown on a display.

For example, the method 100 is provided for enhancing spectral mammography imaging.

The channels representing different tissue structures in the frequency representations may be identified as spatial frequency portions in a spatial frequency domain.

The tissue enhanced image uses spectral information, thus it is a spectrally enhanced image.

The first provision step 102 is also referred to as step a), the second step, comprising the transformation step 103, the identification step 104 and the application step 106, is also referred to as step b), and the second provision step 110 is also referred to as step c).

According to Fig. 4, showing an exemplary embodiment, the X-ray image is a mammography image, and a glandularity image 112 is generated in a generation step 114 from the mammography image. Different image contrast enhancements 116 are provided in a provision step 118 to different channels in the glandularity image 112. For the tissue enhanced image 108, a larger contrast enhancement is applied in the provision step 118 to image portions with more dense tissue than to image portions with less dense tissue.

For providing the glandularity image 112 (glandularity map), the following is noted. Assuming a breast that exclusively contains adipose and glandular tissue, the detected number of counts in a photon counting system is: where qo is the incident number of quanta, Φ is the (normalized) energy spectrum, μ_α and μ_ε are the linear attenuation coefficients for adipose and glandular tissue, g is the glandular fraction, and <¾ is the breast thickness. Subscript Ω is the energy bin index; in a non-energy resolved system, there is only one energy bin, whereas in a spectral system, there are at least two bins (Ω e {low, high}), ris the bin response function (ideally a rectangular function) and η is the quantum efficiency (ideally unity). If all parameters of the equation are known, it can be used to generate pixel values for each energy bin and for a range of thicknesses and glandularities. All pixel values can be gathered in a pixel value to glandularity and thickness lookup table which can be used to produce glandularity and thickness maps of the breast. Under a few simplifying assumptions, two energy bins are necessary and sufficient to accurately map into adipose and glandular space.

The glandularity map is used for modifying a contrast enhancement procedure. The glandularity map is thus generated by an energy separation of a mammogram.

According to Fig. 5, channels of the X-ray image, provided in step 102, are scaled and de-noised in an image modification step 120, generating a control image 122. For the modification in step b), in a further provision step 124, a multiplication 126 of image data of the X-ray image with the control image 122 is provided.

The control image 122 is a scaled and de-noised version of channels of the glandularity image. The control image is thus a result of weighting image frame with spectral means.

According to the example shown in Fig. 6, the following steps are provided. In a first step 128, a glandularity image 129 is provided. In a next step 130, the glandularity image 130 is mapped with a threshold factor. In still further step 132, a lowpass filter is applied for the image points below the threshold factor, thus generating a control image 134, also referred to with C(x, y). For example, for the mapping in step 130, the following is provided:

Map to [ ;l],

wherein a = minimum enhancement for fatty tissue, and wherein a < value < 1 = for dense tissue.

Thus, the control image has high values where tissue is dense. The X-ray image, for example, the mammography image, is then modified with the control image.

According to the example shown in Fig. 7, the following steps are provided in relation with step b). In an application step 136, for the transforming, a band pass filter is applied to the image, generating a band pass image 138. Further, in a combination step 140, the band pass image 138 is combined with the control image 122 by a multiplication 142 of the band pass image with the control image, generating a combined image 144. It is noted that the above described example for generating a control image is depicted in a simplified manner by a single frame 146 for the sake of visibility.

In an addition step 148, the image from step 102 and the combined image 144 are added, generating the tissue enhanced image 108.

The application step 136, and the combination step 140, and the addition step 148, being steps of step b), are also referred to as step bl).

For example, the applied band filter is a highpass filter to the image, generating a highpass image. The highpass image is combined with the control image by a multiplication of the highpass image with the control image, generating the combined image. The steps of step bl) are referred to as "un-sharp masking steps".

Fig. 8, relating to the exemplary method of Fig. 7, shows a provision 150 of an X-ray image 152, which is then subject to highpass filtering 154, resulting in highpass image 156. The highpass image 156 is then combined with the control image 122 by:

out-i = HP(x, y) *C(x, y)

and the result is added to the X-ray image 152 in an addition step 158. The provision of the X-ray image 152 is indicated with an arrow 160. As a result, the tissue enhanced image 108 is provided. The X-ray image may be provided in form of a mammography image.

The enhancements take place in the frequency domain, since the highpass is a kind of frequency domain. The Highpass is a predetermined frequency representations interval, for example a spatial frequency interval. Further, for example, by the addition in step 158, the "normal image" is re-established in the spatial domain.

Fig. 9 shows a further example of a method, in which for step b), the following is provided. In a first application step 162, for the transforming, a transformation generating a plurality of image bands 164 is applied to the image. Further, in a combination step 166, the image bands 164 are combined with the control image 122 by a multiplication 168 of the control image 122 with the respective image bands, generating a plurality of outgoing image bands 170. Further, a reconstruction step 172 is provided to reconstruct the outgoing image bands 170 with a retransformation from the frequency representations into an image, generating the tissue enhanced image 108.

The retransformation is also referred to as a reconstruction. The image bands are also referred to as spatial frequency intervals. The transformation to the frequency representations may be a Laplacian image decomposition , or a Fourier transformation to the spatial frequency domain. The retransformation may be a Fourier retransformation, also known as Fourier synthesis, or may be a Laplacian image decomposition, or re-mapping.

The application step 162, the combination step 166 and the reconstruction step 172 are thus steps of step b), and are thus referred to as b2), also referred to as "tissue enhancement steps".

A further example, relating to the method of Fig. 9, is shown in Fig. 10. First, the image is provided in a provision step 174. Next, the image is transformed to the frequency representations in transformation step 176, for example by a Laplacian image decomposition or Fourier transformation. This then leads to N image bands 178, for example to a first image band HPo for a highpass band, further bands BPi, BPi, and BP₇. Still further, a lowpass band LP₈ is provided. Of course, also less or more image bands 178 can be provided. Each of the image bands is then subject to the following multiplication, taking into account the result of the control image C(x, y) :

Thus, a number of outgoing image bands 180 image is provided, for example a first outgoing band OUT₀, OUTi, OUTi, OUT₇ and OUT₈, relating to the before-described ingoing bands 178. The bands are then reconstructed, as indicated with upwardly directed arrow 182, to provide the tissue enhanced image 108.

The image bands are modified in the frequency representations, for example, the spatial frequency domain.

Fig. 11 shows a further example of a method, in which for step b), the following steps are provided. In a step 184, the image of step 102 is divided in at least a low energy spectral- image 186, and a high energy spectral- image 188. Next, for the transforming, an application step 190 is provided, in which the low energy spectral- image and the high energy spectral-image are each transformed into spectral-image channels relating to at least a highpass band and a lowpass band. This is indicated with two separate connection lines 192 and 194. Next, a combination step 196 is provided in which the respective spectral- image bands of the different spectral- images are combined, wherein the combination step 196 comprises an application 198 of a weighting and enhancement procedure, generating a plurality of outgoing enhanced image channels 200, 202. Further, a reconstruction 204 is provided in which the enhanced image bands 200, 202 are reconstructed with a

retransformation from the frequency representations into an image, generating the tissue enhanced image 108.

The enhancement is thus integrated into the Laplace pyramid. The steps of the division step 184, the application step 190, the combining step 196, and the reconstruction 204 are also referred to as step b3), further also referred to as "tissue enhanced and energy weighting within the location pyramid".

According to a further example, although also shown in combination with Fig. 1 1 , but not being an essential part of the above described exemplary method of Fig. 1 1 , for the combination step 196, a predetermined division 206 is applied to the control image 122, generating a plurality of control image channels 208. The control image channels 208 are then supplied to the combination step 198 for providing the enhanced image channels also considering the information of the control image.

As mentioned above, the predetermined division 206 is provided as an option according to a further example which is indicated with dotted connection arrow 210.

The predetermined division 206 is a Gaussian transformation, for example.

The combining thus comprises a combination of enhancement and weighting. In the transformation application, the frequency space may be decomposed into partial frequencies o f the image .

For step b), different weighting factors are thus applied to the different frequency representations.

Fig. 12 shows a further schematic setup, relating to the exemplary method shown in Fig. 11. The X-ray image provided in step 102 is provided as a low energy image 212 and a high energy image 214. Both images 212, 214 are individually provided to a transformation to the frequency representations, for example a Laplacian image

decomposition 216, respectively 218. Thus, for the different channels, different image bands are provided, which are labelled in a similar manner, as before, only with the difference that the label indicates the channel, for example the low energy channel or the high energy channel.

Further, as an option, also the control image is provided in step 220, and then supplied to a division, for example a Gaussian division 222. Thus, respective control image bands CLP-0, CBP-1 . .. CBP-7, and CLP-8 are provided.

Next, each of the respective frequency bands, arranged in the horizontal arrangement in form of a line of a spreadsheet, whereas the channels are provided in columns, is then subject to a weighting and enhancement procedure 224:

(1 +/* CB_j) * [w„ * LBi + (1 - w_n) * HBi

j = i when i = i_min corresponds to a band that describes structures larger than a minimum diameter d_mi„ (for example 5 mm). j = imin if i represents a smaller structure than d_min.

As a result, outgoing enhanced image bands 226 are provided.

Next, the enhanced image bands for the outgoing part are then subject of a reconstruction 228 to generate the tissue enhanced image 108.

The different image bands (or frequency bands) are modified in the frequency representations, such as the spatial frequency domain.

According to a further example, although not further shown, for step b), different weighting factors are applied to the different frequency representations, wherein different weighting factors are assigned to different determined frequency bands of the image data. Energy subtraction and/or energy weighting is then applied to the different frequency bands according to the assigned weighting factors.

The X-ray image may thus be provided as a weighted sum of the spectral image data.

For example, energy subtraction is provided for large structures and energy weighting is provided for small lesions.

According to a further example, during the combining, weighting factors are applied which are adapted to a structure within the image bands, which is also explained with reference to Fig. 13.

The spectrally processed image is displayed in combination with a regular X- ray image, such as a mammogram. For example, they are arranged next to each other, or temporarily in an alternating manner.

According to a further example (not shown), the spectrally processed image is displayed in combination with a regular X-ray image, e.g. a mammogram, wherein the user can toggle between the spectrally processed image and the regular X-ray image, respectively mammogram.

Fig. 13 shows a graph 230 with a detectability index 232 on the vertical line, and a weighting factor 234 on the horizontal line. The detectability index is in this context a task-weighted signal-to-noise ratio that includes quantum noise and anatomical structure noise. A first curve 236 indicates the detectability of a tumor. A second line 238, in a dotted manner, indicates the detectability of micro-calcifications.

A small circle 240 indicates the situation for a conventional absorption image.

First, a small triangular 242 shows detectability index for the energy weighting scheme applied to the micro-calcification, which is slightly above the conventional absorption image. The detectability index of the tumor curve, indicated with a second small triangular 244, is, however, slightly below the detectability for the conventional absorption image for the energy weighting scheme. The squares indicate detectability for the energy subtraction scheme. For the micro-calcification it is at a minimum, indicated with a first square 246. A second square 248 indicates the respective maximum detectability index for the tumor curve, which is at a maximum. Thus, to achieve an optimum detectability, and thus an image with enhanced spectral information, energy subtraction or energy weighting, with corresponding weighting factors, has to be applied to the image depending on the detection target.

This is further described in "Observer Model Optimization of a Spectral Mammography System" by Erik Fredenberg, Magnus Aslund, Bjorn Cederstrom, Mats Lundqvist and Mats Danielsson, 2010, describes that energy subtraction and energy weighting are special cases of a weighted image combination with different signs on the weight factor. Energy subtraction reduces the anatomical noise, thereby improving the signal- to-anatomical-noise ratio, but simultaneously increases quantum noise. Energy weighting on the other hand improves the signal-to-quantum-noise ratio, but affects the anatomical noise in a similar way and is therefore inefficient or even deteriorating for situations that are dominated by anatomical noise. Anatomical noise is generally assumed to follow a power law (black noise), and mainly affects visibility of large structures. Quantum noise is flat over all spatial frequencies (white) and therefore affects small features to a larger extent.

"Observer Model Optimization of a Spectral Mammography System" proposes to use different weighting factors on the image depending on what the observer is looking for. Thus, Fig. 13 may be referred to as showing the trade-off between quantum and anatomical noise in relation to feature size.

Nevertheless, the observer generally is looking for both small and large structures within the same image and it is a drawback of the prior art that one has to optimize the image either for small or for large structures.

This problem is solved within the present invention. The two schemes, energy subtraction and energy weighting, are optimally applied to the same image by assigning different weight factors for different frequency bands. Thus, further improvements in order to work out, and thus to provide the spectral information provided by the mammography image data, is provided according to the present invention.

In the following, it is referred to Figs. 14A to 14H, showing example tissue enhancement. Fig. 14A shows a low energy image 250, and Fig. 14B shows a high energy image 252. Fig. 14C shows a quotient 254 of low/high image, and Fig. 14D shows a smoothed version 256 of the quotient 254. It is noted that dense tissue is dark. For example, this improved spectral image information may be referred to as an alternative glandularity map according to the present invention. Fig. 14E shows an original image 258 as a sum of a low and high image. The "dense tissue" is the structure that is depicted comparatively bright in this image. The crosshair cursor is located here. Fig. 14F shows a tissue enhanced version 260 according to the present invention. One can see the bright structures being even brighter due to the contrast enhancement that is active here. Fig. 14G shows a further version 262 of the image of Fig. 14E, which is also the case for Fig. 14H showing a further version 264 of the of Fig. 14F. However, Figs. 14G and 14H show the respective position of the cursor, which is indicated with two rectangular lines 266 and a respective box 268 for displaying values. Here the cursor is located in the "fatty tissue" which is not contrast enhanced at all in this example. So the dark parts of the two images 14G and 14H look the same and the values are identical, as the cursor demonstrates. In other words, the position of the cursor of Fig. 14E and 14F indicates the same value, which is also the case for the cursor position of Figs. 14G and 14H.

In the following, the calculation procedure is also addressed in more detail. The calculation of the combined image from two spectral images may be provided by:

M

WSUM w_n * LB + (1— %_;) - HB_{

The weighting factors w, can be adapted to the structure size within the image band. The correlation of structure size and image bands may be provided as follows:

Laplace band typical structure size MicroDose [mm] Example value w;

0 0.15 1.02

1 0.3 1.40

2 0.6 1.10

3 1.2 0.50

4 2.4 -0.25

5 4.7 -0.55

6 9.4 -0.55

7 18.8 -0.55

8 37.6 -0.55

The objective in defining the values is to match the properties that are described as: 1) adding the spectral images for small structures will improve the visibility of for instance micro-calcifications or small tumor structures slightly; and 2) subtracting the spectral images will optimize the visibility for larger structures, such as solid tumours.

For example, the weighting factors, or w_r values, are derived from the graph 230 shown in Fig. 13, thus, for a typical structure size of 0.3 mm, shown in the Laplace band 1, the value is set for 1.4. Further, as typical tumor size, a typical structure size of 5 mm is set in the Laplace band 5, for which the weighting factor value of -0.55 is taken from Fig. 13. Further, the other example values can be derived for example by interpolation. The example value wi could also be adapted to breast thickness by letting Wj vary with spatial position controlled by the thickness map. This means that Wi is a sub-band image of the proper size instead of a scalar value. Another option is to use a non-linear pre-processing of low and high such that adaption to breast thickness can be accomplished with a scalar w,.

It is noted that in instead of the Laplace band, which represents the division into a frequency space, or spatial frequencies of the image, also a Fourier space can be provided.

The spectral enhancement of the combined images is provided by:

N

ENH = ^ (1 + * CBj )

1 = 1

The factor/ controls the amount of spectral contrast enhancement. It is noted that the control image can be noisy and special conditions may be needed to be fulfilled for "j". This may be corresponding to the use of a lowpass filter applied to the "thickness map" as a further example of a material-decomposition map:

j = i when i = i_min corresponds to a band that describes structures larger than a minimum diameter d_mi„ (for example 5 mm).

j = imin if i represents a smaller structure than d_min.

The combination of enhancement and weighting can be expressed as:

N

D n I t = (1 + / - CBi) - [\ v_n * LBi + (1 - w½ ) ' HBj ]

i - 1

In other words, a tissue specific contrast enhancement is provided. Therefore, the glandularity is used within image processing to enhance the glandular tissue selectively. The glandularity image is scaled into an image that is 1.0 at fatty tissue locations and larger than 1.0 at dense locations. The denser the tissue is, the higher is the value. Further, different amount of image enhancement are applied to different frequency representations in the image.

According to one example, a Fourier transformation is provided and the enhancement is provided in the spatial frequency domain, with enhancement factors being dependent of the spatial frequency.

An alternative option provided is to use a multi-scale image decomposition, for instance a Laplace pyramid, which generates various daughter images representing different spatial frequencies or typical structure sizes.

A further approach is to do the enhancement based on un-sharp masking, see below. A further implementation is provided by integrating the enhancement into the location pyramid, for example, also see below.

According to the present invention, contrast enhancement is applied more at locations with dense tissue than elsewhere. Both, the regular image and the tissue enhanced images will be generated and the user can, for example, toggle between them at the viewing station.

In Fig. 15, an algorithm layout is shown in a diagram 270. As can be seen, a low energy image 272 and a high energy image 274 are provided. The two images are supplied to a fast Fourier transformation 276, thus achieving two different parts for the low energy and the high energy, indicated with two different rectangles 278, 280. A weighting factor 282 is applied to the respective part, and the two high and low energy parts are combined in an addition step 284. It is noted that the application of the weighting factor 282 is only shown for one, but also provided for the other part in the frequency domain. Next, an inverse fast Fourier transformation 286 is provided, thus generating a tissue enhanced image 288. Below, a graph 290 indicates an energy weighting factor W, with reference numeral 292, on the vertical axis, and a spatial frequency 294, also indicated with F, on the horizontal axis. A first line 296 indicates an exemplary selection for the weighting factor in relation with the spatial frequency. For example, for small structures, an energy weighting of W > 0, wherein 0 is indicated with a little horizontal line 298, is provided; and for large structures, an energy subtraction with a weighting factor of W < 0 is provided.

Fig. 16A shows a conventional non-energy resolved mammogram 300 with a lesion, inside circle 304, wherein the lesion is difficult to detect. Fig. 16B shows a corresponding spectral image 302 with image combination according to the energy subtraction scheme. The lesion shows up, but the image is severely affected by quantum noise and low-pass filtering is necessary. It is obvious that no fine structures are visible in this image and it would have benefitted from the present invention.

In another exemplary embodiment of the present invention, a computer program or a computer program element is provided that is characterized by being adapted to execute the method steps of the method according to one of the preceding embodiments, on an appropriate system.

The computer program element might therefore be stored on a computer unit, which might also be part of an embodiment of the present invention. This computing unit may be adapted to perform or induce a performing of the steps of the method described above. Moreover, it may be adapted to operate the components of the above described apparatus. The computing unit can be adapted to operate automatically and/or to execute the orders of a user. A computer program may be loaded into a working memory of a data processor. The data processor may thus be equipped to carry out the method of the invention.

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 up-date 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 fulfil the procedure of an exemplary embodiment of the method as described above.

According to a further exemplary embodiment of the present invention, a computer readable medium, such as a CD-ROM, 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.

A computer program may be stored and/or distributed on a suitable medium, such as an optical storage medium or a solid state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the internet or other wired or wireless telecommunication systems.

However, the computer program may also be presented over a network like the World Wide Web and can be downloaded into the working memory of a data processor from such a network. 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 a method according to one of the previously described embodiments of the invention.

It has to be noted that embodiments of the invention are described with reference to different subject matters. In particular, some embodiments are described with reference to method type claims whereas other embodiments are described with reference to the device type claims. However, a person skilled in the art will gather from the above and the following description that, unless otherwise notified, in addition to any combination of features belonging to one type of subject matter also any combination between features relating to different subject matters is considered to be disclosed with this application.

However, all features can be combined providing synergetic effects that are more than the simple summation of the features.

While the invention has been illustrated and described in detail in the drawings and foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive. The invention is not limited to the disclosed embodiments. Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing a claimed invention, from a study of the drawings, the disclosure, and the dependent claims.

In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single processor or other unit may fulfil the functions of several items re-cited in the claims. The mere fact that certain measures are re-cited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. Any reference signs in the claims should not be construed as limiting the scope.

Claims

CLAIMS:

1. An X-ray image processing arrangement (10) for enhanced spectral image information, comprising:

an interface unit (12);

a processing unit (14); and

- a display unit (16);

wherein the interface unit is configured to provide an X-ray image comprising spectral image data;

wherein the processing unit is configured to transform the spectral image data into frequency representations; to identify channels representing different tissue structures in the frequency representations; and to apply tissue specific modification of the X-ray image, to generate a tissue enhanced image; wherein the modification comprises a weighting in the frequency representations; and

wherein the display unit is configured to provide the tissue enhanced image.

2. An X-ray imaging system (11) for providing enhanced spectral image information, comprising:

an X-ray source (20);

an X-ray detector arrangement (22); and

an X-ray image processing arrangement (10) according to claim 1;

wherein the X-ray source is configured to provide X-ray radiation towards the

X-ray detector arrangement; and

wherein the X-ray detector arrangement is configured to detect radiation with at least two different energies in order to provide the spectral image data.

3. System according to claim 2, wherein the X-ray system is a mammography X- ray system and wherein the X-ray image is a mammography image; and

wherein the processing unit is configured to generate a glandularity image from the mammography image; and to provide different image contrast enhancements to different channels in the glandularity image; and to apply a larger contrast enhancement for the tissue enhanced image to image locations with more dense tissue than to image locations with less dense tissue.

4. System according to claim 2 or 3, wherein the processing unit is configured: - to divide the image in at least the following spectral- images: a low energy spectral-image and a high energy spectral-image;

for the transformation, to apply a transformation of the low energy spectral- image and the high energy spectral-image each into spectral-image channels relating to at least a highpass band and a lowpass band;

- to combine the respective spectral- image channels of different spectral- images; wherein the combining comprises the application of a weighting and enhancement procedure, generating a plurality of outgoing enhanced image channels; and

to reconstruct the enhanced image bands with a retransformation from the frequency representations into an image, generating the tissue enhanced image.

5. A method (100) for enhancing spectral X-ray imaging, comprising the following steps:

a) providing (102) an X-ray image comprising spectral image data;

b) - transforming (103) the spectral image data into frequency representations;

- identifying (104) channels representing different tissue structures in the frequency representations;

- applying (106) tissue specific modification of the X-ray image, generating a tissue enhanced image (108); wherein the modification comprises a weighting in the frequency representations; and

c) providing (110) the tissue enhanced image.

6. Method according to claim 5, wherein the X-ray image is a mammography image; and wherein a glandularity image (112) is generated (114) from the mammography image;

wherein different image contrast enhancements (116) are provided (118) to different channels in the glandularity image; and

wherein for the tissue enhanced image, a larger contrast enhancement is applied to image portions with more dense tissue than to image portions with less dense tissue.

7. Method according to claim 5 or 6, wherein channels of the X-ray image are scaled and de-noised (120), generating a control image (122); and

wherein, for the modification in step b), a multiplication (124) of image data of the X-ray image with the control image is provided (126).

8. Method according to claim 7, wherein the following steps are provided:

bl) - for the transforming, a band pass filter is applied (136) to the image, generating a band pass image (138);

- combining (140) the band pass image with the control image by a multiplication (142) of the band pass image with the control image, generating a combined image (144); and

- adding (148) the image and the combined image, generating the tissue enhanced image.

9. Method according to claim 7 or 8, wherein the following steps are provided: b2) - for the transforming, a transformation generating a plurality of image bands (164) is applied (162);

- combining (166) the image bands with the control image by a multiplication (168) of the control image with the respective image band, generating a plurality of outgoing image bands (170); and

- reconstructing (172) the outgoing image bands with a retrans formation from the frequency representations into an image, generating the tissue enhanced image.

10. Method according to one of the claims 7 to 9, wherein the following steps are provided:

b3) - dividing (184) the image in at least the following spectral- images: a low energy spectral-image (186) and a high energy spectral-image (188);

- for the transforming, the low energy spectral- image and the high energy spectral- image are each transformed into spectral- image channels relating to at least a highpass band and a lowpass band;

- combining (196) the respective spectral- image channels of different spectral- images; wherein the combining comprises an application (198) of a weighting and enhancement procedure, generating a plurality of outgoing enhanced image channels (200, 202); and

- reconstructing (204) the enhanced image bands with a retransformation from the frequency representations into an image, generating the tissue enhanced image.

11. Method according to claim 10, wherein in step b3), a predetermined division

(206) is applied to the control image, generating a plurality of control image channels (208).

12. Method according to one of the claims 5 to 11, wherein for step b), different weighting factors are applied to the different frequency representations;

wherein different weighting factors are assigned to different determined frequency channels of the image data; and

wherein energy subtraction and/or energy weighting is applied to the different frequency channels according to the assigned weighting factors.

13. Method according to one of the claims 5 to 12, wherein the spectrally processed image is displayed in combination with a regular X-ray image; and

wherein the user can toggle between the spectrally processed image and the regular image.

14. A computer program element for controlling an arrangement according to claim 1 , or a system according to one of the claims 2 to 4, which, when being executed by a processing unit, is adapted to perform the method according to one of the claims 5 to 13.

A computer readable medium having stored the program element of claim 14.