X-ray imaging system and method

The multi-energy X-ray imaging system addresses the trade-off in DAX and transmission image acquisition by generating a corrected transmission image with improved signal-to-noise ratio and reduced noise, effectively removing bone or soft tissue interference, thereby meeting clinical requirements and enhancing diagnostic accuracy.

JP7948188B2Active Publication Date: 2026-11-11KONINKLIJKE PHILIPS NV
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
JP2023542669
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2021-01-19
Filing Date
2022-01-17
Publication Date
2026-11-11
Estimated Expiration
2042-01-17

AI Technical Summary

Technical Problem

Conventional X-ray imaging techniques face challenges in achieving optimal acquisition parameters for both dark-field and phase-contrast images, particularly in chest X-ray imaging, due to the trade-off between low-energy DAX image acquisition for improved diagnostic accuracy and high-energy transmission image acquisition required by medical guidelines, which is compromised by rib visibility at lower energies.

Method used

A multi-energy dark-field and phase-contrast X-ray imaging system that acquires a series of low-energy phase-step images and at least one high-energy image, using phase retrieval and image processing to generate a corrected transmission image that meets clinical requirements, while simultaneously capturing DAX and PC images.

Benefits of technology

The system produces a clinically acceptable X-ray image with improved signal-to-noise ratio and reduced noise, allowing for accurate diagnosis by effectively removing bone or soft tissue interference, thus aligning with medical guidelines and enhancing diagnostic capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007948188000001
    Figure 0007948188000001
  • Figure 0007948188000002
    Figure 0007948188000002
  • Figure 0007948188000003
    Figure 0007948188000003
Patent Text Reader

Abstract

Concepts of multi-energy dark-field (DAX) and phase contrast (PC) x-ray imaging are proposed. One such concept involves collecting a set of low-energy images with phase stepping and collecting a set of high-energy images (with or without phase stepping). Phase retrieval is used to obtain transmission images, DAX images, and PC images from the phase-stepped low-energy images. Also, high-energy transmission images are obtained from the high-energy images. Based on the transmission images from the low-energy images and the transmission images from the high-energy images, a corrected transmission image is generated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of dark-field X-ray imaging and phase-contrast X-ray imaging. The present invention relates to a multi-energy dark-field and phase-contrast X-ray imaging system, a multi-energy dark-field and phase-contrast X-ray imaging method, and a related computer program product.

Background Art

[0002] Dark-field X-ray (DAX) imaging has recently demonstrated improved diagnostic accuracy in the diagnosis and staging of lung diseases. In conventional X-ray imaging techniques, an object is irradiated with X-rays to obtain a transmission image of the object. However, soft tissues such as blood vessels, cartilage, lung, and breast tissue that barely absorb X-rays exhibit lower attenuation contrast compared to bone images. To overcome this drawback, significant progress has been made in recent years in acquiring DAX images and phase-contrast (PC) images using X-rays.

[0003] Some of the techniques used for DAX and PC imaging are classified as grating interferometry or non-interferometric grating methods. In each of these techniques, an X-ray fringe pattern is generated after the X-rays pass through the object. Changes in the fringe pattern caused by the object are used to extract DAX and PC images.

[0004] DAX and PC imaging uses at least partially coherent emitted X-rays generated by a source grating arranged adjacent to and near an X-ray source such as an X-ray tube. Coherent X-rays penetrating the object enable subsequent acquisition of phase information. To obtain appropriate phase information, so-called phase stepping is performed.

[0005] In phase stepping, one (interference) grating is displaced laterally relative to another grating and X-ray detector by a fraction of its grating pitch (e.g., one-quarter, one-sixth, or one-eighth of the phase grating's pitch). When phase stepping is performed using a specific grating, the phase stepping must cover the entire period of this particular grating. Phase stepping techniques typically require the acquisition of 10 or more images to adequately suppress vibrational disturbances.

[0006] In typical radiographic imaging of human patients, ribs have low contrast at high kVp, so an X-ray energy of approximately 125 kVp is used. However, DAX imaging is more sensitive to lower X-ray energies, so an X-ray energy of approximately 70 kVp is used. At these lower X-ray energies, the ribs become visible to the extent that the bone interferes with the ability to analyze and interpret dark-field and radiographic images at these energies. [Overview of the project] [Problems that the invention aims to solve]

[0007] One advantage of DAX is that it is always acquired together with transmission images. Simultaneous acquisition of DAX and transmission images has the disadvantage of requiring trade-offs in acquisition parameters. In particular, DAX image acquisition is usually optimal at X-ray energies of 60kVp to 70kVp because the DAX signal generated by lung tissue decreases rapidly with increasing X-ray energy. Furthermore, current medical guidelines state that chest X-rays should be performed at energies of ≥120kVp. This is because the ribs have low attenuation contrast at higher kVp levels. [Means for solving the problem]

[0008] The present invention is defined by the claims.

[0009] According to an embodiment of one aspect of the present invention, a multi-energy dark-field (DAX) and phase-contrast (PC) X-ray imaging system is provided. The system includes an imaging unit which acquires a plurality of phase-step X-ray images at a first set of energy values ​​and acquires at least one X-ray image at a second set of energy values, wherein each energy value in the first set is smaller than each energy value in the second set, and a processing unit which processes the plurality of phase-step X-ray images acquired at the first set of energy values ​​using phase retrieval to acquire a first transmission image, a DAX image, and a PC image; processes at least one X-ray image acquired at the second set of energy values ​​to acquire a second transmission image; and processes the first transmission image based on the second transmission image to generate a corrected transmission image.

[0010] This system collects a set of low-energy images using phase stepping and a set of high-energy images (with or without phase stepping). Using phase retrieval, transmission images, DAX images, and PC images are acquired from the phase-stepped low-energy images. High-energy transmission images are acquired from the high-energy images.

[0011] A modified transmission image is generated based on the transmission image from the low-energy image (i.e., the first transmission image) and the transmission image from the high-energy image (i.e., the second transmission image).

[0012] By basing the modified radiograph on a second radiograph, the modified radiograph can retain the characteristics of the second radiograph (e.g., a high signal-to-noise ratio), making it appear as if it were acquired at a high energy (e.g., 125 kVp), thereby meeting the clinical requirements regarding the acquisition energy for chest X-ray imaging.

[0013] Furthermore, by also basing the correction on the first transmission image, the corrected transmission image can retain the features of the first transmission image (e.g., image framing). Since the system can collect low-energy DAX and PC images simultaneously with the first transmission image, the low-energy DAX and PC images can be immediately compared with the corrected transmission image. This is because the image framing is the same in all cases (i.e., the images are aligned / aligned to each other).

[0014] Furthermore, by relying on both the first and second transmission images, high-energy transmission image features and low-energy transmission image features can be compared (e.g., attenuation contrast). The processing unit obtains segmentation information from this comparison and can use this information to remove high-attenuation features (e.g., bone) or, conversely, low-attenuation features (e.g., soft tissue) from the image.

[0015] For example, a corrected radiographic image is a combination of high-energy and low-energy radiographic images such that the system can reduce noise in the corrected radiographic image and / or remove bone or soft tissue. The corrected radiographic image is matched with a DAX image and / or PC image having the same subject and framing as the first radiographic image, and the corrected radiographic image is clinically acceptable.

[0016] These and other aspects of the present invention will become apparent from the embodiments described below and will be explained with reference to those embodiments.

[0017] In some embodiments, the imaging unit includes a grating interferometer, multiple (periodic) apertures, a propagation-based imaging setup, or a rotating crystal that acquires multiple phase-stepped X-ray images at a first set of energy values. The phase-stepped image sequence can be acquired with various hardware (a grating interferometer for grating-based imaging, multiple apertures for edge illumination, a propagation sensor for propagation-based imaging, or a rotating crystal for analyzer-based imaging). Embodiments utilizing different hardware can offer advantages such as improved system sensitivity for specific patients, improved phase-stepping control, and improved spatial resolution, among others.

[0018] In some embodiments, the first set of energy values ​​consists of energy values ​​less than 100 kVp, preferably in the range of 60 kVp to 70 kVp. Thus, the above phase-step image sequence is acquired at least one low-energy state for optimal DAX image acquisition, which is 60 to 70 kVp, but is not limited to this range.

[0019] In some embodiments, the second set of energy values ​​consists of energy values ​​of 100 kVp or higher, preferably 120 kVp or higher, and more preferably at least one energy value that is substantially equal to 125 kVp. Thus, the above at least one X-ray image is acquired at at least one high-energy state for optimal transmission image acquisition, above 120 kVp, ideally at 125 kVp.

[0020] In some embodiments, the processing unit processes at least one of a plurality of phase-step X-ray images and / or at least one X-ray image to remove fringes according to a fringe removal (e.g., phase stepping) algorithm. Furthermore, in some embodiments, the imaging unit processes at least one of a plurality of phase-step X-ray images and / or at least one X-ray image to remove residual fringes according to a fringe removal algorithm using an X-ray beam spectral filter and / or a dynamic interference grating.

[0021] Fringes arise from acquiring images using a DAX system that performs phase stepping using grating interferometry. Therefore, embodiments remove or reduce such fringes. For example, embodiments acquire object-free phase-stepped X-ray images to acquire fringe information. However, fringe information may be acquired by other means. A processing unit can execute a fringe removal algorithm using the fringe information. Some embodiments can remove fringes using physical components of the imaging system. That is, fringes can be removed by moving the interference grating, or by using a spectral filter to prevent fringes from occurring in the first place. Removing fringes improves the contrast, clarity, and overall quality of the X-ray image.

[0022] In some embodiments, the processing unit processes the first transmission image based on a second transmission image to reduce noise in the corrected transmission image. DAX images acquired at optimal low energy values ​​are useful in diagnosing lung diseases. However, transmission images acquired simultaneously at low energy values ​​have an insufficient signal-to-noise ratio and do not conform to medical guidelines. Therefore, the processing unit uses information from the high-energy transmission image to improve the signal-to-noise ratio in the low-energy transmission image. The resulting corrected transmission image becomes a clinically acceptable "virtual" 125kVp image.

[0023] In some embodiments, the processing unit processes the first transmission image based on a second transmission image to remove soft tissue or bone from the modified transmission image. Techniques such as image segmentation are used to determine the difference in attenuation contrast of high-attenuation features (i.e., bone) between the low-energy and high-energy transmission images. This difference is then used to remove bone or soft tissue from the X-ray image. This is beneficial in diagnosing lung diseases because it reduces irrelevant information in the image processed by the clinician. It can also be used to identify features in the transmission image that may be obscured by high-attenuation features.

[0024] In some embodiments, the processing unit further executes a training algorithm. This training algorithm creates a trained neural network that receives a training input and a known output, the training input comprising a first transparent image and the known output comprising a modified transparent image. Furthermore, in some embodiments, the processing unit further executes an image processing algorithm. The image processing algorithm uses the trained neural network to receive an input comprising a first transparent image and generate an output comprising a modified transparent image.

[0025] The difference between the first and second transmission images is not an extremum, and the neural network is trained to output a corrected transmission image from the system without needing to acquire the second transmission image. In some embodiments of the neural network, the neural network is trained for noise reduction, bone or soft tissue removal, and / or a combination of these two actions. These deep learning methods increase the rate at which the system outputs clinically acceptable multi-energy X-ray images. Furthermore, these deep learning methods simplify the inputs required for the system to perform multi-energy DAX and PC imaging.

[0026] Another aspect of the present invention provides a multi-energy dark-field (DAX) and phase-contrast (PC) X-ray imaging method. The method includes the steps of: acquiring a plurality of phase-step X-ray images at a first set of energy values ​​and at least one X-ray image at a second set of energy values, wherein each energy value in the first set is smaller than each energy value in the second set; performing a phase retrieval process on the plurality of phase-step X-ray images acquired at the first set of energy values ​​to acquire a first transmission image, a DAX image, and a phase-contrast (PC) image; processing at least one X-ray image acquired at the second set of energy values ​​to acquire a second transmission image; and processing the first transmission image based on the second transmission image to generate a corrected transmission image.

[0027] In some embodiments, the step of processing the first transmission image comprises the step of reducing noise in the first transmission image based on the second transmission image.

[0028] Furthermore, in some embodiments, the second transmission image is used to adapt the contrast between different materials (e.g., soft tissue and bone) in the first transmission image. Accordingly, in some embodiments, the step of processing the first transmission image comprises the step of removing soft tissue or bone from the first transmission image based on the second transmission image.

[0029] In some embodiments, the first set of energy values consists of energy values less than 100 kVp, preferably energy values in the range of 60 kVp to 70 kVp.

[0030] According to another aspect of the present invention, there is provided a computer program product comprising computer program code means which, when executed on a computing device having a processing system, causes the processing system to perform all the steps of the method according to the presented embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] For a better understanding of the present invention, and to show more clearly how the present invention may be carried into effect, reference will be made by way of example only to the accompanying drawings.

[0032] [Figure 1] Figure 1 is a flow diagram of a multi-energy DAX imaging system. [Figure 2] Figure 2 is an example of a transmission X-ray image and an example of a dark-field X-ray image. [Figure 3] Figure 3 is a schematic diagram of an imaging unit according to one or more embodiments of the present invention. [Figure 4] Figure 4 is a flow diagram of a multi-energy DAX imaging method according to one or more embodiments of the present invention. [Figure 5]Figure 5 is a flowchart of an image processing algorithm that uses a trained neural network to output a corrected X-ray image. [Modes for carrying out the invention]

[0033] The present invention will be described with reference to the figures.

[0034] The detailed descriptions and specific examples illustrate exemplary embodiments of the apparatus, system, and method, but should be understood to be for illustrative purposes only and not intended to limit the scope of the invention. These and other features, aspects, and advantages of the apparatus, system, and method of the present invention will be better understood from the following description, the appended claims, and the appended drawings. It should be understood that the figures are schematic diagrams and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the figures to indicate the same or similar parts.

[0035] A multi-energy dark-field (DAX) and phase-contrast (PC) X-ray imaging system has been proposed. This system includes an imaging unit that acquires multiple phase-step X-ray images at a first set of energy values ​​and acquires at least one X-ray image at a second set of energy values, where each energy value in the first set is smaller than each energy value in the second set. Furthermore, the multi-energy DAX imaging system includes a processing unit that processes the multiple phase-step X-ray images acquired at the first set of energy values ​​using phase retrieval to acquire a first transmission image, a DAX image, and a phase-contrast image. The processing unit further processes at least one X-ray image acquired at a second set of energy values ​​to acquire a second transmission image, and processes the first transmission image based on the second transmission image to generate a corrected transmission image.

[0036] The embodiments provide a concept of multi-energy dark-field (DAX) and phase-contrast X-ray imaging. This includes the acquisition of a series of images using phase stepping at low kVp, the acquisition of one (or more) additional images with or without phase stepping at high kVp, phase retrieval of the low kVp stepping series (resulting in DAX images and low kVp transmission images), stripe removal from high kVp transmission images, e.g., combination of transmission images, denoising of high kVp transmission images using additional structural information taken from low kVp transmission images, and / or dual-energy combination of low kVp and high kVp transmission images using state-of-the-art techniques for bone or soft tissue removal. Thus, the concept of combining DAX image acquisition with multi-energy transmission image acquisition is proposed. In general, the proposed embodiments are not limited to only two-energy acquisition. That is, in principle, each new step of phase stepping can be acquired at a different energy. In this case, phase retrieval is performed considering all images at once, while taking into account the effect of X-ray energy on the visibility (performance) of the grating interferometer.

[0037] DAX imaging includes an X-ray source and a configuration that can measure the magnitude of X-ray diffraction passing through an object, rather than the attenuation of X-rays passing through the same object.

[0038] Phase stepping in X-ray imaging is a process known to those skilled in the art and involves determining phase information about an object. In particular, phase stepping is performed by moving the grating laterally relative to the sensor over a distance equal to the period of the grating. In other embodiments, phase stepping is performed by moving the sensor laterally relative to the grating over a distance equal to the period of the grating. This is done in steps shorter than the period of the grating, so that when all steps are added together, the total distance moved is equal to the period of the grating. In yet another embodiment, phase stepping is performed using electromagnetic phase stepping, in which an electromagnet is used to move the target of the X-ray beam after the X-ray beam has been emitted from the X-ray source. The movement of the target of the X-ray beam is lateral relative to the sensor over a distance equal to the period of the (phase) grating.

[0039] "Energy" or "energy value" refers to the peak potential, peak voltage, and / or maximum voltage applied to both ends of an X-ray tube to generate X-rays, and is measured in kVp, i.e., peak kilovoltage. The higher the kVp, the greater the energy of the X-rays emitted from the X-ray source. For example, a low-energy image refers to an X-ray image collected at a lower kVp compared to a high-energy image, and conversely, a high-energy image refers to an X-ray image collected at a higher kVp compared to a low-energy image. Low-energy images typically have a low signal-to-noise ratio and low attenuation contrast in transmission images, while DAX images have a high signal-to-noise ratio. High-energy images typically have a high signal-to-noise ratio and high attenuation contrast in transmission images, while DAX images have a low signal-to-noise ratio.

[0040] Referring to Figure 1, a simplified block diagram of a system 100 for multi-energy DAX and PC X-ray imaging according to one embodiment is shown. The system includes an X-ray imaging unit 102 and a processing unit 112.

[0041] The imaging unit 102 acquires multiple phase-step X-ray images 108 at a first set of energy values ​​104 and acquires at least one X-ray image 110 at a second set of energy values ​​106, where each energy value in the first set 104 is smaller than each energy value in the second set 106.

[0042] For example, the first set of energy values ​​104 consists of energy values ​​less than 100kVp, preferably in the range of 60kVp to 70kVp (i.e., low energy values). In this way, the first set of energy values ​​104 is configured for optimal acquisition of the DAX image 114. Optimal acquisition of the DAX image 114 occurs when the signal-to-noise ratio is maximized. The first set of energy values ​​104 is determined automatically, by a clinician, or by default values ​​for optimal acquisition of the DAX image 114.

[0043] The second set of energy values ​​106 consists of energy values ​​of 100 kVp or higher, preferably 120 kVp or higher, and more preferably at least one energy value that is substantially equal to 125 kVp (e.g., 125 kVp ± 1%, or equal to 125 kVp) (i.e., high energy values). In this way, the second set of energy values ​​106 is configured for optimal acquisition of the second transmission image 416. Optimal acquisition of the second transmission image 416 occurs when the signal-to-noise ratio is maximized and / or when the second set of energy values ​​106 conforms to medical guidelines, i.e., when it exceeds 120 kVp. The second set of energy values ​​106 is determined automatically, by a clinician, or by a default value for optimal acquisition of the second transmission image 416.

[0044] The imaging unit 102 further provides the processing unit 112 with a plurality of phase-step X-ray images 108 with a first set of energy values ​​104, and with at least one X-ray image 110 with a second set of energy values ​​106.

[0045] The processing unit 112 processes multiple phase-stepped X-ray images acquired at a first set of energy values ​​using phase retrieval to obtain a first transmission image, a DAX image 114, and a phase contrast (PC) image 116. Phase retrieval in DAX imaging is a process known to those skilled in the art and involves determining phase information from X-ray images acquired by phase stepping. In particular, the phase of an X-ray beam cannot be measured directly, and the phase shift must be converted into intensity modulation by interfering two or more waves. Phase retrieval may refer to the implementation of an algorithm that performs phase retrieval. The first transmission image 108 is a low-energy transmission image 204, and both the DAX image 114 and the PC image 116 are low-energy X-ray images.

[0046] Furthermore, the processing unit 112 processes at least one X-ray image 110 acquired at a second set of energy values ​​106 to obtain a second transmission image 416. Thus, the second transmission image 416 is a high-energy transmission image 206.

[0047] The processing unit 112 further processes the first transmission image 108 based on the second transmission image 416 to generate a modified transmission image 118. In other words, the modified transmission image 118 is a combination of the first transmission image 108 and the second transmission image 416. That is, the modified transmission image 118 is a combination of the low-energy transmission image 204 and the high-energy transmission image 206.

[0048] The output of the processing unit 112 is a DAX image 114, a PC image 116, and / or a modified transparency image 118, all of which are in an optimal state.

[0049] The processing unit 112 further processes the first transmission image 108 based on the second transmission image 410 to reduce noise in the corrected transmission image 118. The processing unit 112 determines the optimal signal-to-noise ratio in the transmission image based on the second transmission image 410, i.e., the high-energy transmission image 206, because the signal-to-noise ratio of the second transmission image 410 is high and clinically acceptable. Next, the processing unit 112 processes the first transmission image 108, i.e., the low-energy transmission image 204, which has a low signal-to-noise ratio and is clinically unacceptable, using contrast information from the second transmission image 410. Then, the processing unit 112 generates a corrected transmission image 118 that has the framing of the first transmission image 108 but has the high signal-to-noise ratio and clinical acceptability of the second transmission image 410.

[0050] In some implementations of the system in Figure 1, the processing unit 112 further processes the first transmission image 108 based on the second transmission image 410 to remove soft tissue or bone in the modified transmission image 418. The processing unit 112 determines contrast information from the first transmission image 108 and the second transmission image 410. The contrast information indicates which areas of the image are highly absorbent features (i.e., bone). The processing unit 112 determines the contrast information of the first transmission image 108 and the second transmission image 410 using an image segmentation method. The contrast information can be based on any X-ray image information passed to the processing unit 112. After areas of the image are identified as highly absorbent features, the processing unit 112 removes these features. The removal of these features is performed by applying one of the following effects to the identified areas: opacity mask, partial opacity mask, brightness reduction, and / or contrast reduction.

[0051] The processing unit 112 processes at least one of the phase-step X-ray images 108 and / or at least one X-ray image 110 according to a fringe removal algorithm to remove fringes. Fringes are artifacts that remain in the X-ray image as a result of using grating interferometry, and appear as bands or stripes perpendicular to the X-ray image. The fringe removal algorithm uses phase-step X-ray images in which no objects exist in order to obtain fringe information. This is not intended as a limitation. The processing unit can run the fringe removal algorithm using fringe information obtained by any method. Removing fringes improves the contrast, clarity, and overall quality of the X-ray image.

[0052] Referring to Figure 2, examples of transmission images 202 acquired at low energy 204 and high energy 206, and examples of DAX images 208 acquired at low energy 210 and high energy 212 are shown.

[0053] The transmission image 202 is generated from the X-ray signal transmitted through object 308. Therefore, the X-ray beam 306 is attenuated and / or absorbed by object 308. Energy is transferred from the X-ray beam 306 to object 308, and the X-ray signal is weaker because more energy is transferred to features with high absorption properties, which are usually high-density features such as bone. This is shown in the areas of the transmission image 202 that appear whiter, indicating that more absorption of the X-ray signal has occurred in these areas. As an example, the transmission image 202 is useful for inspecting signs of damage or deterioration in high-absorption features of object 308, such as bone. High-energy X-ray signals are more readily absorbed by high-absorption features compared to low-energy X-ray signals. Therefore, the high-energy transmission image 206 has a higher contrast, known as attenuation contrast, between the low-absorption and high-absorption features of object 308. Conversely, the low-energy transmission image 204 has a lower attenuation contrast between the low-absorption and high-absorption features.

[0054] DAX image 208 is generated from the X-ray signal diffracted through object 308, and therefore the phase of the X-ray beam 306 is changing. Since the diffraction of the X-ray beam 306 is dominated by small-angle scattering, the magnitude of diffraction through features of object 308 with different electron densities is different. Therefore, DAX image 208 is good for obtaining information from low-absorptural features of object 308, such as soft tissue. Diffraction information from high-energy 106 X-ray signals degrades more quickly through object 308 compared to low-energy 104 X-ray signals. Therefore, the low-energy DAX image 212 has higher contrast (compared to the high-energy DAX image) during slight changes in electron density in the soft tissue of object 308. Conversely, the high-energy DAX image 210 has lower contrast during slight changes in electron density in the soft tissue of object 308.

[0055] In human chest X-ray imaging, soft tissue is of greater interest than bone in diagnosing lung diseases. Therefore, DAX images 208 are useful to clinicians, but they need to be matched with radiographic images 202 to be clinically acceptable, for example. Radiographic images 202 can be acquired simultaneously with DAX images 208 on the DAX imaging system using phase retrieval 406. However, as mentioned above, there is a trade-off in the acquisition parameters of these two images; namely, radiographic images 202 are best acquired at high energy 106, while DAX images are best acquired at low energy 104.

[0056] However, according to the proposed concept, a modified transmission image is provided that appears to have been acquired at high energy (e.g., 125 kVp) but retains the characteristics of low-energy acquisition, thereby meeting clinical requirements regarding acquisition energy for, for example, chest X-ray imaging.

[0057] For further illustration, a model imaging unit 102 that may be used according to the proposed embodiment is described with reference to Figure 3. Figure 3 shows a simplified schematic diagram of the imaging unit 102 according to one embodiment. Specifically, in this embodiment, the imaging unit 102 includes grating interferometers 304, 310, 312 that collect multiple phase-step X-ray images 108 at a first set of energy values ​​104.

[0058] The imaging unit 102 includes an X-ray source 300, an X-ray beam spectral filter 302, a source grating 304, an X-ray beam 306, an object 308, a phase grating 310, an analysis grating 312, and / or a sensor 314.

[0059] The source grating 304, the phase grating 310, and / or the analysis grating 312 are configured according to a first set of energy values ​​104, a second set of energy values ​​106, and / or the configuration of the X-ray source 300. In other words, in some embodiments, the distance between any of the gratings 304, 310, and 312 is modified to match the energy level of the X-ray beam 306 passing through the gratings 304, 310, and 312 to maximize the phase information in the phase-stepped X-ray image 108, and / or the signal-to-noise ratio quality in the phase-stepped X-ray image 108 and / or at least one X-ray image 110.

[0060] The imaging unit includes grating interferometers 304, 310, and 312 that collect multiple phase-step X-ray images 108 at a first set of energy values ​​104. The grating interferometer setup includes a phase grating 310 and an analysis grating 312, and may additionally include a source grating 304. The grating consists of a repeating structure of lines of a first material and lines of a second material or space. Thus, the grating has an alternating line structure. The grating has important dimensions such as a period, which is the perpendicular distance between the end of a line of the first material and the end of the next line of the first material. The period of the grating is determined to fit different energy values ​​104, 106 of the X-ray beam 306. Another important dimension of the grating is the distance between the phase grating 310 and the analysis grating 312, which is also determined to fit the energy values ​​104, 106 of the X-ray beam 306.

[0061] Wave phase cannot be measured directly; rather, phase shift requires the interference of two or more waves to be converted into intensity modulation. A phase grid 310 is used to generate the corresponding interference pattern. The phase grid is placed between the object 308 under inspection and the sensor 314 (or between the light source grid 304 and the object 308). The interference pattern generated by the phase grid 310 may be too small to be detected by the sensor 314 due to the sensor's lack of spatial resolution; therefore, an analysis grid 312 is placed between the phase grid 310 and the sensor 314 to provide a sufficiently large interference pattern that is detectable by the sensor 314. Preferably, the introduction of the analysis grid 312 does not introduce new fringes, but the analysis grid is used solely to "analyze" the interference pattern of the phase grid 310.

[0062] Phase stepping is performed by moving the analysis grid 312 in steps perpendicular to the direction of the grid lines of the analysis grid 312, and so that the total distance is equal to the period of the analysis grid 312. Phase step X-ray images 108 are acquired by the sensor in multiple instances (i.e., different steps). These instances are at a fraction of the period of the analysis grid 312 (e.g., 1 / 4, 1 / 6, 1 / 8) (i.e., a distance equal to the period of the analysis grid 312). Multiple instances are required to acquire an image sequence in which phase information can be obtained. The analysis grid 312 can be moved by implementing a motor or a high-precision actuator that can achieve fractional accuracy of the period. Phase stepping requires the collection of 10 or more images to suppress vibration disturbances.

[0063] Alternatively, as an example, phase stepping is, (i) Moving the sensor 314 perpendicular to the direction of the line grid of the analysis grid 312, (ii) Moving the grid perpendicular to the direction of the line grid of the analysis grid 312, (iii) Move the X-ray source 300 perpendicular to the direction of the line grid of the analysis grid 312. (iv) Moving the X-ray beam 306 perpendicular to the direction of the line grid of the analysis grid 312 (i.e., using electromagnetic phase stepping), and (v) Rotating the phase grid 310 and the analysis grid 312 together around the axis at a constant angle along the orientation of the grid bars (for example, these two grids are maintained in a position aligned with each other or mechanically fixed), It is executed by [the specified method / system].

[0064] In order to obtain the phase stepping effect by other means, the function of the grating interferometry within the imaging unit is X-ray mask for sensor 314 (the X-ray mask has multiple openings, i.e., coded openings for edge illumination), A propagation sensor that can move parallel to the direction of the X-ray beam 306, or A rotating crystal that directs the X-ray beam after it has passed through object 308 on sensor 314, It may be replaced with this.

[0065] The imaging unit may include an X-ray beam spectral filter 302 and / or a dynamic interference grating 304. The X-ray beam spectral filter 302 and / or the dynamic interference grating 304 remove fringes from at least one of the phase-step X-ray images 108 and / or at least one X-ray image 110. The X-ray beam spectral filter 302 can prevent fringes from occurring in the first place by ensuring that the X-ray beam 306 is of sufficient quality to not produce fringes. The dynamic interference grating 304 moves the interference grating during image acquisition to remove (smear out) fringes from the resulting image. Removing fringes improves the contrast, clarity, and overall quality of the X-ray image.

[0066] Referring to Figure 4, a simplified block diagram of a multi-energy DAX and PC X-ray imaging method 400 according to one embodiment is shown. Specifically, in this embodiment, the processing unit 112 performs the steps of method 400.

[0067] In steps 402 and 404, method 400 receives a plurality of phase-stepped X-ray images acquired at a first set of energy values ​​104 and at least one X-ray image (with or without phase stepping) acquired at a second set of energy values ​​106. Each energy value in the first set is smaller than each energy value in the second set.

[0068] For example, the first set of energy values ​​104 consists of energy values ​​less than 100kVp, preferably in the range of 60kVp to 70kVp (i.e., low energy values). The first set of energy values ​​104 is determined automatically by the processing unit 112 for optimal acquisition of DAX images 114, or otherwise determined by the clinician, or determined by a default value.

[0069] The second set of energy values ​​106 consists of energy values ​​of 100 kVp or higher, preferably 120 kVp or higher, and more preferably at least one energy value that is substantially equal to 125 kVp (i.e., a high energy value). The second set of energy values ​​106 is determined automatically, by a clinician, or by a default value for optimal acquisition of the second transmission image 416.

[0070] In step 406, phase retrieval is performed on multiple phase-step X-ray images from step 402.

[0071] In step 408, the output of the phase retrieval process 406 is the acquired first transmission image, DAX image, and PC image. The DAX image and PC image are outputs of this method.

[0072] In step 410, at least one X-ray image received from step 404 is processed to obtain a second transmission image.

[0073] In step 412, the image combination application processes the first transmission image from step 408 and the second transmission image from step 410. The image combination application determines contrast information from the first transmission image from step 408 and the second transmission image from step 410. The contrast information includes the optimal signal-to-noise ratio of the transmission image and is determined based on the second transmission image 410. This is because the second transmission image 410 is acquired at a high energy value, resulting in a high signal-to-noise ratio that is clinically acceptable. The contrast information also indicates which areas of the image in both the first and second transmission images are features with high absorption rates (i.e., bone). This method uses an image segmentation method to determine the contrast information.

[0074] In step 414, the first transparent image from the image combination application 412 is processed based on contrast information to reduce and / or remove noise from the first transparent image, thereby generating a noise-reduced image.

[0075] In step 416, the first transparent image from the image combination application 412 or the noise-reduced image from step 414 is processed based on contrast information to remove bone or soft tissue from the first transparent image or noise-reduced image. The removal of bone or soft tissue is performed by applying one of the following effects to the determined area: opacity mask, partial opacity mask, brightness reduction, and / or contrast reduction. A feature-removed image is generated.

[0076] In step 418, a modified transparency image is generated based on the noise reduction image from step 414 or the feature removal image from step 416. The modified transparency image is the output of this method.

[0077] Method 400 may also use a computer program product including computer program code means. When the computer program code means is executed on a computing device having a processing system, it causes the processing system to execute all the steps of Method 400.

[0078] Referring to Figure 5, a simplified block diagram of the image processing algorithm 500 according to one embodiment is shown. Specifically, in this embodiment, the processing unit 112 executes the steps of the image processing algorithm 500.

[0079] The processing unit 112 then executes the image processing algorithm 500. The image processing algorithm uses the trained neural network 504 to receive an input 502 containing the first transmission image and generate an output 506 containing the corrected transmission image. Since the difference in the appearance of the images between 70kVp and 125kVp is not very large (in particular, only areas with high absorption rates in the chest, such as the rib cage, appear more clearly in the image), the neural network trained on these differences works very well to generate an image suitable for diagnostic purposes.

[0080] The structure of an artificial neural network (or simply a neural network) is inspired by the human brain. A neural network consists of layers, each containing multiple neurons. Each neuron performs a mathematical operation. In particular, each neuron may contain different weightings of a single type of transformation (for example, the same type of transformation, such as a sigmoid, but with different weightings). In the input data processing process, the mathematical operations of each neuron are performed on the input data to produce a numerical output, and the output of each layer of the neural network is passed sequentially to the next layer, with the final layer providing the output.

[0081] There are several types of neural networks, including convolutional neural networks (CNNs) and recurrent neural networks (RNNs). Embodiments of this invention use a CNN-based learning algorithm because CNNs have proven particularly well for image analysis, as they can classify images with a much lower error rate than other types of neural networks.

[0082] A CNN typically contains several layers, including convolutional layers, pooling layers, fully connected layers, and softmax layers. A convolutional layer consists of a set of learnable filters that extract features from the input. A pooling layer is a form of nonlinear downsampling that reduces data size by combining the outputs of multiple neurons in one layer with a single neuron in the next layer. A fully connected layer connects each neuron in one layer to all neurons in the next layer. A softmax layer determines the probability distribution of each output.

[0083] The training methods for machine learning algorithms are well known. Typically, such methods involve obtaining a training dataset containing training input data entries and corresponding training output data entries. An initialized machine learning algorithm is applied to each input data entry to generate predicted output data entries. The machine learning algorithm is then modified using the error between the predicted output data entries and the corresponding training output data entries. This process is repeated until the error converges, i.e., until the predicted output data entries are sufficiently similar to the training output data entries (e.g., ±1%). This is generally known as supervised learning.

[0084] For example, the weighting of the mathematical operations of each neuron is modified until the error converges. Known methods for modifying neural networks include gradient descent and backpropagation algorithms.

[0085] The training input data entries for the neural network used in Figure 5 correspond to transparent image example 408. The training output data entries correspond to the modified transparent image 418.

[0086] Several preprocessing methods may be used to improve training samples. Transmission images from the same imaging system can be normalized, for example, by display features.

[0087] Multiple neural networks can be generated by modifying one or more existing CNN-based learning algorithms such as VGG, Inception, and ResNet.

[0088] Modifications of the disclosed embodiments can be understood and implemented by those skilled in the art in carrying out the claimed invention, based on a review of the drawings, disclosures, and appended claims. In the claims, the word “including” does not exclude other elements or steps, and singular elements do not exclude plural elements.

[0089] A single processor or other unit can perform the functions of several items described in the claims.

[0090] The mere fact that certain means are described in mutually different dependent claims does not mean that combinations of these means cannot be used advantageously.

[0091] Computer programs may be stored and distributed on any suitable medium, such as optical storage media or solid-state media, supplied together with or as part of other hardware, but they may also be distributed in other forms, such as via the Internet or other wired or wireless communication systems.

[0092] Please note that when the term "adapted to..." is used in the claims or description, it is intended to be equivalent to the term "configured to...".

[0093] Any reference numeral in the claims should not be construed as limiting the scope.

Claims

1. A system for multi-energy dark-field and phase-contrast X-ray imaging, An imaging unit that collects multiple phase-step X-ray images at energy values ​​selected from a first set and collects at least one X-ray image at energy values ​​selected from a second set, wherein each energy value in the first set is smaller than each energy value in the second set, and includes an imaging unit and a processing unit. The aforementioned processing unit is The plurality of phase-step X-ray images acquired at energy values ​​selected from the first set are processed using phase retrieval to obtain a first transmission image, a dark-field X-ray image, and a phase-contrast image. A second transmission image is obtained by processing the at least one X-ray image acquired with an energy value selected from the second set, The first transparent image is processed based on the second transparent image to generate a corrected transparent image. The processing unit is a system that processes at least one of the plurality of phase-step X-ray images and / or the at least one X-ray image to remove fringes according to a fringe removal algorithm.

2. The system according to claim 1, wherein the imaging unit includes a grating interferometer, a plurality of (periodically coded) apertures, a propagation-based imaging arrangement, or a rotating crystal, which collects the plurality of phase-step X-ray images at energy values ​​selected from the first set.

3. The system according to claim 1 or 2, wherein the energy values ​​of the first set consist of energy values ​​less than 100 kVp.

4. The system according to claim 1 or 2, wherein the energy values ​​of the first set are composed of energy values ​​in the range of 60 kVp to 70 kVp.

5. The system according to any one of claims 1 to 4, wherein the energy values ​​of the second set consist of energy values ​​of 100 kVp or more.

6. A system for multi-energy dark-field and phase-contrast X-ray imaging, An imaging unit that collects multiple phase-step X-ray images at energy values ​​selected from a first set and collects at least one X-ray image at energy values ​​selected from a second set, wherein each energy value in the first set is smaller than each energy value in the second set, and includes an imaging unit and a processing unit. The aforementioned processing unit is The plurality of phase-step X-ray images acquired at energy values ​​selected from the first set are processed using phase retrieval to obtain a first transmission image, a dark-field X-ray image, and a phase-contrast image. A second transmission image is obtained by processing the at least one X-ray image acquired with an energy value selected from the second set, The first transparent image is processed based on the second transparent image to generate a corrected transparent image. The imaging unit further includes an X-ray beam spectral filter and / or a dynamic interference grating, wherein the X-ray beam spectral filter and / or the dynamic interference grating removes fringes from at least one of the plurality of phase-step X-ray images and / or from the at least one X-ray image.

7. A system for multi-energy dark-field and phase-contrast X-ray imaging, An imaging unit that collects multiple phase-step X-ray images at energy values ​​selected from a first set and collects at least one X-ray image at energy values ​​selected from a second set, wherein each energy value in the first set is smaller than each energy value in the second set, and includes an imaging unit and a processing unit. The aforementioned processing unit is The plurality of phase-step X-ray images acquired at energy values ​​selected from the first set are processed using phase retrieval to obtain a first transmission image, a dark-field X-ray image, and a phase-contrast image. A second transmission image is obtained by processing the at least one X-ray image acquired with an energy value selected from the second set, The first transparent image is processed based on the second transparent image to generate a corrected transparent image. The processing unit further processes the first transmission image based on the second transmission image to reduce noise in the modified transmission image, comprising a system.

8. A system for multi-energy dark-field and phase-contrast X-ray imaging, An imaging unit that collects multiple phase-step X-ray images at energy values ​​selected from a first set and collects at least one X-ray image at energy values ​​selected from a second set, wherein each energy value in the first set is smaller than each energy value in the second set, and includes an imaging unit and a processing unit. The aforementioned processing unit is The plurality of phase-step X-ray images acquired at energy values ​​selected from the first set are processed using phase retrieval to obtain a first transmission image, a dark-field X-ray image, and a phase-contrast image. A second transmission image is obtained by processing the at least one X-ray image acquired with an energy value selected from the second set, The first transparent image is processed based on the second transparent image to generate a corrected transparent image. The processing unit further processes the first transmission image based on the second transmission image to remove soft tissue or bone in the modified transmission image.

9. A system for multi-energy dark-field and phase-contrast X-ray imaging, An imaging unit that collects multiple phase-step X-ray images at energy values ​​selected from a first set and collects at least one X-ray image at energy values ​​selected from a second set, wherein each energy value in the first set is smaller than each energy value in the second set, and includes an imaging unit and a processing unit. The aforementioned processing unit is The plurality of phase-step X-ray images acquired at energy values ​​selected from the first set are processed using phase retrieval to obtain a first transmission image, a dark-field X-ray image, and a phase-contrast image. A second transmission image is obtained by processing the at least one X-ray image acquired with an energy value selected from the second set, The first transparent image is processed based on the second transparent image to generate a corrected transparent image. The processing unit further executes a training algorithm, which creates a trained neural network that receives a training input and a known output, wherein the training input includes the first transparent image and the known output includes the modified transparent image.

10. The system according to claim 9, wherein the processing unit further executes an image processing algorithm, the image processing algorithm using the trained neural network to receive an input including the first transparent image and to generate an output including the modified transparent image.

11. A method for multi-energy dark-field and phase-contrast X-ray imaging, A step of acquiring multiple phase-step X-ray images at energy values ​​selected from a first set and at least one X-ray image at energy values ​​selected from a second set, wherein each energy value in the first set is smaller than each energy value in the second set. The steps include performing a phase retrieval process on the plurality of phase-step X-ray images collected at energy values ​​selected from the first set in order to acquire a first transmission image, a DAX image, and a phase contrast image, The steps include processing the at least one X-ray image acquired with an energy value selected from the second set to obtain a second transmission image, The process includes the step of processing the first transparent image based on the second transparent image to generate a modified transparent image, A method comprising the step of processing the first transmission image, the step of reducing noise in the first transmission image based on the second transmission image.

12. A method for multi-energy dark-field and phase-contrast X-ray imaging, A step of acquiring multiple phase-step X-ray images at energy values ​​selected from a first set and at least one X-ray image at energy values ​​selected from a second set, wherein each energy value in the first set is smaller than each energy value in the second set. The steps include performing a phase retrieval process on the plurality of phase-step X-ray images collected at energy values ​​selected from the first set in order to acquire a first transmission image, a DAX image, and a phase contrast image, The steps include processing the at least one X-ray image acquired with an energy value selected from the second set to obtain a second transmission image, The process includes the step of processing the first transparent image based on the second transparent image to generate a modified transparent image, A method comprising the step of processing the first transmission image, the step of removing soft tissue or bone from the first transmission image based on the second transmission image.

13. The method according to claim 11 or 12, wherein the energy values ​​of the first set consist of energy values ​​less than 100 kVp.

14. The method according to any one of claims 11 to 13, wherein the energy values ​​of the first set are comprised of energy values ​​in the range of 60 kVp to 70 kVp.

15. A computer program comprising computer program code means, wherein the computer program code means, when executed on a computing device having a processing system, causes the processing system to perform all steps of the method according to any one of claims 11 to 14.

Citation Information

Patent Citations

  • Step imaging system and imaging method for X-ray source diffraction gratings

    JP2013524897A

  • X-ray phase difference imaging apparatus

    JP2018029777A

  • JPP4286012B

  • Correction in slit-scanning phase contrast imaging

    US20160128665A1