Automatic regional lung disease quantification from chest x-ray images

Abstract

Systems, devices, and methods provide techniques for automatically evaluating a diagnostic image, the techniques comprising: receiving a diagnostic image related to a condition of a patient; performing registration of the diagnostic image with reference to the anatomical structure; identifying one or more ROIs from the registered image; generating a feature distribution based on the one or more ROIs; analyzing the feature distribution to determine a quantification, the quantification reflecting the condition of the patient; and providing a diagnostic output based on the quantization. In an embodiment, identifying the ROI includes: identifying a domain of interest in the registered image, the domain of interest including the one or more regions of interest; and dividing the domain of interest into a plurality of sub-regions. In an embodiment, generating the feature distribution includes generating an intensity histogram for each ROI. In an embodiment, analyzing the feature distribution includes determining one or more metrics based on the feature distribution.

Description

Technical Field

[0001] Embodiments relate generally to computing techniques. More specifically, embodiments relate to automatic disease quantification from chest X-ray images. Background Art

[0002] X-ray images are used to detect and stage a variety of different lung diseases. As an example, after the initial PCR diagnosis or throughout the hospitalization period, hospitalized patients are diagnosed and staged for COVID-19 and other diseases using, for example, portable X-ray imaging systems. In fact, patients admitted to the intensive care unit (ICU) receive portable X-ray images regularly (e.g., once a day, once every other day, once every three days, etc.) to determine their stage of illness or check the status of their installed equipment at more or less regular intervals. Disease evaluation using such images is mainly based on visual inspection and verbal description by radiologists. Summary of the invention

[0003] Therefore, there may be a need for improving medical diagnostic imaging in terms of providing a method for automatically quantifying a disease (e.g., a lung disease) from a chest X-ray image. The objects of the disclosed technology are solved by the subject matter of the attached independent claims, wherein further embodiments are contained in the dependent claims, the drawings and the following description.

[0004] Improved computing systems, methods, and computer-readable media for automatically evaluating diagnostic images are disclosed herein. According to one or more embodiments, a computer-implemented method includes: receiving a diagnostic image related to a patient's condition; performing registration of the diagnostic image with reference to an anatomical structure to form a registered image; identifying one or more regions of interest from the registered image; generating a feature distribution based on the one or more regions of interest; analyzing the feature distribution to determine a quantification, the quantification reflecting the patient's condition; and providing a diagnostic output based on the quantification.

[0005] According to one or more embodiments, a computer-implemented system includes a processor and a memory coupled to the processor, the memory including instructions that, when executed by the processor, cause the computing system to perform operations including: receiving a diagnostic image related to a patient's condition; performing registration of the diagnostic image with reference to an anatomical structure to form a registered image; identifying one or more regions of interest from the registered image; generating a feature distribution based on the one or more regions of interest; analyzing the feature distribution to determine a quantification that reflects the condition of the patient; and providing a diagnostic output based on the quantification.

[0006] According to one or more embodiments, at least one non-transitory computer-readable storage medium includes instructions that, when executed by a computing system, cause the computing system to perform operations including: receiving a diagnostic image related to a patient's condition; performing registration of the diagnostic image with reference to an anatomical structure to form a registered image; identifying one or more regions of interest from the registered image; generating a feature distribution based on the one or more regions of interest; analyzing the feature distribution to determine a quantification that reflects the condition of the patient; and providing a diagnostic output based on the quantification.

[0007] The features, functions, and advantages of the disclosed technology can be achieved independently in various examples or can be combined in yet other examples, further details of which can be seen with reference to the following description and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Various advantages of the embodiments will become apparent to those skilled in the art upon reading the following description and appended claims and by referring to the following drawings, in which:

[0009] Figure 1 A block diagram illustrating an overview of an example of an automatic diagnostic system is provided according to one or more embodiments;

[0010] Figure 2 A block diagram illustrating an example of an automatic diagnostic system according to one or more embodiments is provided;

[0011] Figure 3 An exemplary chest X-ray image illustrating application of the contour detection step according to one or more embodiments is provided;

[0012] Figure 4 An example chest X-ray image illustrating identification and isolation of a region of interest according to one or more embodiments is provided;

[0013] Figures 5A-5D Examples of identifying a region of interest and dividing the region of interest into sub-regions according to one or more embodiments are provided;

[0014] Figure 6 An example of generating a feature distribution based on an identified region of interest according to one or more embodiments is provided;

[0015] Figure 7 An example visual display of a regional diagnostic output is provided in accordance with one or more embodiments;

[0016] Figure 8 An example visual display of a diagnostic output timeline is provided according to one or more embodiments;

[0017] Figures 9A-9D A flowchart illustrating an example method related to performing automatic diagnostic image evaluation according to one or more embodiments is provided; and

[0018] Fig.10 is a diagram illustrating an example of a computing system for use in an automated diagnostic system in accordance with one or more embodiments. DETAILED DESCRIPTION

[0019] Improved computing systems, methods, and computer-readable media for automatically evaluating diagnostic images are disclosed herein. As described herein, the technology is operable to identify one or more regions of interest, generate feature distributions for the regions of interest, analyze the feature distributions to quantify the patient's condition, and provide a diagnostic output. The disclosed technology helps improve the overall performance of the diagnostic system by providing an automatic regional quantification algorithm and a user interface for the patient, and the described technology can be used for lung diseases / conditions (e.g., COVID-19, pneumonia, pleural effusion, pneumothorax, etc.). The technology enables the derivation of quantitative values ​​to characterize the disease, which can be correlated with other clinical parameters and can serve as the basis for disease staging, outcome prediction, and treatment decisions for the patient.

[0020] With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Figure 1 A block diagram illustrating an overview of an example of an automatic diagnostic system 100 is provided according to one or more embodiments. Figure 1 As shown, the diagnostic system 100 includes an image evaluation module 110, which is operable to process and analyze a diagnostic image 120 and provide a diagnostic output 130. The image evaluation module 110 includes several modules (e.g., algorithms) (e.g., image registration, region of interest (ROI) identification, ROI processing and quantification, which are referred to herein as Figure 2 , which is further described herein. The diagnostic system 100 so equipped enables robust derivation of comparable quantitative values ​​from variable imaging, such as portable chest x-ray images, which have significant variability in acquisition geometry, superimposed structures, patient shape, and positioning. The diagnostic output 130 can include, for example, visualizations, such as images enhanced (e.g., with contours of lung regions), features derived from ROI processing (e.g., histograms), quantifications (e.g., scores based on feature analysis), and timelines showing progression or disease progression over time. Other diagnostic outputs are also possible.

[0021] Diagnostic image 120 is an image generated by a diagnostic imaging system, and can be an image of multiple types or modalities. Typically, diagnostic image 120 will include two-dimensional (2D) chest imaging (e.g., chest X-ray (including upright or supine), computed tomography (CT) scan, forward projection CT scan, X-ray projection from an image-guided therapy (IGT) system, and / or images obtained via other imaging techniques (e.g., ultrasound). As another example, in some embodiments, diagnostic image 120 is obtained from an X-ray system (e.g., conventional (absorption), dual energy / spectral (detector or tube-based) or phase contrast X-ray system). In some embodiments, in addition to or in place of diagnostic image 120, non-conventional imaging provides different images (e.g., soft tissue / bone images for spectrum, or absorption / phase / dark field images for phase contrast), which are used in one or more processing tasks described herein. For example, when diagnostic image 120 is generated according to CT system scan / forward projection CT, the CT system can also be or include spectrum / phase contrast imaging. Diagnostic image 120 generally relates to the patient's condition. For example, diagnostic image 120 can be a chest X-ray related to the patient's lung condition (e.g., whether there is pneumonia, COVID-19, or other lung diseases / conditions (including chronic obstructive pulmonary disease (COPD), tuberculosis, pleural effusion, pneumothorax, etc.)).

[0022] With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Figure 2 A block diagram illustrating an example of an automatic diagnostic system 200 according to one or more embodiments is provided. In an embodiment, the automatic diagnostic system 200 corresponds to the automatic diagnostic system 100 ( Figure 1 , already discussed). The diagnostic system 200 is operable to automatically evaluate the diagnostic image 120 ( Figure 1 , already discussed) and provide diagnostic output (e.g., Figure 1 Diagnostic output 130 in , already discussed). Figure 2 As shown, the diagnostic system 200 includes an image registration module 210, a region of interest (ROI) identification module 220, a ROI processing module 230, a quantification module 240, and a diagnostic output module 250. In some embodiments, the ROI processing module 230 and the quantification module 240 are performed via a mapping module. In some embodiments, the diagnostic system 200 includes one or more of a bone removal module 260 and / or an intensity normalization module 270.

[0023] The image registration module 210 is operable to reduce or eliminate variability between different diagnostic images 120, the variability arising, for example, based on imaging settings, acquisition geometry, overlay structures, changes in patient positioning and other such changes (e.g., changes that typically occur in portable X-ray units), changes in patient size / anatomy, etc. Performing image registration as described herein enhances the ability to compare and evaluate diagnostic images acquired from a patient over time, acquired from different patients, etc. in an automated manner. In an embodiment, registration of diagnostic images 120 (e.g., portable chest X-ray images) is performed relative to an average standard lung image (e.g., an anatomical atlas) to achieve equivalent, anatomically aligned lung domain representations. For example, in some embodiments, diagnostic image registration is accomplished via a method of a lung contour probability map based on a trained convolutional neural network (CNN) model. The method is operable to automatically register diagnostic images 120 according to (i) collimation, (ii) patient rotation, and (iii) inspiratory state of chest PA radiographs through elastic deformation of the images (e.g., image warping). Anatomical features are robustly detected by a combination of three convolutional neural networks and two probabilistic anatomical atlases. An example of such a technique is described in Jens von Berg et al., "Robust chest x-ray quality assessment using convolutional neural networks and atlas regularization" (Medical Imaging 2020: Image Processing, Vol. 11313, International Society for Optics and Photonics, 2020, which is incorporated herein by reference in its entirety). In some embodiments, other image registration techniques (one or more) are employed.

[0024] In some embodiments, an additional contour detection step is included (as part of the image registration process) to improve the detection of lung contours in diagnostic images. For example, additional contour detection is performed and the results are provided as input to the atlas registration process. Performing the additional contour detection step enhances lung contour detection in images (e.g., images produced by portable (i.e., mobile) chest X-ray systems, which are typically of lower image quality than standard chest radiographs). Reference is now made to Figure 3, with reference to the components and features described herein (including but not limited to the drawings and associated descriptions), the application of additional contour detection steps according to one or more embodiments is illustrated using an example of a chest X-ray image. The image on the left shows an example diagnostic chest X-ray image 310. The middle image illustrates an inverse contour probability map 320 output by a trained CNN, where dark values ​​correspond to high contour probabilities for detected lung contours 322. The image 330 on the right illustrates contour points 332 detected by analyzing the contour probability map 320 superimposed on the original image. For example, the contour points 332 are detected by a separate CNN, multiple separate tasks of a single CNN, or by a contour point classification task, so that it is known whether a point belongs to a certain segment of the contour. These contour points are then used to map the image to an atlas. The contour points 332 more accurately identify the lung contours, thereby illustrating the enhancement provided by the additional lung contour detection.

[0025] Return now Figure 2 , the ROI identification module 220 is operable to identify and extract or isolate the (one or more) regions of interest in the diagnostic image 120 (as registered via the image registration module 210) for further analysis. For example, for a lung image, the (one or more) regions of interest are typically the lung fields (both lungs). In an embodiment, the (one or more) regions of interest are identified, for example, based on images output from an image registration process. For example, in some embodiments, the lung fields are identified and isolated based on drawing a mask on the atlas mapping image and then using the mask to isolate the (one or more) regions of interest in the registered diagnostic image. For example, in an embodiment, masking is performed by a clinical expert with computer assistance provided via a user interface. As another example, in some embodiments, the lung fields are identified and isolated based on the lung contours identified during the image registration process. In an embodiment, these techniques are combined.

[0026] Reference now Figure 4, with reference to the components and features described herein (including but not limited to the figures and associated descriptions), examples of chest X-ray images of a patient taken at two different times illustrate the identification and isolation of ROIs according to one or more embodiments. The diagnostic image of the patient at the first time (Case A) is in the top row, and the image of the patient at the second time (Case B) is in the bottom row. For each case, image 410 in the left column shows the original diagnostic chest X-ray image. In the next column, image 420 shows the result of registering image 410 to an anatomical atlas. Image 420 also shows the result of applying an optional bone removal process. The optional bone removal module 260 will be further described below. In the next column, image 430 shows the result of applying an optional intensity normalization process to image 420. The optional intensity normalization module 270 will be further described below. In the right column, image 440a shows the identified and isolated lung fields for patient case A, and image 440b shows the identified and isolated lung fields for patient case B.

[0027] In some embodiments, a domain of interest is identified that contains the region(s) of interest. The domain of interest (region(s) of interest) is then divided into a plurality of regions (e.g., subdivided into sub-regions). In embodiments, the division of the domain of interest into a plurality of regions (e.g., a plurality of sub-regions) is based on anatomical landmarks. For example, a lung field (e.g., Figure 4 The lung fields shown in images 440a and 440b of the present invention are divided into a plurality of regions or sub-regions. In an example, the division into a plurality of regions or sub-regions is based on anatomical landmarks (e.g., dividing each lung into approximately equivalent lung volumes per region) or based on fitting a lung model (e.g., a 3D lung lobe model to a 2D projection image).

[0028] With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Figures 5A-5D An example of identifying a ROI and dividing the ROI into sub-regions according to one or more embodiments is illustrated. Figure 5A , shows an example of an X-ray chest image 510, in which an overlay of the identified lung fields (light grey overlay) is used as a region of interest (region of interest). Figure 5B The figure shows that roughly equivalent lung volumes will come from each region based on the division of each lung into Figure 5A An example of a region of interest subdivided into six (6) sub-regions, labeled 1 through 6 (image 520). Figure 5C The figure shows the relationship between the lung model and the Figure 5A Another example of a region of interest of is subdivided into six (6) sub-regions, labeled 1 to 6 (image 530). In some embodiments, the fitting is based on identifying specific ribs in the thorax from the original image. Figure 5D The diagram shows the Figure 5B A graphical depiction 540 ( Figure 5D The shading in FIG. 1 is merely to help illustrate the area covered by each zone 1-6 and does not reflect the condition of the patient's lungs).

[0029] Return now Figure 2 , the ROI processing module 230 can be operated to process the (one or more) regions of interest identified in the ROI identification module 220 to generate a feature distribution based on the (one or more) regions of interest. As an example, generating the feature distribution can include generating a histogram of the (one or more) regions of interest, which provides a distribution of the intensity of pixels in the (one or more) regions. With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Figure 6 Illustrated is an example of a histogram 600 for a region of interest in accordance with one or more embodiments. Figure 6 A histogram of pixel intensity (X-axis) is depicted, which shows the number of pixels in the ROI having a particular intensity (Y-axis). In some embodiments, where the region of interest is divided into sub-regions, a histogram is generated for each sub-region. Figure 7 An example of a histogram generated for a plurality of sub-regions is illustrated (as further described herein). In an embodiment, one or more other characteristic distributions are generated in addition to or as an alternative to the histogram. Examples of such other characteristic distributions include local intensity statistical distributions (e.g., mean and standard deviation), edge distributions (e.g., via edge detection), area estimates, etc.

[0030] Return now Figure 2, the quantification module 240 is operable to analyze the feature distribution to determine a quantification (e.g., a metric, a score, etc.). Wherein, the quantification reflects the condition of the patient related to the disease or condition being considered. As an example, in some embodiments, the feature distribution includes an intensity histogram, and the quantification module 240 analyzes the resulting histogram to determine the peak location and the width of the peak area. Such analysis can be used to identify disease conditions or progression. For example, if a histogram of the lung fields of a chest x-ray for a patient is obtained, and if the peak of the histogram has a relatively low intensity, this tends to indicate that the patient's lungs are relatively clear, and therefore the patient's lungs are in good condition. As another example, if a histogram of the lung fields of a chest x-ray for a patient is obtained, and if the peak of the histogram is of medium to high intensity, this tends to indicate that the lung fields of the x-ray are at least partially blurred, and therefore the patient's lungs are in medium to poor condition. In some embodiments, other analyses (e.g., multi-Gaussian distribution fitting) are performed on the feature distribution to provide a more fine-grained analysis of the features. In some embodiments, a histogram is obtained for each of a plurality of sub-regions of the lung fields and the histograms are analyzed, which provides a more fine-grained evaluation (based on multiple histograms) than analyzing a single histogram of the lung fields.

[0031] In an embodiment, the quantification module 240 can also be operated to provide a score indicating a disease condition or a stage of disease progression. The score can be based on markers such as comparing the feature analysis with a similar analysis for patients with known conditions, a predetermined scale (e.g., based on professional or expert experience), etc. For example, in some embodiments, the feature distribution includes an intensity histogram and an analysis is performed (e.g., peak position / width, multiple Gaussian fits, etc.), in which case the quantification module 240 determines a score based on the analysis. The score includes, for example, one or more numerical scores indicating the severity of the disease (e.g., a score of 1-5), a qualitative rating indicating the condition (e.g., a series of qualitative ratings from good to poor), etc. In some embodiments, the score can be based on a comparison of the feature analysis results with one or more thresholds.

[0032] For example, when a histogram is obtained for the lung field of a chest X-ray of a patient and the peak of the histogram has a relatively low intensity (which tends to indicate that the patient's lungs are relatively clear and therefore the patient's lungs are in good condition), a "good" score is assigned. As another example, when a histogram is obtained for the lung field of a chest X-ray of a patient and the peak of the histogram has a high intensity and a relatively wide width (which tends to indicate that the lung field of the X-ray is very blurry and therefore the patient's lungs are in poor condition), a "poor" score is assigned.

[0033] In some embodiments, the score is based on an evaluation of multiple feature distributions. For example, in the case where a histogram is obtained for each of the multiple sub-regions of the lung field and the histogram is analyzed, a sub-score is generated for each sub-region, and the sub-scores are combined to generate a total score. In some embodiments, the score is determined using a location related to the feature. For example, in the case where a histogram is obtained for each of the multiple sub-regions of the lung field and the histogram is analyzed, each sub-region receives a sub-score, wherein the sub-score is weighted based on the location. An example of a scoring scheme related to COVID-19 patients is discussed in A. Borghese's "COVID-19outbreak in Italy: experimental chest X-ray scoring system for quantifyingand monitoring disease progression" (Radiology, 2020, [https: / / doi.org / 10.1007 / s11547-020-01200-3], which is incorporated herein by reference in its entirety). In some embodiments, a neural network is used to evaluate feature distributions and provide scores, and the neural network is trained using data such as clinical parameters, radiologist ratings, etc.

[0034] The diagnostic output module 250 is operable to provide an output (result) of a quantification (evaluation) process. For example, in an embodiment, an output / result comprising a diagnosis of a disease or condition is provided. As another example, in an embodiment, the output / result comprises a progression or staging of a disease / condition. In an embodiment, the output / result comprises a score indicating the severity or progression of the disease / condition. In an embodiment, the output / result in the form of a diagnosis, staging / progression, score, etc. is provided over time (e.g., via a timeline). The output / result provided as described herein has a variety of uses, such as to perform triage of new patients and / or track the progression of a disease over time, etc.

[0035] In embodiments, the diagnostic output module 250 provides visualizations of outputs / results, including, for example, those identified herein, e.g., a diagnosis of a disease or condition, a progression or staging of a disease / condition, a score indicating severity or progression of a disease / condition, and / or a timeline showing the diagnosis, staging / progression, score over time, etc. In embodiments, the visualization includes, for example, one or more of: a visual display (e.g., a screen on a laptop, tablet, smartphone, etc.), a web-based dashboard, a document printout (including electronically generated documents), etc.

[0036] With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Figure 7An example visual display 700 of a regional diagnostic output is provided in accordance with one or more embodiments. Figure 7 A display (e.g., screen display) in includes images and histograms for normal (healthy) patients on the right side of the display and a set of images and histograms for COVID-19 patients on the left / center of the display (showing progression over several time samples). The display illustrates a visualization of a regional statistical analysis of the distribution of grayscale values ​​of pixels in a lung field region for diseased and healthy patients, including showing changes in the distribution due to COVID-19 disease progression.

[0037] More specifically, for a normal patient, the first row of the display (below the header) shows the identified / isolated lung fields (two lungs) for a normal patient and a graphical illustration (similar to Figure 5D ), which illustrates how the lung field is divided into 6 sub-regions (numbered 1-6). Below the first row is a set of six histograms, each of which represents an intensity histogram for one of the six sub-regions of the lung field that have been isolated (based on the corresponding image portion corresponding to the sub-region). Each histogram is labeled (1)-(6) (corresponding to the labeled sub-regions 1-6). Figure 7 The histograms illustrated in FIG. 1 are encoded by grayscale values ​​indicating scores ranging from "good" to "poor" (as shown to the left of the "healthy" histogram). As reflected in the histograms for normal patients, the peaks in each histogram are associated with relatively low intensities (scores of "good"), indicating that the patient is healthy. In an embodiment, color coding (e.g., green, yellow, orange, red) is used instead of grayscale to represent scores from "good" to "poor."

[0038] Steering Figure 7On the left side of the figure, there are images and histograms taken for COVID-19 patients at three times (labeled T1, T2, and T3 in the figure). The first two columns show the images of the identified / isolated lung fields at time T1 (top row) and a graph showing how the lung fields are divided into six sub-regions. The graph corresponds to the same graph and the same sub-division into six sub-regions as the above-mentioned healthy patients. The next three rows show six histograms corresponding to the intensity histograms of each of the six sub-regions for COVID patients at time T1. The histograms are arranged in the same order as shown on the right side of the figure for healthy patients. The histograms for time T1 show that for various lung sub-regions, the histogram evaluation ranges from "good" (top two histograms) to relatively "poor" (bottom two histograms) (indicating that the patient's lungs are relatively clear, but other parts are negatively affected by the disease). Similarly, the next two columns show the identified / isolated lung field images (top row) and dividing line graphics at time T2, followed by six histograms corresponding to the intensity histograms for each of the six sub-regions for COVID patients at time T2. It is worth noting that the histogram at the bottom right of time T2 shows a relatively high intensity peak and a score of "poor". The histograms and (therefore) the scores show the deterioration of the patient's condition between time T1 and T2. The next two columns also show the identified / isolated lung field images (top row) and dividing line graphics at time T3, followed by six histograms corresponding to the intensity histograms for each of the six sub-regions for COVID patients at time T3. It is worth noting that the three histograms at the bottom right of time T3 show relatively high intensity peaks and "poor" scores. The histograms and (therefore) the scores show the further deterioration of the patient's condition between time T2 and T3.

[0039] With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Figure 8 An example visual display 800 of a diagnostic output timeline is provided according to one or more embodiments. Display 800 (e.g., a web-based dashboard) includes a series of timelines for a group of COVID patients, indicated by coded numbers on the left side of the timeline. For each patient, there is a line accompanied by a series of points indicating the results or scores automatically obtained from imaging events on a specific date via the diagnostic system 200 (fictitious dates are shown along the top of the display). In an embodiment, this series of points is encoded with markers such as numbers, grayscale, color, etc. (indicating the diagnostic score or disease stage on a specific date). Such a timeline can be compared to map the disease stage (including, for example, progression or recovery) of different patients over time, and such a timeline is used in an embodiment to, for example, identify trends in disease progression or recovery.

[0040] Return now Figure 2 In some embodiments, the optional bone removal module 260 is operable to remove ribs and clavicles that overlay lung tissue in the diagnostic image after the image registration process. An example of a bone removal technique is described in Jens von Berg, "A novel bonesuppression method that improves lung nodule detection" (International Journal of Computer-Assisted Radiology and Surgery 11.4, 2016, pp. 641-655, which is incorporated herein by reference in its entirety). Figure 4 An example illustrating the result of using a bone removal technique - image 420 is shown (bone removal is performed after image registration). In some embodiments, other bone removal technique(s) are used as an alternative.

[0041] In some embodiments, optional intensity normalization module 270 can be operated to obtain a standardized representation of density values ​​at pixel gray levels. In one example, intensity normalization is performed by obtaining an intensity histogram of pixel values ​​in the lung fields and spine and normalizing the pixel values ​​so that the histogram becomes substantially flat across all intensities. Figure 4 An example of an image 430 illustrating the results of using an intensity normalization technique is shown (intensity normalization is performed after bone removal). In some embodiments, other intensity normalization techniques (one or more) are used as alternatives. In some embodiments, intensity normalization is performed after bone removal; in some embodiments, intensity normalization is performed before bone removal; in some embodiments, intensity normalization is performed without bone removal; in some embodiments, bone removal is performed without intensity normalization. In some embodiments, one or more of bone removal and / or intensity normalization is performed before identifying the ROI via the ROI identification module 220.

[0042] Although certain examples have been provided herein (including examples of diagnostic imaging related to COVID-19 patients), the improved techniques described herein are useful for evaluating a variety of lung diseases / conditions, including pneumonia, COVID-19, chronic obstructive pulmonary disease (COPD), tuberculosis, pleural effusion, pneumothorax, etc. In some embodiments, one or more modules of the diagnostic system 200 are implemented at least in part via a trained neural network (trained using data suitable for the particular module). For example, as described herein, in some embodiments, one or more trained neural networks are used to at least partially implement the image registration module 210. Additionally, as described herein, in some embodiments, the trained neural network is used to at least partially implement the quantification module 240.

[0043] Some or all of the components of the diagnostic system 100 and / or the diagnostic system 200 can be implemented using one or more of a central processing unit (CPU), a graphics processing unit (GPU), an artificial intelligence (AI) accelerator, a field programmable gate array (FPGA) accelerator, an application-specific integrated circuit (ASIC), and / or implemented via a processor with software, or implemented in a combination of a processor with software and an FPGA or ASIC. More specifically, the components of the diagnostic system 100 and / or the diagnostic system 200 can be implemented in one or more modules as a set of programs or logic instructions stored in a machine or computer readable storage medium (e.g., a random access memory (RAM), a read-only memory (ROM), a programmable ROM (PROM), firmware, flash memory, etc.); can also be implemented in hardware; can also be implemented in any combination thereof. For example, a hardware implementation can include a configurable logic unit, a fixed function logic unit, or any combination thereof. Examples of configurable logic units include appropriately configured programmable logic arrays (PLAs), FPGAs, complex programmable logic devices (CPLDs), and general-purpose microprocessors. Examples of fixed function logic cells include appropriately configured ASICs, combinational logic circuits, and sequential logic circuits. Configurable logic cells or fixed function logic cells can be implemented using complementary metal oxide semiconductor (CMOS) logic circuits, transistor-transistor logic (TTL) logic circuits, or other circuits.

[0044] For example, computer program code to be executed by diagnostic system 100 and / or diagnostic system 200 can be written in any combination of one or more programming languages, including object-oriented programming languages ​​(e.g., JAVA, SMALLTALK, C++, etc.) and conventional procedural programming languages ​​(e.g., "C" programming language or similar programming languages). In addition, program or logic instructions may include assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, state setting data, configuration data for integrated circuits, state information that personalizes other structural components inherent to electronic circuits and / or hardware (e.g., main processor, central processing unit / CPU, microcontroller, etc.).

[0045] With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Fig. 9A A flowchart illustrating an example method 900 for performing automatic diagnostic image evaluation according to one or more embodiments is provided. The method 900 can generally be implemented in the diagnostic system 100 ( Figure 1 , already discussed) and / or diagnostic system 200 ( Figure 2, discussed above). More specifically, method 900 can be implemented as one or more modules as a set of logic instructions stored in a machine or computer readable storage medium (e.g., RAM, ROM, PROM, firmware, flash memory, etc.); can also be implemented in hardware; can also be implemented in any combination thereof. For example, a hardware implementation can include a configurable logic unit, a fixed-function logic unit, or any combination thereof. Examples of configurable logic units include appropriately configured PLAs, FPGAs, CPLDs, and general-purpose microprocessors. Examples of fixed-function logic units include appropriately configured ASICs, combinational logic circuits, and sequential logic circuits. Configurable logic units or fixed-function logic units can be implemented using CMOS logic circuits, TTL logic circuits, or other circuits.

[0046] For example, computer program code to perform method 900 and / or functions associated therewith can be written in any combination of one or more programming languages, including object-oriented programming languages ​​(e.g., JAVA, SMALLTALK, C++, etc.) and conventional procedural programming languages ​​(e.g., "C" programming language or similar programming languages). In addition, program or logic instructions may include assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, state setting data, configuration data for integrated circuits, state information that personalizes other structural components inherent to electronic circuits and / or hardware (e.g., main processor, central processing unit / CPU, microcontroller, etc.).

[0047] Illustrated processing block 910 provides for receiving a diagnostic image related to a patient's condition. In an embodiment, the diagnostic image corresponds to diagnostic image 120 ( Figure 1-2 , discussed above). The illustrated processing block 915 provides for performing registration of the diagnostic image with reference to the anatomical structure to form a registered image. In some embodiments, the anatomical structure relates to a lung field. In some embodiments, the anatomical structure is associated with an anatomical atlas. The illustrated processing block 920 provides for identifying one or more regions of interest from the registered image. In an embodiment, the one or more regions of interest relate to a lung field of the patient. The illustrated processing block 925 provides for generating a feature distribution based on the one or more regions of interest. The illustrated processing block 930 provides for analyzing the feature distribution to determine a quantification that reflects the patient's condition. The illustrated processing block 935 provides for providing a diagnostic output based on the quantification. In some embodiments, providing a diagnostic output includes one or more of the following: determining a diagnostic score for the patient's condition based on the quantification, providing a visualization showing one or more regions of interest using the feature distribution, and / or providing a visualization showing the progression of the patient's condition over time.

[0048] In some embodiments, illustrated processing block 940 provides for performing a bone removal process on the registered image. In some embodiments, illustrated processing block 945 provides for performing an image normalization process. In some embodiments, intensity normalization is performed after bone removal; in some embodiments, intensity normalization is performed before bone removal; in some embodiments, intensity normalization is performed without bone removal; in some embodiments, bone removal is performed without intensity normalization. In some embodiments, one or more of bone removal and / or intensity normalization is performed before ROI identification (block 920).

[0049] With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Fig. 9B A flowchart illustrating an example method 950 for identifying one or more regions of interest according to one or more embodiments is provided. In an embodiment, the method 950 can replace the illustrated processing block 920 ( Fig. 9A , already discussed). Method 950 can generally be implemented in diagnostic system 100 ( Figure 1 , already discussed) and / or diagnostic system 200 ( Figure 2 , discussed above). More specifically, method 950 can be implemented as one or more modules as a set of logic instructions stored in a machine or computer readable storage medium (e.g., RAM, ROM, PROM, firmware, flash memory, etc.); can also be implemented in hardware; can also be implemented in any combination thereof. For example, a hardware implementation can include a configurable logic unit, a fixed-function logic unit, or any combination thereof. Examples of configurable logic units include appropriately configured PLAs, FPGAs, CPLDs, and general-purpose microprocessors. Examples of fixed-function logic units include appropriately configured ASICs, combinational logic circuits, and sequential logic circuits. Configurable logic units or fixed-function logic units can be implemented using CMOS logic circuits, TTL logic circuits, or other circuits.

[0050] For example, computer program code to perform method 950 and / or functions associated therewith can be written in any combination of one or more programming languages, including object-oriented programming languages ​​(e.g., JAVA, SMALLTALK, C++, etc.) and conventional procedural programming languages ​​(e.g., "C" programming language or similar programming languages). In addition, program or logic instructions may include assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, state setting data, configuration data for integrated circuits, state information that personalizes other structural components inherent to electronic circuits and / or hardware (e.g., main processor, central processing unit / CPU, microcontroller, etc.).

[0051] Illustrated processing block 955 provides for identifying a region of interest in the registered image, the region of interest comprising one or more regions of interest.Illustrated processing block 960 provides for dividing the region of interest into a plurality of sub-regions.

[0052] With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Fig. 9C A flowchart illustrating an example method 970 for generating a feature distribution according to one or more embodiments is provided. In an embodiment, the method 970 can replace the illustrated processing block 925 ( Fig. 9A , already discussed). Method 970 can generally be implemented in diagnostic system 100 ( Figure 1 , already discussed) and / or diagnostic system 200 ( Figure 2 , discussed above). More specifically, method 970 can be implemented as one or more modules as a set of logic instructions stored in a machine or computer readable storage medium (e.g., RAM, ROM, PROM, firmware, flash memory, etc.); can also be implemented in hardware; can also be implemented in any combination thereof. For example, a hardware implementation can include a configurable logic unit, a fixed-function logic unit, or any combination thereof. Examples of configurable logic units include appropriately configured PLAs, FPGAs, CPLDs, and general-purpose microprocessors. Examples of fixed-function logic units include appropriately configured ASICs, combinational logic circuits, and sequential logic circuits. Configurable logic units or fixed-function logic units can be implemented using CMOS logic circuits, TTL logic circuits, or other circuits.

[0053] For example, the computer program code to perform the method 970 and / or the functions associated therewith can be written in any combination of one or more programming languages, including object-oriented programming languages ​​(e.g., JAVA, SMALLTALK, C++, etc.) and conventional procedural programming languages ​​(e.g., "C" programming language or similar programming languages). In addition, the program or logic instructions may include assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, state setting data, configuration data for integrated circuits, state information that personalizes other structural components inherent to electronic circuits and / or hardware (e.g., main processor, central processing unit / CPU, microcontroller, etc.).

[0054] Illustrated processing block 975 provides for generating an intensity histogram for each of the one or more regions of interest.In an embodiment, the intensity histogram provides a distribution of intensities of pixels in the one or more regions of interest.

[0055] With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Fig.9D A flowchart illustrating an example method 980 for analyzing feature distribution according to one or more embodiments is provided. In an embodiment, the method 980 can replace the illustrated processing block 930 ( Fig. 9A , already discussed). Method 980 can generally be implemented in diagnostic system 100 ( Figure 1 , already discussed) and / or diagnostic system 200 ( Figure 2 , discussed above). More specifically, method 980 can be implemented as one or more modules as a set of logic instructions stored in a machine or computer readable storage medium (e.g., RAM, ROM, PROM, firmware, flash memory, etc.); can also be implemented in hardware; can also be implemented in any combination thereof. For example, a hardware implementation can include a configurable logic unit, a fixed-function logic unit, or any combination thereof. Examples of configurable logic units include appropriately configured PLAs, FPGAs, CPLDs, and general-purpose microprocessors. Examples of fixed-function logic units include appropriately configured ASICs, combinational logic circuits, and sequential logic circuits. Configurable logic units or fixed-function logic units can be implemented using CMOS logic circuits, TTL logic circuits, or other circuits.

[0056] For example, computer program code to perform method 980 and / or functions associated therewith can be written in any combination of one or more programming languages, including object-oriented programming languages ​​(e.g., JAVA, SMALLTALK, C++, etc.) and conventional procedural programming languages ​​(e.g., "C" programming language or similar programming languages). In addition, program or logic instructions may include assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, state setting data, configuration data for integrated circuits, state information that personalizes other structural components inherent to electronic circuits and / or hardware (e.g., main processor, central processing unit / CPU, microcontroller, etc.).

[0057] Illustrated processing block 985 provides for determining one or more metrics based on the feature distribution.In some embodiments, quantification is based on comparing the one or more metrics to one or more predetermined thresholds.

[0058] With reference to the components and features described herein (including but not limited to the drawings and associated descriptions), Fig.10 is a diagram illustrating a computing system 1000 for use in diagnostic system 100 and / or diagnostic system 200 according to one or more embodiments. Fig.10Certain components are shown, but computing system 1000 can include additional components or multiple components connected in various ways. It should be understood that not all examples necessarily include Fig.10 Each component shown. Fig.10 As shown, computing system 1000 includes one or more processors 1002, I / O subsystem 1004, network interface 1006, memory 1008, data storage device 1010, artificial intelligence (AI) accelerator 1012, user interface 1016 and / or display 1020. These components are coupled, connected or otherwise communicate data via interconnect 1014. In some embodiments, computing system 1000 is connected to a separate display interface. Computing system 1000 can implement one or more components or features of diagnostic system 100, diagnostic system 200 and / or the like as referenced herein. Figures 1 to 9D Any component or method described.

[0059] The processor 1002 includes one or more processing devices, such as a microprocessor, a central processing unit (CPU), a fixed application specific integrated circuit (ASIC) processor, a reduced instruction set computing (RISC) processor, a complex instruction set computing (CISC) processor, a field programmable gate array (FPGA), a digital signal processor (DSP), etc., and associated circuits, logic units and / or interfaces. Where necessary or appropriate, the processor 1002 can include or be connected to a memory (e.g., memory 1008) that stores executable instructions and / or data. The processor 1002 can execute such instructions to implement, control, operate, or interface with any component or feature of the diagnostic system 100, the diagnostic system 200, and / or the diagnostic system 100 referenced herein. Figures 1 to 9D The processor 1002 can transmit, send messages, requests, notifications, data, etc. to other devices / receive messages, requests, notifications, data, etc. from other devices. The processor 1002 can be implemented as any type of processor capable of performing the functions described herein. For example, the processor 1002 can be implemented as (one or more) single-core or multi-core processors, digital signal processors, microcontrollers, or other processors or processing / control circuits. The processor can include embedded instructions (e.g., processor code).

[0060] I / O subsystem 1004 includes circuits and / or components suitable for facilitating input / output operations with processor 1002 , memory 1008 , and other components of computing system 1000 .

[0061] The network interface 1006 includes suitable logic units, circuits and / or interfaces that use one or more communication network protocols to send and receive data on one or more communication networks. The network interface 1006 can operate under the control of the processor 1002 and can send / receive various requests and messages to / from one or more other devices. The network interface 1006 can include wired or wireless data communication capabilities; these capabilities can support data communication with a wired or wireless communication network (e.g., network 1007, and also includes the Internet, a wide area network (WAN), a local area network (LAN), a wireless personal area network, a wide body area network, a cellular network, a telephone network, any other wired or wireless network for sending and receiving data signals, or any combination thereof (including, for example, a Wi-Fi network or a corporate LAN)). The network interface 1006 can support communication via a short-range wireless communication field (e.g., Bluetooth, NFC or RFID). Examples of network interface 1006 include, but are not limited to, one or more of an antenna, a radio frequency transceiver, a wireless transceiver, a Bluetooth transceiver, an Ethernet port, a universal serial bus (USB) port, or any other device configured to send and receive data.

[0062] The memory 1008 includes suitable logic units, circuits and / or interfaces to store executable instructions and / or data as necessary or appropriate, which when executed implement, control, operate the diagnostic system 100, any component or feature of the diagnostic system 200 and / or any other component or feature referenced herein. Figures 1 to 9D 1000. The memory 1008 may be implemented as any type of volatile or non-volatile memory or data storage device capable of performing the functions described herein, and may include random access memory (RAM), read-only memory (ROM), write-once-read-many memory (e.g., EEPROM), removable storage drives, hard disk drives (HDDs), flash memory, solid-state memory, etc., and any combination thereof. In operation, the memory 1008 may store various data and software (e.g., operating systems, applications, programs, libraries, and drivers) used during operation of the computing system 1000. Thus, the memory 1008 may include at least one non-transitory computer-readable medium, the non-transitory computer-readable medium including instructions that, when executed by the computing system 1000, cause the computing system 1000 to perform operations to execute the diagnostic system 100, the diagnostic system 200, and / or the diagnostic system 100 referred to herein. Figures 1 to 9D The memory 1008 can be communicatively coupled to the processor 1002 directly or via the I / O subsystem 1004.

[0063] Data storage device 1010 can include any type of one or more devices configured for short-term or long-term storage of data, such as memory devices and circuits, memory cards, hard drives, solid-state drives, non-volatile flash memory, or other data storage devices. Data storage device 1010 can include or be configured as a database (e.g., a relational database or a non-relational database) or a combination of more than one database. In some examples, the database or other data storage device can be physically separated from computing system 1000 and / or remote from computing system 1000, and / or can be located in / on another computing device, a database server, a cloud-based platform, or any storage device that communicates data with computing system 1000.

[0064] Artificial intelligence (AI) accelerator 1012 includes suitable logic units, circuits, and / or interfaces to accelerate artificial intelligence applications (e.g., artificial neural networks, machine vision, and machine learning applications, including operating the above applications through parallel processing techniques). In one or more examples, AI accelerator 1012 can include a graphics processing unit (GPU). AI accelerator 1012 can implement one or more components or features of diagnostic system 100, diagnostic system 200, and / or the diagnostic system 100 described herein. Figures 1 to 9D In some examples, computing system 1000 includes a second AI accelerator (not shown).

[0065] The interconnect 1014 includes any one or more separate physical buses, point-to-point connections, or both connected through appropriate bridges, adapters, or controllers. The interconnect 1014 can include, for example, a system bus, a peripheral component interconnect (PCI) bus, a HyperTransport or Industry Standard Architecture bus, a Small Computer System Interface (SCSI) bus, a Universal Serial Bus (USB), an IIC (I2C) bus, or an Institute of Electrical and Electronics Engineers (IEEE) Standard 694 bus (e.g., "Firewire"), or any other interconnect suitable for coupling or connecting components of the computing system 1000.

[0066] The user interface 1016 includes code to present information or screens to a user on a display and receive input (including commands) from the user via an input device. The display 1020 can be any type of device (e.g., a computer monitor, a flat panel display, or a mobile device screen) for presenting visual information, and can include a liquid crystal display (LCD), a light emitting diode (LED) display, a plasma panel, or a cathode ray tube display, etc. The display 1020 can include a display interface for communicating with the display. In some examples, the display 1020 can include a display interface for communicating with a display external to the computing system 1000.

[0067] In some examples, one or more of the illustrative components of computing system 1000 can be incorporated (in whole or in part) into another component, or form part of another component. For example, memory 1008 or a portion thereof can be incorporated into processor 1002. As another example, user interface 1016 can be incorporated into code in processor 1002 and / or memory 1008. In some examples, computing system 1000 can be embodied as, but not limited to, a mobile computing device, a smartphone, a wearable computing device, an Internet of Things device, a laptop, a tablet, a notebook computer, a workstation, a server, a multi-processor system, and / or a consumer electronic device. In some examples, computing system 1000 or portions thereof are implemented in one or more modules using circuit technology (e.g., application specific integrated circuit (ASIC) technology, complementary metal oxide semiconductor (CMOS) technology, or transistor-transistor logic (TTL) technology, or any combination thereof) as a set of logic instructions stored in a machine or computer readable storage medium (e.g., random access memory (RAM), read-only memory (ROM), programmable ROM (PROM), firmware, flash memory, etc.); implemented in a configurable logic unit (e.g., a programmable logic array (PLA), a field programmable gate array (FPGA), a complex programmable logic device (CPLD)); implemented in fixed function logic hardware.

[0068] Embodiments of each of the above systems, devices, components and / or methods including diagnostic system 100, diagnostic system 200, method 900, method 950, method 970 and / or method 980 and / or any other system components can be implemented in hardware, software or any suitable combination thereof. For example, a hardware implementation may include a configurable logic unit, a fixed function logic unit or any combination thereof. Examples of configurable logic units include appropriately configured PLAs, FPGAs, CPLDs and general-purpose microprocessors. Examples of fixed function logic units include appropriately configured ASICs, combinational logic circuits and sequential logic circuits. Configurable logic units or fixed function logic units can be implemented using CMOS logic circuits, TTL logic circuits or other circuits.

[0069] Alternatively or additionally, all or part of the foregoing systems and / or components and / or methods can be implemented in one or more modules as a set of program or logic instructions stored in a machine or computer readable storage medium (e.g., RAM, ROM, PROM, firmware, flash memory, etc.) to be executed by a processor or computing device. For example, the computer program code that performs the operation of the components can be written in any combination of programming languages ​​applicable / appropriate to one or more operating systems (OS), including object-oriented programming languages ​​(e.g., PYTHON, PERL, JAVA, SMALLTALK, C++, C#, etc.) and conventional procedural programming languages ​​(e.g., "C" programming language or similar programming languages).

[0070] Additional notes and examples:

[0071] Example 1 includes a computer-implemented method comprising: receiving a diagnostic image related to a patient's condition; performing registration of the diagnostic image with reference to an anatomical structure to form a registered image; identifying one or more regions of interest from the registered image; generating a feature distribution based on the one or more regions of interest; analyzing the feature distribution to determine a quantification that reflects the condition of the patient; and providing a diagnostic output based on the quantification.

[0072] Example 2 includes the method of Example 1, further comprising one or more of the following: performing a bone removal process on the registered image; or performing an image normalization process.

[0073] Example 3 includes the method of Example 1 or 2, wherein identifying one or more regions of interest comprises: identifying a region of interest in the registered image, the region of interest containing the one or more regions of interest; and dividing the region of interest into a plurality of sub-regions.

[0074] Example 4 includes the method of Example 1, 2, or 3, wherein generating the feature distribution includes generating an intensity histogram for each of the one or more regions of interest.

[0075] Example 5 includes the method of any of Examples 1-4, wherein analyzing the feature distribution includes determining one or more metrics based on the feature distribution.

[0076] Example 6 includes the method of any of Examples 1-5, wherein the quantifying is based on comparing the one or more metrics to one or more predetermined thresholds.

[0077] Example 7 includes the method of any one of Examples 1-6, wherein providing a diagnostic output comprises one or more of: determining a diagnostic score for the condition of the patient based on the quantification; utilizing the feature distribution to provide a visualization showing the one or more regions of interest; or providing a visualization showing the progression of the condition of the patient over time.

[0078] Example 8 includes a computing system comprising a processor and a memory coupled to the processor, the memory comprising instructions that, when executed by the processor, cause the computing system to perform operations comprising: receiving a diagnostic image related to a condition of a patient; performing registration of the diagnostic image with reference to an anatomical structure to form a registered image; identifying one or more regions of interest from the registered image; generating a feature distribution based on the one or more regions of interest; analyzing the feature distribution to determine a quantification that reflects the condition of the patient; and providing a diagnostic output based on the quantification.

[0079] Example 9 includes the system of Example 8, wherein the instructions, when executed, further cause the computing system to perform operations including one or more of: performing a bone removal process on the registered image; or performing an image normalization process.

[0080] Example 10 includes the system of Example 8 or 9, wherein identifying one or more regions of interest comprises: identifying a region of interest in the registered image, the region of interest containing the one or more regions of interest; and dividing the region of interest into a plurality of sub-regions.

[0081] Example 11 includes the system of Example 8, 9, or 10, wherein generating the feature distribution comprises generating an intensity histogram for each of the one or more regions of interest.

[0082] Example 12 includes the system of any of Examples 8-11, wherein analyzing the feature distribution comprises determining one or more metrics based on the feature distribution.

[0083] Example 13 includes the system of any of Examples 8-12, wherein the quantifying is based on comparing the one or more metrics to one or more predetermined thresholds.

[0084] Example 14 includes the system of any of Examples 8-13, wherein providing a diagnostic output comprises one or more of: determining a diagnostic score for the condition of the patient based on the quantification; utilizing the feature distribution to provide a visualization showing the one or more regions of interest; or providing a visualization showing the progression of the condition of the patient over time.

[0085] Example 15 includes at least one non-transitory computer-readable storage medium comprising instructions that, when executed by a computing system, cause the computing system to perform operations comprising: receiving a diagnostic image related to a condition of a patient; performing registration of the diagnostic image with reference to an anatomical structure to form a registered image; identifying one or more regions of interest from the registered image; generating a feature distribution based on the one or more regions of interest; analyzing the feature distribution to determine a quantification that reflects the condition of the patient; and providing a diagnostic output based on the quantification.

[0086] Example 16 includes at least one non-transitory computer-readable storage medium of Example 15, wherein the instructions, when executed, further cause the computing system to perform operations including one or more of: performing a bone removal process on the registered image; or performing an image normalization process.

[0087] Example 17 includes at least one non-transitory computer-readable storage medium of Example 15 or 16, wherein identifying one or more regions of interest comprises: identifying a domain of interest in the registered image, the domain of interest containing the one or more regions of interest; and dividing the domain of interest into a plurality of sub-regions.

[0088] Example 18 includes at least one non-transitory computer-readable storage medium of Example 15, 16, or 17, wherein generating the feature distribution comprises generating an intensity histogram for each of the one or more regions of interest.

[0089] Example 19 includes at least one non-transitory computer-readable storage medium of any of Examples 15-18, wherein analyzing the feature distribution comprises determining one or more metrics based on the feature distribution, and wherein the quantification is based on comparing the one or more metrics to one or more predetermined thresholds.

[0090] Example 20 includes at least one non-transitory computer-readable storage medium of any of Examples 15-19, wherein providing a diagnostic output comprises one or more of: determining a diagnostic score for the condition of the patient based on the quantification; utilizing the feature distribution to provide a visualization showing the one or more regions of interest; or providing a visualization showing the progression of the condition of the patient over time.

[0091] Example 21 includes an apparatus comprising means for implementing the method of any of Examples 1-7.

[0092] Embodiments are applicable to all types of semiconductor integrated circuit ("IC") chips. Examples of these IC chips include, but are not limited to, processors, controllers, chipset components, programmable logic arrays (PLA), memory chips, network chips, systems on chips (SOC), SSD / NAND controller ASICs, etc. In addition, in some of the drawings, signal conductors are represented by lines. Some lines may be different to indicate more component signal paths; some lines may have digital labels to indicate multiple component signal paths; and / or some lines may have arrows at one or more ends to indicate the main information flow direction. However, this should not be interpreted as limiting. Instead, such added details can be used in conjunction with one or more exemplary embodiments to facilitate easier understanding of the circuit. Any signal line represented, whether or not it has additional information, can actually include one or more signals that can propagate in multiple directions and can be implemented with any suitable type of signal scheme (e.g., digital lines or analog lines implemented using differential pairs, fiber optic lines, and / or single-ended lines).

[0093] Example sizes / models / values / ranges may have been given, but the embodiments are not limited thereto. As manufacturing technology (e.g., lithography) matures over time, it is expected that devices of smaller sizes can be manufactured. In addition, in order to simplify the illustration and discussion, and in order not to obscure certain aspects of the embodiments, the known power / ground connections to the IC chip and other components may or may not be shown in the drawings. In addition, in order to avoid obscuring the embodiments, and also taking into account the fact that the details of the implementation of such block diagram arrangements are highly dependent on the platform in which the embodiments will be implemented (i.e., such details should be within the knowledge of those skilled in the art), the arrangements may be shown in block diagram form. In the case of elaborating specific details (e.g., circuits) to describe example embodiments, it is obvious to those skilled in the art that the embodiments can be implemented without these specific details or with variations of these specific details. Therefore, the description is considered to be illustrative, not restrictive.

[0094] The term "coupled" may be used herein to refer to any type of direct or indirect relationship between the components in question, and may apply to electrical, mechanical, fluid, optical, electromagnetic, electromechanical, or other connections, including logical connections via intermediate components (e.g., device A may be coupled to device C via device B). In addition, unless otherwise specified, the terms "first", "second", etc., are used herein merely to facilitate discussion and do not have a specific temporal or chronological meaning.

[0095] A list of items connected by the term "one or more of..." as used in the present application and claims may mean any combination of the listed terms. For example, the phrase "one or more of A, B, or C" may mean A, B, C; A and B; A and C; B and C; or A; B; C.

[0096] From the foregoing description, those skilled in the art will appreciate that the broad techniques of the embodiments can be implemented in a variety of forms. Therefore, although the embodiments have been described in conjunction with specific examples of the embodiments, the true scope of the embodiments should not be limited thereto, as other modifications will be apparent to those skilled in the art after studying the drawings, the specification, and the appended claims.

Claims

1. A computer-implemented method comprising: receiving diagnostic images related to the patient's condition; performing registration of the diagnostic image with reference to the anatomical structure to form a registered image; identifying one or more regions of interest from the registered images; generating a feature distribution based on the one or more regions of interest; analyzing the characteristic distribution to determine a quantification, the quantification reflecting the condition of the patient; and A diagnostic output is provided based on the quantification.

2. The method according to claim 1, further comprising one or more of the following: performing a bone removal process on the registered image; or Perform image normalization process.

3. The method according to claim 1, wherein: Identifying one or more regions of interest includes: identifying a region of interest in the registered image, the region of interest containing the one or more regions of interest; and The region of interest is divided into a plurality of sub-regions.

4. The method according to claim 1, wherein: Generating the feature distribution includes generating an intensity histogram for each region of interest of the one or more regions of interest.

5. The method according to claim 4, wherein: Analyzing the feature distribution includes determining one or more metrics based on the feature distribution.

6. The method according to claim 5, wherein: The quantification is based on comparing the one or more metrics to one or more predetermined thresholds.

7. The method according to claim 1, wherein: Provides diagnostic output including one or more of the following: determining a diagnostic score for the condition of the patient based on the quantification; providing a visualization showing the one or more regions of interest using the feature distribution; or A visualization is provided showing the progression of the condition of the patient over time.

8. A computing system comprising: processor; as well as a memory coupled to the processor, the memory comprising instructions that, when executed by the processor, cause the computing system to perform operations comprising: receiving diagnostic images related to the patient's condition; performing registration of the diagnostic image with reference to the anatomical structure to form a registered image; identifying one or more regions of interest from the registered images; generating a feature distribution based on the one or more regions of interest; analyzing the characteristic distribution to determine a quantification, the quantification reflecting the condition of the patient; and A diagnostic output is provided based on the quantification.

9. The system according to claim 8, wherein: When executed, the instructions further cause the computing system to perform operations including one or more of the following: performing a bone removal process on the registered image; or Perform image normalization process.

10. The system according to claim 8, wherein: Identifying one or more regions of interest includes: identifying a region of interest in the registered image, the region of interest containing the one or more regions of interest; and The region of interest is divided into a plurality of sub-regions.

11. The system according to claim 8, wherein: Generating the feature distribution includes generating an intensity histogram for each region of interest of the one or more regions of interest.

12. The system according to claim 11, wherein: Analyzing the feature distribution includes determining one or more metrics based on the feature distribution.

13. The system according to claim 12, wherein: The quantification is based on comparing the one or more metrics to one or more predetermined thresholds.

14. The system according to claim 8, wherein: Provides diagnostic output including one or more of the following: determining a diagnostic score for the condition of the patient based on the quantification; providing a visualization showing the one or more regions of interest using the feature distribution; or A visualization is provided showing the progression of the condition of the patient over time.

15. At least one non-transitory computer-readable storage medium comprising instructions that, when executed by a computing system, cause the computing system to perform operations comprising: receiving diagnostic images related to the patient's condition; performing registration of the diagnostic image with reference to the anatomical structure to form a registered image; identifying one or more regions of interest from the registered images; generating a feature distribution based on the one or more regions of interest; analyzing the characteristic distribution to determine a quantification, the quantification reflecting the condition of the patient; and A diagnostic output is provided based on the quantification.

16. The at least one non-transitory computer-readable storage medium of claim 15, wherein: When executed, the instructions further cause the computing system to perform operations including one or more of the following: performing a bone removal process on the registered image; or Perform image normalization process.

17. The at least one non-transitory computer-readable storage medium of claim 15, wherein: Identifying one or more regions of interest includes: identifying a region of interest in the registered image, the region of interest containing the one or more regions of interest; and The region of interest is divided into a plurality of sub-regions.

18. The at least one non-transitory computer-readable storage medium of claim 15, wherein: Generating the feature distribution includes generating an intensity histogram for each region of interest of the one or more regions of interest.

19. The at least one non-transitory computer-readable storage medium of claim 18, wherein: Analyzing the feature distribution includes determining one or more metrics based on the feature distribution, and wherein the quantifying is based on comparing the one or more metrics to one or more predetermined thresholds.

20. The at least one non-transitory computer-readable storage medium of claim 15, wherein: Provides diagnostic output including one or more of the following: determining a diagnostic score for the condition of the patient based on the quantification; providing a visualization showing the one or more regions of interest using the feature distribution; or A visualization is provided showing the progression of the condition of the patient over time.

Images