System and method for evaluating lung ventilation and perfusion gradient

Abstract

A system and method for quantifying pulmonary ventilation and perfusion gradients in a subject's lungs, wherein local Hounsfield density histograms are generated from computed tomography data and each local histogram is cross-correlated with a global Hounsfield density histogram at a number of different shift values, and a final shift value is determined for each local histogram based on the correlation values ​​obtained by the cross-correlation.

Description

Technical Field

[0001] The present invention relates to the field of evaluating computed tomography (CT) images.

Background Art

[0002] Gravity can create a ventilation and perfusion gradient between the ventral and dorsal regions of the lungs, particularly in subjects who have been lying down for a long time.

[0003] The potential adverse effects that can be caused by this gradient should be taken into account when treating patients with acute respiratory distress syndrome (ARDS) caused by infectious lung diseases such as COVID-19 by applying mechanical ventilation while the patient is in the prone position in addition to the standard supine position.

[0004] However, the presence and magnitude of the gradient due to gravity can vary greatly among patients. Some diseases increase or decrease the magnitude of this gradient. For example, patients with vascular disorders associated with systemic sclerosis exhibit a lower gravity-dependent attenuation effect due to a decrease in the elastic compliance of the pulmonary arterial tree. Some diffuse lung diseases (e.g., hypersensitivity pneumonitis, cellular nonspecific interstitial pneumonia, and subacute diffuse alveolar damage) show a subtle and relatively uniform increase in lung attenuation, which can overlap with the gradient induced by gravity.

[0005] The ventilation and perfusion gradient in the lungs of a patient can appear as a slight shift in the overall Hounsfield density between the ventral and dorsal lung regions in the three-dimensional image of a chest CT scan (when the subject is in the supine position, with the dorsal region having a high Hounsfield density). However, it is difficult to objectively quantify this effect.

[0006] Furthermore, the lung parenchyma surrounds normal anatomical structures (blood vessels, bronchi, lymph nodes, heart, etc.). In patients with lung disease, lesions, pleural effusion, atelectasis, mesothelioma, and scarring can be embedded within the lung parenchyma. This "anatomical noise" means that the Hounsfield density in lung CT images is far greater than the changes caused by lung ventilation and perfusion gradients. Therefore, simple quantitative measures such as local mean Hounsfield density cannot be used to provide information about ventral-dorsal gravity effects.

[0007] Treatment decisions for patients with acute respiratory syndrome often need to be made urgently in emergency situations, but in such circumstances, there is often insufficient time to perform careful manual measurements to determine information regarding lung ventilation and perfusion gradient.

[0008] Therefore, improved information regarding lung ventilation and perfusion gradients between the ventral and dorsal regions of the subject's lungs is needed.

[0009] U.S. Patent Application Publication 2021 / 065361 discloses a method and system for determining areas of dense lung parenchyma in lung images.

[0010] International Patent Application Publication No. WO2020 / 231904 discloses a method for imaging a patient's lungs.

[0011] The paper "Volumetric xenon-CT imaging of conventional and high-frequency oscillatory ventilation" by Mulreany, DG et al., Academic Radiology, 16(6):718~725, June 1, 2009, discloses a method for obtaining high-resolution volumetric ventilation maps of the lungs.

[0012] The paper "Regional lung perfusion analysis in experimental ARDS by electrical impedance and computed tomography" by Kircher, Michael et al., IEEE Transactions on Medical Imaging, 40(1):251~261, September 21, 2020, discloses a method for detecting lung diffusion capacity based on indicator-enhanced electrical impedance tomography.

[0013] The paper "Quantification of ventilation distribution in regional lung injury by electrical impedance tomography and xenon computed tomography" by Gunnar, Elke et al., Physiological Measurement, 34(10):1303~1318, September 11, 2013, describes the evaluation of local ventilation in pathological conditions based on electrical impedance tomography. [Overview of the project] [Problems that the invention aims to solve]

[0014] Improved information regarding lung ventilation and perfusion gradients between the ventral and dorsal regions of the subjects' lungs is needed. [Means for solving the problem]

[0015] The present invention is defined by the claims.

[0016] According to an example in accordance with one aspect of the present invention, a processing system is provided for evaluating the lung ventilation and perfusion gradient of at least one lung of a subject.

[0017] The processing system is configured to: receive CT imaging data containing representations of at least one lung of a subject; perform segmentation on the CT imaging data to identify representations of at least one lung of the subject; process the CT imaging data to generate multiple local histograms of the Houndsfield density frequency distribution, each of which represents a different region of at least one lung represented in the CT imaging data; generate a global histogram of the Houndsfield density frequency distribution for at least one lung of the subject; and for each of the multiple local histograms, generate multiple shifted local histograms, each of which is a version shifted by a different shift value of the local histogram; cross-correlate each shifted local histogram with the global histogram to generate a corresponding set of correlation values; and process the shift values ​​and correlation values ​​to generate a final shift value, the total range of the magnitude of the final shift value representing the magnitude of the lung ventilation and perfusion gradients.

[0018] The inventors recognized that lung ventilation and perfusion gradients can be quantified by shift values ​​in Hounsfield units from cross-correlated local histograms. Local histograms in regions with higher densities typically have a higher correlation with the global histogram for shifts in a first direction (i.e., positive shift values), while local histograms in regions with lower densities generally have a higher correlation with the global histogram for shifts in a second direction opposite to the first direction (i.e., negative shift values).

[0019] The overall range of the final shift value represents the magnitude of the lung ventilation and perfusion gradients, while the direction in which the range of the final shift value is greatest represents the direction of the gradient. The difference in magnitude between the final shift values ​​of adjacent regions represents the local gradient between those regions.

[0020] The CT imaging data can be two-dimensional or three-dimensional imaging data. The three-dimensional imaging data can provide improved accuracy and statistical detection power compared to the two-dimensional imaging data.

[0021] In some examples, the final shift value of each of the plurality of local histograms is the shift value corresponding to the correlation value having the maximum value among the plurality of correlation values of the local histogram.

[0022] The shift value corresponding to the maximum histogram cross-correlation (i.e., the shift value at which the correlation value is maximum) can be considered to best represent the difference between the local histogram and the global histogram.

[0023] In some examples, the final shift value of each of the plurality of local histograms is a correlation weighted average shift value.

[0024] Using the correlation weighted average shift value as the final shift value improves the robustness against noise and outliers.

[0025] In some examples, the processing system is configured to provide visualization information of the final shift value in a user interface.

[0026] Visualization of the shift value enables the clinician to immediately confirm the presence, amount, and distribution of the lung ventilation and perfusion gradients, and enables the clinician to take this into account during treatment.

[0027] The visualization also enables the clinician to detect the presence of attenuation differences due to disease rather than gravity. This is because local attenuation gradients that do not coincide with the direction of gravity are revealed by the visualization.

[0028] Negative final shift values can be represented by a first color, and positive final shift values can be represented by a second different color.

[0029] Using different colors for positive and negative shift values ​​immediately reveals the direction of the gradient, as it corresponds to the direction of the color change.

[0030] In some examples, the visualization of the final shift value is generated by: for each local histogram, generating visualization of the maximum correlation value of that local histogram; and generating an overlay representing the final shift value on the visualization of the maximum correlation value.

[0031] This recognizes that the correlation value with the highest cross-correlation also provides valuable clinical information. Typical parenchymal regions will have a higher maximum correlation value than affected regions, which will have a lower correlation value at any shift in the global histogram. Affected regions include lesions, exudates, atelectasis, mesothelioma, etc.

[0032] By overlaying the representation of the final shift value onto the visualization of the maximum correlation value, clinicians can obtain all of this information from a single image, enabling them to make rapid treatment decisions.

[0033] This visualization can use various effects to represent the final shift value and maximum correlation value. For example, grayscale visualization can be used to represent the maximum correlation value, with brighter shades representing higher maximum correlation values, and a color scale overlay can be used to represent the final shift value. The shift value can be within a predetermined range, for example, between -200 Houndsfield units and +200 Houndsfield units.

[0034] Multiple local histograms may include ray-wise histograms in the lateral and / or axial (i.e., head-to-tail) directions.

[0035] For subjects in the prone or supine position, the lateral and axial directions are perpendicular to the direction of gravity. Therefore, the effect of gravity is generally lowest in these directions. Thus, systematic cross-shifts between different histograms in either the lateral or axial direction can be used to quantify and visualize gravity-dependent gradients.

[0036] When the results are presented as a sagittal overview image, the lateral linear histogram can be used as the multiple local histograms for improved visual resolution and smoothness.

[0037] To evaluate lung ventilation and perfusion gradients in three dimensions, both horizontal and axial linear histograms can be used as the multiple local histograms. For example, three-dimensional visualization information of the final shift value can be generated by cross-correlating the linear histograms in both directions with the global histogram.

[0038] These multiple local histograms may include two-dimensional patch-like histograms in the coronal plane.

[0039] For subjects in the prone or supine position, the coronal plane is perpendicular to the direction of gravity. Patch histograms in the coronal plane can yield the most visually favorable results for the results presented in the coronal view.

[0040] In some examples, each of the multiple local histograms is generated by: generating multiple sublocal histograms, each representing a different subregion of at least one lung region corresponding to the local histogram; and accumulating these sublocal histograms into the local histogram.

[0041] If a local histogram contains multiple sublocal histograms, these sublocal histograms have a finer resolution than the local histogram. For example, sublocal histograms may be linear histograms (e.g., transverse and / or axial) or two-dimensional patch histograms (e.g., coronal plane).

[0042] First, histograms are generated at a resolution finer than necessary (for example, the finest possible resolution), and then these histograms are accumulated into a histogram with the desired resolution. This allows for the efficient generation of local histograms at various resolutions according to user requirements.

[0043] For example, if a user determines that the resolution of the visualization generated based on the cross-correlation of local histograms is not sufficiently useful, the user can provide user input requesting visualizations at a different resolution. In other examples, the user may find multiple visualizations at different resolutions useful. These can be efficiently generated by accumulating partial local histograms into a new local histogram according to the required resolution.

[0044] In some cases, a global histogram is generated by accumulating local histograms into a global histogram.

[0045] A system for evaluating at least one lung of a subject is also proposed. The system comprises: a CT scanner configured to generate CT imaging data; and the aforementioned processing system configured to receive CT imaging data from the CT scanner.

[0046] The above processing system can be configured to generate visualization information of the final shift value, and the system further includes a user interface configured to receive and display the generated visualization information of the final shift value from the above processing system.

[0047] Another aspect of the present invention provides a computer-aided method for evaluating the lung ventilation and perfusion gradient of at least one lung of a subject. The computer-aided method includes: receiving CT imaging data including representations of at least one lung of a subject; performing segmentation on the CT imaging data to identify representations of at least one lung of the subject; processing the CT imaging data to generate a plurality of local histograms of Hounsfield density frequency distributions, each of which local histograms represents a different region of at least one lung represented in the CT imaging data; generating a global histogram of the Hounsfield density frequency distribution of at least one lung of the subject; and for each of the plurality of local histograms, generating a plurality of shifted local histograms, each of which is a version of the local histogram shifted by a different shift value; cross-correlation of each shifted local histogram with the global histogram to generate a plurality of corresponding correlation values; and processing the shift values ​​and correlation values ​​to generate a final shift value, wherein the entire range of the magnitude of the final shift value represents the magnitude of the lung ventilation and perfusion gradient.

[0048] A computer program product has also been proposed, which, when executed on a computer device having a processing system, includes computer program code means that causes the processing system to execute all of the steps of the method described above.

[0049] The above and other aspects of the present invention will become apparent from the embodiments described below and will be explained with reference to such embodiments.

[0050] The accompanying drawings are provided for illustrative purposes only, in order to better understand the present invention and to more clearly demonstrate how it can be implemented. [Brief explanation of the drawing]

[0051] [Figure 1]Figure 1 shows a system for evaluating at least one lung of a subject according to one embodiment of the present invention. [Figure 2] Figure 2 shows an example image generated from CT scan data of the lungs of a COVID-19 patient. [Figure 3] Figure 3 shows a computer-assisted method for evaluating at least one lung of a subject according to one embodiment of the present invention. [Modes for carrying out the invention]

[0052] The present invention will be described below with reference to the drawings.

[0053] The detailed description and specific examples provided 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 invention will be better understood from the following description, appended claims, and accompanying drawings. It should be understood that the same reference numerals are used throughout the drawings to indicate identical or similar parts.

[0054] According to the concept of the present invention, a system and method are proposed for quantifying the lung ventilation and perfusion gradient of a subject's lungs. Local Hounsfield density histograms are generated from computed tomography data, and each local histogram is cross-correlated with the global Hounsfield density histogram at several different shift values. The final shift value is determined for each local histogram based on the correlation value obtained by the cross-correlation.

[0055] Each embodiment is at least partially based on the understanding that lung ventilation and perfusion gradients can be represented by Hounsfield unit shift values, in which local histograms correlate most closely with global histograms.

[0056] Exemplary embodiments may be used, for example, in computed tomography systems, particularly PACS and workstation reading modules (e.g., Philips Intellispace-Portal, Pacs, and Illumeno), as well as in intensive care and mechanical ventilation.

[0057] Figure 1 shows a system 100 for evaluating at least one lung of a subject 110 according to one embodiment of the present invention. The system comprises a computed tomography (CT) scanner 120, a processing system 130, and optionally a user interface 140. The processing system 130 itself is one embodiment of the present invention.

[0058] The CT scanner 120 can be any suitable CT scanner. Conventional CT scanners include an X-ray emission generator mounted opposite one or more integrated detectors on a rotatable gantry or C-arm. The X-ray generator rotates completely or partially around an examination area located between the X-ray generator and the one or more detectors, emitting (usually polychromatic) radiation across the examination area and the subject and / or objects placed within it. The one or more detectors detect the radiation across the examination area and generate signals (or projection data) indicating the examination area and the subject and / or objects placed within it. The projection data represents the raw detector data and can be used to form a projection sinogram, the latter being a visual representation of the projection data captured by the detectors.

[0059] A reconstructor is further used to process the above projection data and reconstruct a volume image of the subject or object. The volume image consists of multiple cross-sectional image slices, each of which is generated from the projection data through a tomographic reconstruction process such as the application of a filtered back projection algorithm. The reconstructed image data is, in effect, the inverse Radon transform of the raw projection data.

[0060] The CT scanner 120 acquires / generates CT imaging data 125 that includes a representation of at least one lung of the subject 110. The CT imaging data may be any imaging data that can evaluate gravity-dependent lung ventilation and perfusion gradients in at least one lung. In some examples, the CT imaging data may be two-dimensional imaging data that includes a component in the direction of gravity. When the subject is in a supine or prone position, the direction of gravity is parallel or approximately parallel to the anterior-posterior (y-) direction. Therefore, the CT imaging data may include two-dimensional lateral or axial slices through the subject's chest. In other examples, the CT imaging data may be three-dimensional (e.g., a whole chest scan).

[0061] The processing system 130 receives CT imaging data 125 from the CT scanner 120 and performs segmentation on the CT imaging data to identify representations of at least one lung of the subject 110. Any suitable segmentation method can be used to identify at least one lung representation, such as a voxel-based segmentation method, a mesh model-based segmentation method, and an AI-based segmentation method (e.g., using one or more convolutional neural networks or other machine learning methods). The segmentation may include the separation / identification of the subject's left lung, right lung, or both lungs (or their boundaries).

[0062] The processing system 130 processes the CT imaging data 125 to generate multiple local histograms of the Hounsfield density frequency distribution. A histogram is a data structure that identifies the frequency of each data value within a group (for a group of data values). In this context, a group of data values ​​is a group of Hounsfield unit values ​​(e.g., of different pixels).

[0063] Each of the multiple local histograms represents a different region of at least one lung represented in the CT imaging data, identified based on the segmentation results. The multiple local histograms may correspond to one or more image planes perpendicular to the direction of gravity to reduce the effect of gravity on changes in Hounsfield density within each local histogram.

[0064] In the case of three-dimensional imaging data, multiple local histograms may, for example, have linear histograms in the lateral (x-) direction. In other words, a Hounsfield density frequency distribution may be established for multiple yz points in the lateral view. Alternatively or additionally, multiple local histograms may have linear histograms in the axial (z-) direction (i.e., a Hounsfield density frequency distribution may be established for multiple xy points in the axial view).

[0065] In other examples, multiple local histograms may have two-dimensional patch-like histograms in the coronal (xz-) plane.

[0066] In some cases, multiple local histograms may each have multiple sublocal histograms. In other words, for each local histogram, multiple sublocal histograms may be generated, each representing a sub-region of at least one lung region represented by that local histogram. In this case, the sublocal histograms corresponding to a particular local histogram may be accumulated to generate a local histogram.

[0067] A partial local histogram can have the finest possible resolution. This allows for the generation of a local histogram of any desired resolution by accumulating these partial local histograms.

[0068] For example, a partial local histogram may include a linear histogram (e.g., in the horizontal (x-) and / or axial (z-) directions). In other examples, a partial local histogram may include a two-dimensional patch histogram (e.g., within the coronal plane).

[0069] The number of sublocal histograms accumulated in a single local histogram can be any number from 1 to N, where N is the total number of sublocal histograms across at least one lung. In some examples, the number of sublocal histograms accumulated in a single local histogram can be a coefficient of N, so that each local histogram may have the same number of sublocal histograms.

[0070] The number of sublocal histograms accumulated in a single local histogram can be predetermined or determined based on a desired resolution, which can be obtained by user input.

[0071] Hounsfield density histograms are commonly used when processing CT imaging data, and methods for calculating multiple local or sublocal histograms will be apparent to those skilled in the art. These calculations can be performed using a massively parallel processing approach.

[0072] Histograms are accumulated by combining them. For example, histograms can be accumulated using summation operations (e.g., sum or weighted sum of histograms). Other methods of accumulating histograms will be apparent to those skilled in the art, such as using averaging, multiplication, or weighted multiplication. In some examples, when the accumulation involves a weighted sum, Gaussian weighting can be applied when accumulating sublocal histograms into local histograms (e.g., using standard three-dimensional Gaussian filtering when the x-axis of the image volume is the axis of the histogram bins). In other examples, other types of filtering or weighting (e.g., mean filtering, median filtering, etc.) can be applied when accumulating sublocal histograms.

[0073] The processing system 130 then generates a global histogram of the overall Hounsfield density frequency distribution for at least one lung of the subject 110. The global histogram may be generated by accumulating local histograms. The global histogram may be generated for a single lung or both lungs, or separate global histograms may be generated for each lung.

[0074] In some cases, a global histogram can be generated by first accumulating local histograms into a regional histogram, and then accumulating these regional histograms into a global histogram. A regional histogram can correspond to multiple neighbor histograms extending perpendicular to the direction of a linear histogram. In other words, a linear histogram corresponding to a specific location within a neighbor histogram can be accumulated into a regional histogram. For example, a linear histogram in the horizontal (x-) direction can be accumulated into a regional histogram corresponding to a yz-neighbor histogram, and a linear histogram in the axial (i.e., head-to-tail) (z-) direction can be accumulated into a regional histogram corresponding to an xy-neighbor histogram.

[0075] For each of the multiple local histograms, the processing system 130 generates multiple shifted local histograms and cross-correlated each shifted local histogram with the global histogram. Each of the multiple shifted local histograms is a version of the local histogram shifted by a different shift value (along the Hounsfield density axis). The shift value to which the local histogram is shifted can fall within a predetermined range. For example, the shift value may be in the range between -200 Hounsfield units and +200 Hounsfield units.

[0076] Cross-correlation is a well-known technique in the field of image processing for determining the similarity between two histograms, providing a highly sensitive and robust estimate of goodness of fit. See, for example, Wu and Hudson, "An image-clustering method based on cross-correlation of color histograms," Proc. SPIE 5682, Storage and Retrieval Methods and Applications for Multimedia 2005, and Guthier et al., "Parallel implementation of a real-time high dynamic range video system," Integrated Computer-Aided Engineering, 21(2):189-202.

[0077] The cross-correlation between each of the multiple shifted local histograms and the global histogram yields a number of corresponding correlation values. The processing system 130 then processes the shift values ​​and correlation values ​​to generate a final shift value for each of the multiple local histograms. The final shift value of the local histogram can be, for example, the shift value corresponding to the correlation value having the maximum value among the multiple correlation values ​​of the local histogram.

[0078] As another example, the correlation-weighted average shift value can be used as the final shift value to reduce the effects of noise and outliers. The correlation value is expected to increase approximately smoothly as it approaches the “true” shift value corresponding to the largest correlation. Therefore, the “true” peak shift value can be estimated as the correlation-weighted average shift, even if the peak position is affected by noise. The correlation-weighted average shift is:

number

[0079] In some cases, the processing system 130 may be configured to further generate visualization information of final shift values ​​for multiple local histograms and output the generated visualization information to the user interface 140. Lung ventilation and perfusion gradients become more apparent in such visualization information compared to visualization information of Hounsfield density, enabling clinicians to identify the presence and magnitude of gravity-dependent gradients and to decide on treatment options for subjects accordingly. Visualization of final shift values ​​also allows clinicians to identify regions where local gradients do not coincide with the direction of gravity, which can assist in the diagnosis of certain lung diseases.

[0080] In this visualization, any appropriate scale can be used to represent the final shift value. In some examples, a negative final shift value can be represented by the first color, and a positive final shift value can be represented by a second different color. The magnitude of the final shift value can be represented by the intensity of the first or second color.

[0081] In some cases, visualizations of the final shift value can be overlaid on visualizations of the maximum correlation value (or correlation-weighted average shift value). For example, the processing system 130 can generate grayscale visualizations of the maximum correlation value for multiple local histograms and generate a color overlay representing the final shift value on top of the visualizations of the maximum correlation value.

[0082] Local histograms corresponding to typical parenchymal tissue areas are highly correlated with global histograms at shift values ​​corresponding to the maximum correlation values, while disease areas generally have very low maximum correlation values.

[0083] By combining visualizations of final shift values ​​with visualizations of maximum correlation values, it is possible to generate "key images" that show the local presence, volume, and distribution of lung ventilation and perfusion (diffusion) gradients together with diseased parenchymal tissue areas.

[0084] In some cases, if the CT imaging data 125 is three-dimensional imaging data, three-dimensional visualization information representing lung ventilation and perfusion gradients can be generated. For example, the processing system 130 can generate a first set of linear histograms in the lateral (x-) direction and a second set of linear histograms in the axial (z-) direction, and then cross-correlate each of these histograms with the global histogram as described above to generate three-dimensional visualization information of the final shift value.

[0085] Three-dimensional visualizations can reveal local variations in ventral-dorsal gravitational effects, particularly those in the coronal (xz-) plane, which are not apparent from two-dimensional visualizations.

[0086] In another example, two-dimensional visualization information for the coronal view can be generated in addition to two-dimensional visualization information for the lateral and / or axial views, allowing for the observation of changes in the axial (z-) and lateral (x-) directions. This visualization information can be generated, as mentioned above, for example, by using a linear histogram in the anterior-posterior (y-) direction as a local histogram.

[0087] Figure 2 shows an exemplary image generated from CT scan data of the lungs of subjects infected with COVID-19.

[0088] Image 210 shows the standard Hounsfield values ​​for a single slice in sagittal view.

[0089] Figure 220 shows the visualization of the maximum correlation value generated by cross-correlation of histograms as described above. In Figure 220, the local histogram used for cross-correlation with the global histogram was generated by accumulating a sublocal histogram having a transverse linear histogram with a Gaussian weighted neighborhood radius of 9 pixels in the sagittal plane.

[0090] Image 230 shows a visualization of the shift values ​​that reached maximum cross-correlation, with negative shifts represented by darker shading and positive shifts by brighter shading. Lung ventilation and perfusion gradients are clearly visible in this visualization.

[0091] Image 240 is a composite image of Images 220 and 230, showing both lung ventilation and perfusion gradients, as well as the affected area of ​​the lung (in this case, the dorsal COVID-19 lesion appearing as a missing value in the visualization information of the maximum correlation value). Although Image 240 is a grayscale image, lung ventilation and perfusion gradients can be more easily distinguished from disease-related changes by using color in one or both of the visualizations.

[0092] Figure 3 shows a computer-assisted method 300 for evaluating at least one lung of a subject, according to one embodiment of the present invention.

[0093] The method begins in step 310, in which CT imaging data including representations of at least one lung of the subject is acquired.

[0094] In step 320, segmentation is performed on the CT imaging data to identify representations of at least one lung of the subject.

[0095] In step 330, the CT imaging data is processed to generate multiple local histograms of the Hounsfield density frequency distribution. Each of these local histograms represents a different region of at least one lung as shown in the CT imaging data.

[0096] In step 340, a global histogram of the Hounsfield density frequency distribution for at least one lung of the subject is generated.

[0097] In step 350, multiple shifted local histograms are generated for each of the multiple local histograms. Each shifted local histogram is a version of the local histogram shifted by a different shift value.

[0098] In step 360, for each of the multiple local histograms, each shifted local histogram is cross-correlated with the global histogram to generate a corresponding number of correlation values.

[0099] In step 370, the shift value and correlation value are processed for each of the multiple local histograms to generate the final shift value.

[0100] It will be understood that the disclosed methods are methods to be executed on a computer. Therefore, the concept of a computer program having coding means for carrying out any of the described methods, when executed on a processing system, is also proposed.

[0101] As mentioned above, the system uses a processor to perform data processing. The processor can be implemented in various ways using software and / or hardware to perform the various necessary functions. Typically, a processor employs one or more microprocessors that can be programmed using software (e.g., microcode) to perform the required functions. A processor can be implemented as a combination of dedicated hardware for performing some functions and one or more programmed microprocessors and associated circuits for performing other functions.

[0102] Examples of circuits that may be used in various embodiments of this disclosure include, but are not limited to, conventional microprocessors, application-specific integrated circuits (ASICs), and field-programmable gate arrays (FPGAs).

[0103] In various implementations, the processor may be associated with one or more storage media, such as volatile and non-volatile computer memory like RAM, PROM, EPROM, and EEPROM®. These storage media may be encoded with one or more programs that perform the necessary functions when executed on one or more processors and / or controllers. The various storage media may be fixed within the processor or controller, or they may be portable so that the one or more programs stored within them can be loaded into the processor.

[0104] Variations of the disclosed embodiments can be understood and implemented by those skilled in the art by examining the drawings, this disclosure, and the appended claims when carrying out the invention described in the claims. In the claims, the word “having (including)” does not exclude other elements or steps, and the singular form does not exclude the plural. A single processor or other unit can perform the functions of several items described in the claims. The mere fact that certain means are described in different dependent claims does not mean that combinations of these means cannot be used advantageously. Computer programs can be stored / distributed not only on suitable media such as optical storage media or solid-state media supplied together with or as part of other hardware, but also in other forms, such as via the Internet or other wired or wireless communication systems. Note that where the term “adapted” is used in the claims or description, such “adapted” is intended to be equivalent to the term “configured.” No reference numeral in the claims should be construed as limiting the scope. The embodiments of the present invention are described below. (Note 1) A processing system for evaluating lung ventilation and perfusion gradient in at least one lung of a subject, CT imaging data including representation of at least one lung of the subject is received, Segmentation is performed on the CT imaging data to identify representations of at least one lung of the subject. By processing the CT imaging data, a plurality of local histograms of the Hounsfield density frequency distribution are generated, where each of the plurality of local histograms represents a different region of at least one lung as shown in the CT imaging data. A global histogram of the Hounsfield density frequency distribution for at least one lung of the subject is generated. For each of the aforementioned local histograms, Multiple shifted local histograms are generated, each being a version of the local histogram shifted by a different shift value. Each shifted local histogram is cross-correlated with the global histogram to generate a number of corresponding correlation values. The shift value and the correlation value are processed to generate the final shift value. A processing system in which the entire range of the magnitude of the final shift value represents the magnitude of the lung ventilation and perfusion gradients. (Note 2) The processing system as described in Appendix 1, wherein the final shift value of each of the multiple local histograms is the shift value corresponding to the correlation value having the maximum value among the multiple correlation values ​​of the local histogram. (Note 3) The processing system as described in Appendix 1, wherein the final shift value of each of the aforementioned multiple local histograms is the correlation-weighted average shift value. (Note 4) A processing system according to any one of the appendices 1 to 3, which provides visualization information of the final shift value in a user interface. (Note 5) The processing system described in Appendix 4, in which a negative final shift value is represented by a first color, and a positive final shift value is represented by a second different color. (Note 6) The visualization information of the final shift value is, For each local histogram, visualization information of the maximum correlation value of that local histogram is generated. An overlay representing the final shift value is generated on the visualization information of the maximum correlation value. The processing system described in Appendix 4 or 5, which is generated by the above. (Note 7) The processing system according to any one of the appendices 1 to 6, wherein the shift value is within a predetermined range, for example, between -200 Houndsfield units and +200 Houndsfield units. (Note 8) The processing system according to any one of the appendices 1 to 7, wherein the plurality of local histograms include linear histograms in the lateral and / or axial directions, i.e., in the head-to-tail direction. (Note 9) The processing system according to any one of the appendices 1 to 7, wherein the plurality of local histograms include a two-dimensional patch-like histogram in the coronal plane. (Note 10) Each of the aforementioned local histograms is Multiple sublocal histograms are generated, each representing a different subregion of at least one lung region corresponding to the local histogram in question. The aforementioned partial local histogram is accumulated into the local histogram. A processing system described in any one of the appendices 1 to 7, which is generated by the above. (Note 11) The processing system according to any one of Appendix 1 to 10, wherein the global histogram is generated by accumulating the local histograms into the global histogram. (Note 12) A system for evaluating at least one lung of a subject, A CT scanner that generates CT imaging data, A processing system according to any one of appendices 1 to 11 that receives the CT imaging data from the CT scanner. A system that has (Note 13) The system according to Appendix 12, wherein the processing system generates visualization information of the final shift value, and the system further has a user interface that receives and displays the generated visualization information of the final shift value from the processing system. (Note 14) A computer-aided method for evaluating lung ventilation and perfusion gradient in at least one lung of a subject, The steps include receiving CT imaging data that includes a representation of at least one lung of the subject, The steps include performing segmentation on the CT imaging data to identify representations of at least one lung of the subject, A step of generating a plurality of local histograms of Hounsfield density frequency distribution by processing the CT imaging data, wherein each of the plurality of local histograms represents a different region of at least one lung represented in the CT imaging data; The steps include generating a global histogram of the Hounsfield density frequency distribution of at least one lung of the subject, For each of the aforementioned local histograms, Multiple shifted local histograms are generated, each being a version of the local histogram shifted by a different shift value. Each shifted local histogram is cross-correlated with the global histogram to generate a number of corresponding correlation values. The shift value and the correlation value are processed to generate the final shift value. Steps and It has, A computer-aided method wherein the entire range of the magnitude of the final shift value represents the magnitude of the lung ventilation and perfusion gradients. (Note 15) A computer program having computer program code means that, when executed on a computer device having a processing system, causes the processing system to execute all the steps of the computer implementation method described in Appendix 14.

Claims

1. A processing system for evaluating the lung ventilation and perfusion gradient of at least one lung of a subject, CT imaging data including representation of at least one lung of the subject is received. Segmentation is performed on the CT imaging data to identify representations of at least one lung of the subject. By processing the CT imaging data, a plurality of local histograms of the Hounsfield density frequency distribution are generated, where each of the plurality of local histograms represents a different region of at least one lung of the subject as shown in the CT imaging data. By processing the CT imaging data, a global histogram is generated showing the Hounsfield density frequency distribution of at least one lung of the subject, where the global histogram represents the entire at least one lung of the subject as shown in the CT imaging data. For each of the aforementioned local histograms, Multiple shifted local histograms are generated, each being a version of the local histogram shifted by a different shift value. Each shifted local histogram is cross-correlated with the global histogram to generate a number of corresponding correlation values. The shift value and the correlation value are processed to generate the final shift value. The aforementioned final shift value represents the difference between the local histogram and the global histogram. A processing system in which the entire range of the magnitude of the final shift value represents the magnitude of the lung ventilation and perfusion gradients.

2. The processing system according to claim 1, wherein the final shift value of each of the plurality of local histograms is the shift value corresponding to the correlation value having the maximum value among the plurality of correlation values ​​of the local histogram.

3. The processing system according to claim 1, wherein the final shift value of each of the plurality of local histograms is the correlation-weighted average shift value.

4. A processing system according to any one of claims 1 to 3, which provides visualization information of the final shift value in a user interface.

5. The processing system according to claim 4, wherein a negative final shift value is represented by a first color, and a positive final shift value is represented by a second different color.

6. The visualization information of the final shift value is, For each local histogram, visualization information of the maximum correlation value of that local histogram is generated. An overlay representing the final shift value is generated on the visualization information of the maximum correlation value. The processing system according to claim 5, which is generated by the above.

7. The processing system according to any one of claims 1 to 3, wherein the shift value is within a predetermined range.

8. The processing system according to any one of claims 1 to 3, wherein the plurality of local histograms include linear histograms in the lateral and / or axial directions, i.e., in the head-to-tail direction.

9. The processing system according to any one of claims 1 to 3, wherein the plurality of local histograms include a two-dimensional patch-like histogram in the coronal plane.

10. Each of the aforementioned local histograms is Multiple sublocal histograms are generated, each representing a different subregion of at least one lung region corresponding to the local histogram in question. The aforementioned partial local histogram is accumulated into the local histogram. A processing system according to any one of claims 1 to 3, which is generated by the above.

11. The processing system according to any one of claims 1 to 3, wherein the global histogram is generated by accumulating the local histograms into the global histogram.

12. A system for evaluating at least one lung of a subject, A CT scanner that generates CT imaging data, A processing system according to any one of claims 1 to 3, which receives the CT imaging data from the CT scanner. A system that has

13. The system according to claim 12, wherein the processing system generates visualization information of the final shift value, and the system further has a user interface that receives and displays the generated visualization information of the final shift value from the processing system.

14. A computer-aided method for evaluating lung ventilation and perfusion gradient in at least one lung of a subject, The steps include receiving CT imaging data that includes a representation of at least one lung of the subject, The steps include performing segmentation on the CT imaging data to identify representations of at least one lung of the subject, A step of generating a plurality of local histograms of Hounsfield density frequency distribution by processing the CT imaging data, wherein each of the plurality of local histograms represents a different region of at least one lung of the subject as shown in the CT imaging data; A step of generating a global histogram showing the Hounsfield density frequency distribution of at least one lung of the subject by processing the CT imaging data, wherein the global histogram represents the entire at least one lung of the subject as shown in the CT imaging data, For each of the aforementioned local histograms, Multiple shifted local histograms are generated, each being a version of the local histogram shifted by a different shift value. Each shifted local histogram is cross-correlated with the global histogram to generate a number of corresponding correlation values. The shift value and the correlation value are processed to generate the final shift value. Steps and It has, The aforementioned final shift value represents the difference between the local histogram and the global histogram. A computer-aided method wherein the entire range of the magnitude of the final shift value represents the magnitude of the lung ventilation and perfusion gradients.

15. A computer program having computer program code means that, when executed on a computer device having a processing system, causes the processing system to execute all of the steps of the computer implementation method described in claim 14.

Images