Providing normalized medical images

Abstract

A normalized medical image 110' representing a region of interest 120 within a subject. 1..n The method provides a warped region of interest 120' for comparison with the region of interest 120 in the atlas image 130. 1..n A normalized medical image 110' having 1.. n The method also includes warping the medical image 110' to the 2D atlas image 130 to provide the normalized medical image 110' 1..n and / or outputting the normalized medical image 110' 1..n The warped region of interest 120' between 1..n The method includes outputting the magnitude of the change in

Description

[Technical Field]

[0001] This disclosure relates to the provision of normalized medical images representing regions of interest within a subject. Computer implementation methods, computer program products, and systems are disclosed. [Background technology]

[0002] Clinical examinations often involve acquiring medical images of the patient. These images are acquired at different points in time; that is, they are acquired as a time-series image. The images are used to identify changes in the region of interest over time and thereby assess the progression of the patient's condition. However, differences in image acquisition techniques can hinder clinical examinations. For example, differences in the patient's posture, differences in the field of view of the medical imaging system, and differences in the amount of ionizing radiation used to acquire the images all exacerbate the difficulty in identifying changes in the region of interest over time. As a result, such differences can reduce the diagnostic value of the images and potentially lead to misdiagnosis of the patient's condition.

[0003] In some clinical settings, it is possible to position the patient in an optimal location relative to the imaging system. In such settings, the variability between images resulting from differences in image acquisition methods is relatively small. However, in other clinical settings, it may be impractical or impossible to position the patient in an optimal location. For example, in an intensive care setting, it is impractical to acquire images of the patient in a desired posture or with the patient in a desired position relative to the medical imaging system. [Overview of the project] [Problems that the invention aims to solve]

[0004] For example, taking diagnostic chest X-rays in an intensive care setting to assess the progression of lung condition is a challenging task for radiographers. The immobility of typically bedridden patients, the limited space for operating mobile X-ray imaging systems, and the bulky nature of such systems are just some of the confounding factors that pose challenges for radiographers. The resulting X-ray images are often of lower quality due to suboptimal field of view, suboptimal viewpoint, limitations on X-ray dose, and limitations on patient posture, such as the patient's inspiratory state, arm position, and incomplete upright position. Therefore, X-ray images taken in an intensive care setting may be of significantly lower quality than those taken during a standard chest X-ray examination. This makes interpreting the X-ray images difficult. In particular, since these confounding factors are not constant over time and vary from image to image, it is difficult to assess long-term changes in the patient's condition. As a result, radiologists interpreting such images may unconsciously compensate for these differences, potentially leading to misdiagnosis of lung condition.

[0005] Therefore, improvements are needed to make it easier to identify changes in medical images over time.

[0006] The document US2007 / 003117A1 discloses a technique for comparative image analysis and / or change detection using a computer-aided detection and / or diagnostic "CAD" algorithm. This technique includes registering two or more images, comparing the images to each other to generate a change map, and detecting anomalies in the images based on the change map. [Means for solving the problem]

[0007] According to one aspect of this disclosure, a computer-aided method is provided for providing normalized medical images representing regions of interest within a subject. This method is The steps include receiving image data including a time-series 2D medical image containing a region of interest within the subject, The steps include receiving an atlas image representing the region of interest, For each medical image in a time series, in order to provide a normalized medical image with a warped region of interest for comparison with the region of interest in the atlas image, the medical image is warped into the atlas image. This method includes the steps of outputting a normalized medical image and / or outputting the magnitude of the change in warped regions of interest between the normalized medical images.

[0008] In the method described above, time-series medical images are warped into an atlas image, resulting in a greater similarity in the shape of each normalized medical image. This facilitates reliable comparisons between regions of interest within images, enabling a more reliable assessment of long-term changes in the subject.

[0009] Further aspects, features, and advantages of this disclosure will become apparent from the following description of embodiments made with reference to the accompanying drawings. [Brief explanation of the drawing]

[0010] [Figure 1] Figure 1 is a flowchart illustrating an embodiment of a method for providing a normalized medical image representing a region of interest within a subject according to several aspects of the present disclosure. [Figure 2] Figure 2 is a schematic diagram showing an embodiment of System 200 for providing normalized medical images representing regions of interest within a subject according to several aspects of the present disclosure. [Figure 3] Figure 3 shows examples of lung field atlas images 130 according to several embodiments of the present disclosure. [Figure 4] Figure 4 shows examples of thoracic atlas images 130 according to several embodiments of the present disclosure. [Figure 5] Figure 5 shows an example of warping two time-series 2D medical images 1101 and 1102 into an atlas image to provide their respective normalized medical images 110'1 and 110'2, according to some aspects of the present disclosure. [Figure 6]Figure 6 shows an example of applying a feature detector to a medical image 1101 to identify multiple landmarks in the medical image, according to some aspects of this disclosure. [Figure 7] Figure 7 shows an example of warping two time-series 2D medical images 1101 and 1102 into an atlas image and adjusting the intensity of the medical images, according to some aspects of the present disclosure. [Figure 8] Figure 8 shows examples of warping two time-series 2D medical images 1101 and 1102 into an atlas image, adjusting the intensity of the medical images, and suppressing image features outside the lung region of interest, according to some aspects of the present disclosure. [Figure 9] Figure 9 shows examples of time-series 2D medical images 1101..n (top), a time sequence of the corresponding normalized images 110'1..n (middle) in which the time-series 2D medical images 1101..n are warped into an atlas image and adjusted to provide adjusted intensity values, and a time sequence of the corresponding normalized images (bottom) in which image features outside the lung region of interest are suppressed. [Modes for carrying out the invention]

[0011] Examples of the present disclosure are provided with reference to the following description and figures. For illustrative purposes, many specific details of several examples are provided below. References to “examples,” “implementations,” or similar terms herein mean that the features, structures, or characteristics described in relation to an example are included in at least one of those examples. Furthermore, it should be understood that features described in relation to one example may also be used in another example, and that not all features necessarily overlap in each example for the sake of brevity. For example, features described in relation to a computer implementation may be implemented in corresponding ways in computer program products and systems.

[0012] The following description refers to computer-aided methods that provide normalized medical images representing regions of interest within a subject. In some embodiments, the region of interest is referred to as the subject's lungs. However, this region of interest is merely an example, and it should be understood that the methods disclosed herein may be used alternatively to provide normalized medical images representing other regions of interest in the subject.

[0013] Furthermore, the computer implementation disclosed herein may be provided as a non-temporary computer-readable storage medium storing computer-readable instructions. When the computer-readable instructions are executed by at least one processor, they cause that at least one processor to execute the method described above. In other words, the computer implementation may be implemented in a computer program product. The computer program product can be provided by dedicated hardware, or dedicated hardware capable of running the software in conjunction with appropriate software. When provided by a processor, the functionality of the method features may be provided by a single dedicated processor, a single shared processor, or multiple individual processors (some of which may be shared). For example, one or more functionality of the method features may be provided by a processor shared within a networked processing architecture such as a client / server architecture, a peer-to-peer architecture, the internet, or a cloud.

[0014] The explicit use of the terms “processor” or “controller” should not be interpreted as referring only to hardware capable of executing software, but implicitly includes, but is not limited to, digital signal processor (DSP) hardware, read-only memory (ROM), random access memory (RAM), and non-volatile storage devices for storing software. Furthermore, embodiments of this disclosure may take the form of computer program products accessible from computer-usable storage media or computer-readable storage media, which provide program code for use by or in connection with a computer or any instruction execution system. For the purposes of this description, computer-usable storage media or computer-readable storage media may be any device including storage, communication, propagation, or transport of programs for use by or in connection with an instruction execution system, apparatus, or device. Media may be electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or propagation media. Examples of computer-readable media include semiconductor or solid-state memory, magnetic tape, removable computer disks, random access memory (RAM), read-only memory (ROM), rigid magnetic disks, and optical disks. Current examples of optical discs include Compact Disc - Read-Only Memory (CD-ROM), Compact Disc - Read / Write (CD-R / W), Blu-Ray®, and DVD.

[0015] As mentioned above, improvements are needed to make it easier to identify changes in medical images over time.

[0016] FIG. 1 is a flowchart showing an example of a method for providing a normalized medical image representing a region of interest within a subject, according to some aspects of the present disclosure. FIG. 2 is a schematic diagram showing an example of a system 200 for providing a normalized medical image representing a region of interest within a subject, according to some aspects of the present disclosure. The operations described in connection with the method shown in FIG. 1 can also be performed in the system 200 shown in FIG. 2, and vice versa. Referring to FIG. 1, a normalized medical image 110' representing a region of interest 120 within a subject 1..n , , 1..n , , 1..n , 1.. n , , 1..n , , 1..n , , 1..n ,

[0018] , , 1..n ,

[0017] , The computer-implemented method for providing is Receiving image data including a time series of 2D medical images 110 including a region of interest 120 within a subject 1..n at (S110); Receiving an atlas image 130 representing the region of interest 120 at (S120); For each medical image 110 in the time series 1..n warping the medical image to the atlas image 130 to provide a normalized medical image 110' having a warped region of interest 120' 1..n within the atlas image 130 for comparison with the region of interest 120 at (S130); 1.. n and further including outputting the normalized medical image 110' at (S140a) and / or outputting the magnitude of change of the warped region of interest 120' 1..n between the normalized medical images 110' 1..n at (S140b). 1..n In the above method, since the time series of medical images are warped to the atlas image, the shapes of the regions of interest will be more similar in each of the normalized medical images. This facilitates a more reliable comparison between the regions of interest in the images and enables a more reliable assessment of the long-term changes in the subject's condition.

[0017] Referring to FIG. 1, at operation S110, image data is received. The image data includes a time series of 2D medical images 110 including a region of interest 120 within a subject

[0018] 1..n ​This includes the following. The received 2D medical images form a time series in the sense that the images were taken at different points in time. The images may be taken regularly, i.e., at regular intervals, or intermittently, i.e., at irregular intervals. Time-series 2D medical images are taken over a certain period of time, such as several minutes, several hours, several days, several weeks, or longer. Time-series 2D medical images represent a so-called long-term study of the subject. Generally, the region of interest represented in the images is any anatomical region. For example, the region of interest may be the lungs, heart, liver, kidneys, etc. For example, time-series 2D medical image 110 received in operation S110 1..n This represents the subject's chest. For example, a time-series 2D medical image represents daily images of the chest taken as part of a long-term study of the subject to assess the progression of COVID-19 in the subject's lungs.

[0019] Time-series 2D medical image 110 received in operation S110 1..n These are generally generated by any 2D medical imaging system. 2D medical images include X-ray images and ultrasound images. 2D X-ray images are generally generated by any projection X-ray imaging system. 2D medical image 110 1..n Examples of current X-ray projection imaging systems capable of generating 2D ultrasound images include the MobileDiagnost M50 mobile X-ray imaging system, the DigitalDiagnost C90 ceiling-mounted X-ray imaging system, and the Azurion7, which includes a source-detector configuration mounted on a C-arm, all of which are sold by Philips Healthcare, located in Best, Netherlands. 2D ultrasound images can be generated by any 2D ultrasound imaging system.

[0020] For example, a 2D medical image 110 received in operation S110 1..nThis includes X-ray images generated by the projection X-ray imaging system 220 shown in Figure 2. The projection X-ray imaging system shown in Figure 2 is a so-called mobile X-ray imaging system, commonly used in intensive care settings to acquire images of patients whose movement is restricted. The projection X-ray imaging system 220 shown in Figure 2 includes an X-ray source 230 and a wirelessly connected X-ray detector 240. The versatility of this type of X-ray imaging system facilitates imaging of patients in complex settings where patient movement is restricted, such as the bedridden setting shown in Figure 2. However, given that patient movement is restricted (the patient cannot position themselves in an optimal position or a repeatable position for each imaging session), it is difficult to assess the patient's condition from images acquired in such settings. An example of the variability in pose and image quality of images acquired from the bedridden setting shown in Figure 2 is shown on the left side of Figure 5. Figure 5 shows one embodiment of warping two time-series 2D medical images 1101 and 1102 into an atlas image to provide their respective normalized medical images 110'1 and 110'2, according to several aspects of this disclosure. The two 2D X-ray images 1101 and 1102 on the left side of Figure 5 represent the chest of a subject at two different points in time. As can be seen, identifying changes in the state of the subject's lungs in these images is complicated by differences in the subject's posture and the viewpoint of the X-ray imaging system relative to the subject. Note that the setup shown in Figure 2 is merely an example, and the principles disclosed herein are not limited to this type of imaging system or setup.

[0021] Returning to Figure 1, generally, the time-series 2D medical image 110 received in operation S110 1..n It can be received via any data communication method, such as wired, optical, and wireless communication. As some examples, when using wired or optical communication, communication takes place via signals transmitted through electrical or optical cables, and when using wireless communication, communication is via, for example, RF signals or optical signals. Time-series 2D medical images 1101..n The data is received by one or more processors, such as the one or more processors 210 shown in Figure 2. The one or more processors 210 receive the 2D medical image 110 from a 2D medical imaging system such as the projection X-ray imaging system 220 shown in Figure 2, from a 2D ultrasound imaging system, or from another source such as a computer-readable storage medium, the internet, or the cloud. 1..n It can receive.

[0022] Continuing to refer to Figure 1, in operation S120, an atlas image 130 representing the region of interest 120 is received. Figure 3 shows examples of lung field atlas images 130 according to some aspects of the present disclosure. Figure 4 shows examples of thoracic atlas images 130 according to some aspects of the present disclosure. The atlas images 130 shown in Figures 3 and 4 indicate, by their intensity values, the probability that a structure belongs to the lung boundary and the probability that a structure belongs to the ribs, respectively. The atlas images 130 are received by, for example, one or more processors 210 as shown in Figure 2. The atlas images can be received from an atlas image database. The atlas images may be selected from the database based on the similarity between the subject and a reference subject represented in the atlas image. For example, the atlas images are selected based on the similarity of age, sex, and size between the subject and the reference subject. The atlas images serve as reference images for the region of interest. The atlas images represent preferred viewpoints of the region of interest within the reference subject. Atlas images are acquired while the reference subject maintains a preferred posture. For example, if the region of interest is the lungs, a lung atlas image is a so-called dorsal-ventral (PA) view of the chest, representing the lungs, bonythoracic cavity, mediastinum, and major vessels. Such atlas images are acquired with the subject standing upright, facing an upright X-ray detector, in a specified inspiratory state, with the top of the X-ray detector at a specified distance above the shoulder joint, the chin raised so as to be outside the image field, and the shoulders rotated forward to move the scapula laterally away from the lung field, under specified operating settings of the X-ray imaging system (e.g., X-ray kVp energy, exposure time). More generally, atlas images 130 are obtained from time-series 2D medical images 110. 1..nThis represents a preferred viewpoint in the region of interest for evaluating the condition of a patient. Using such atlas images facilitates reliable and clinically meaningful comparisons between regions of interest in 2D medical images. For example, it may be used to determine clinically meaningful changes in a patient's lung volume, as these changes are observed from a clinically recognized viewpoint.

[0023] In one embodiment, the atlas image 130 is a time-series 2D medical image 110. 1..n This represents a preferred viewpoint in the area of ​​interest for evaluating the condition in the patient. The method described with reference to Figure 1 includes the operation of selecting an atlas image 130 from a database of atlas images based on the condition. In this embodiment, the atlas image 130 may be selected from the database of atlas images based on the similarity between the subject and a reference subject represented in the atlas image, as described in the above embodiment.

[0024] In another embodiment, the atlas image 130 is a time-series 2D medical image 110. 1..n This represents a preferred viewpoint of the region of interest for evaluating the condition in the region of interest. The method described with reference to Figure 1 includes the operation of selecting an atlas image 130 from a database of atlas images based on the type of imaging examination performed on the region of interest. In this embodiment, if the region of interest is the lungs, for example, the lung atlas image is a so-called dorsal-ventral (PA) view of the chest, representing the lungs, bonythoracic cavity, mediastinum, and great vessels. Such an atlas image is called a “PA” chest image type imaging examination. The database stores multiple atlas images, each representing a different preferred viewpoint of a different type of imaging examination. In this embodiment, the atlas image is selected from the atlas images in the database based on the type of imaging examination. In this embodiment, the atlas image 130 may also be selected from the database of atlas images based on the similarity between the subject and a reference subject represented in the atlas image, as described in the above embodiment.

[0025] Returning to Figure 1, in operation S130, the time-series medical image 110 received in operation S110 1..n The data is warped to the atlas image. The warp operation is performed using various known transformations. One example of a suitable transformation is the affine transformation. Other examples of suitable transformations include B-splines, thin-plate splines, and radial basis functions. The warp operation S130 warps multiple corresponding landmarks 140 that are represented in both the medical image and the atlas image 130. 1..k It is performed based on the mapping between them. Warp operation S130 can be performed using a neural network. When warp is performed by a neural network, the neural network may or may not explicitly use such landmarks. Landmark 140 1..k These are provided by anatomical features or reference markers. In some embodiments, anatomical features such as bones (e.g., ribs, scapula, etc.) or organ contours (e.g., lung contours, diaphragm, cardiac silhouette, etc.) serve as anatomical landmarks. Such features are identifiable in 2D medical images due to their radiofrequency attenuation. Reference markers formed from radiofrequency attenuating materials are also identifiable in medical images and may also function as landmarks. Reference landmarks can be placed at known reference locations, either superficially or internally. In the latter case, reference markers may be implanted for use as surgical guides, etc. Implantable devices such as pacemakers may also function as reference markers. Reference markers may also be provided by interventional devices inserted into the body.

[0026] 110 2D medical images in time series 1..n 140 landmarks within the area 1..k The corresponding landmarks within the atlas image can be identified using various techniques. In one embodiment, a feature detector is used to identify a time-series 2D medical image 110 1..nIdentify the anatomical landmarks within the tissue. The identified landmarks are mapped to the corresponding anatomical landmarks labeled in the atlas image. In this embodiment, atlas image 130 contains multiple labeled anatomical landmarks. The method described with reference to Figure 1 is also: To identify multiple landmarks in medical images corresponding to labeled anatomical landmarks, each medical image in time series 110 1..n This includes applying a feature detector to it. The mapping is determined using the identified corresponding landmarks.

[0027] This embodiment shows a time-series medical image 110 1..n The operation of applying a feature detector is performed using edge detectors, model-based segmentation, or neural networks, etc. For example, Figure 6 shows one embodiment of the results of applying a feature detector to a medical image 1101 to identify multiple landmarks in the medical image, according to some aspects of this disclosure. The 2D medical image 1101 shown in Figure 6 represents the chest and includes bony regions such as ribs and vertebrae, as well as the outlines of the cardiac silhouette and lungs. In this embodiment, the region of interest is the lungs, and the feature detector identifies the outlines of the right lung, the right side of the cardiac silhouette, the right side of the diaphragm, the left lung, the left side of the cardiac silhouette, and the left side of the diaphragm as landmarks 140 in the 2D medical image 1101, respectively. 1..6 These landmarks have been identified. These landmarks correspond to landmarks in atlas images, such as the lung field atlas image shown in Figure 3. Landmarks identified in 2D medical image 1101 140 1..6 These are mapped to corresponding landmarks in the atlas image, and in operation S130, the medical image 1101 is warped to the atlas image. As shown in Figure 6, one or more of the landmarks identified and used in the mapping operation represent a region of interest, which in this embodiment is the lung.

[0028] The result of warp operation S130 is a time-series medical image 110 1..nFor each of these, warp region of interest 120' for comparison with region of interest 120 of atlas image 130. 1..n Normalized medical image 110' 1..n The objective is to provide normalized medical images 110'. 1..n This includes the warped region of interest 120' within each normalized medical image 110'1..n. 1..n This means that the difference in shape between the region of interest 120 in the atlas image 130 and the warped region of interest 120' for comparison with the region of interest 120 in the atlas image 130 is reduced. 1..n There is a warp operation. The warped regions of interest 120'1 and 120'2 are suitable for comparison with region of interest 120 in the atlas image 130, and as a result they can be compared with each other. Therefore, the warp operation S130 reduces the difference in shape between regions of interest 120'1 and 120'2, and makes it possible to more reliably identify the temporal changes of the region of interest resulting from changes in the subject's condition. In other words, the warp operation S130 normalizes the medical image 110' 1..n Warped region of interest 120' 1..n These will have similar shapes. The image on the right in Figure 5 shows the normalized medical image 110' provided in this way. 1..n An example is shown. As can be seen, the lung regions of interest 120'1 and 120'2 in these images have a more similar shape in the normalized medical images 110'1 and 110'2, which allows for a more reliable identification of changes in the region of interest over time due to the progression of the lung condition.

[0029] Returning to Figure 1, in operations S140a and S140b, the method is to normalize the medical image 110' 1..n Operation S140a to output the normalized medical image 110'. 1..n Warped region of interest between 120' 1..n Operation S140b is performed to output the magnitude of the change. The normalized medical image 110' in operation S140a is then output. 1..n The output is a normalized medical image 110' 1..nThis can be done in various ways, such as displaying, printing, and saving. The image can be displayed on a display device such as the monitor 250 shown in Figure 5. For example, a normalized medical image 110' 1..n These may be displayed side-by-side as shown in Figure 5, as a temporal sequence or overlay image, or in other ways.

[0030] The output of the magnitude of the change in the warped region of interest in operation S140b can also be displayed, printed, and saved in various ways. Generally, this change is normalized in the medical image 110'. 1..n This represents a change in intensity or shape of the region of interest between any two of the following images. The change is determined, for example, between consecutive images. In some embodiments, statistical analysis is applied to the warped region of interest to determine the magnitude of the change. In some embodiments, if the region of interest is the lung, the magnitude of the change can quantify the volume of pneumothorax (i.e., the volume of the "collapsed lung"), the amount of pleural effusion (i.e., the amount of "fluid around the lung"), the amount of pulmonary edema (i.e., the amount of "fluid inside the lung"), or a certain stage of pneumonia. Such changes are normalized in the medical image 110'. 1..n The change in intensity values ​​within the lung over time can be determined, or calculated by comparing the image intensity with the normal reference values ​​for the lung in the atlas image. In one embodiment, the change is, for example, in a normalized medical image 110' 1..n The pixels within the lungs are then assigned to either air or water based on their intensity values ​​to represent the lung contour, and the volume of each air and water region can be estimated for each normalized medical image 110'1.

[0031] In some embodiments, outputting the magnitude of change in the warped region of interest in operation S140b includes outputting the magnitude as a numerical value (e.g., as a percentage) or outputting the magnitude of change graphically. In one embodiment, the magnitude of change is output to the normalized medical image 110' 1..nThis can be represented graphically as an overlay on one or more normalized medical images, by showing increases in lung volume between consecutive images in green and decreases in lung volume between consecutive images in red.

[0032] By outputting a normalized medical image in operation S140a and / or outputting the magnitude of change in the warped region of interest in operation S140b, this method facilitates more reliable comparisons between regions of interest within images for the reviewing clinician. This allows the clinician to more accurately assess long-term changes in the patient's condition.

[0033] You can also perform various additional operations by referring to Figure 1 and following the method described above.

[0034] In some embodiments, the method described with reference to Figure 1 is also used. Normalized medical image 110' 1..n To provide adjusted intensity values ​​within the time series of medical images 110, based on the intensity at one or more locations within the atlas image 130. 1..n Based on the intensity at one or more locations within, or based on the intensity at one or more locations within the reference image, each medical image 110 in time series 1..n Adjusting the intensity of, and / or Normalized medical image 110' 1..n To provide adjusted intensity values ​​within each medical image 110 in time series, an image style transfer algorithm is used. 1..n This includes adjusting the intensity.

[0035] Next, normalized medical image 110' 1..n Warped region of interest between 120' 1..n The magnitude of the change can be calculated and output based on the adjusted intensity values ​​within the normalized medical image.

[0036] The inventors have observed that evaluating changes in medical images over time is hindered not only by differences in the shape of regions of interest between images, but also by differences in the intensity scale of the images. More specifically, differences in intensity scale resulting from differences in the methods used to acquire time-series 2D medical images can hinder comparisons. For example, the intensity at any point in a 2D X-ray image depends on the value of the X-ray energy "kVp" used to acquire the image, the irradiation time, and the sensitivity of the X-ray detector. Since time-series images are acquired over a period of time, time-series images may be generated by X-ray imaging systems with different settings, or even by different X-ray imaging systems. In computed tomography images, Hounsfield units can be assigned to each voxel, but in 2D X-ray images, there is no absolute decay scale for the corresponding pixel intensity value. Normalized medical image 110' 1..n To provide adjusted intensity values ​​within the time series of medical images 110, based on the intensity at one or more locations within the atlas image 130. 1..n Based on the intensity at one or more locations within, or based on the intensity at one or more locations within the reference image, each medical image 110 in time series 1..n By adjusting the intensity, the variability between images resulting from differences in the methods used to acquire 2D medical images is further reduced. Based on the intensity at one or more locations within the atlas image 130, each medical image 110 in the time series 1..n The operation to adjust the intensity is the normalized medical image 110', which has the same intensity value as one or more locations in the atlas image 130. 1..n Adjusted intensity values ​​can be provided at one or more corresponding locations within the same. Similarly, time-series medical images 110 1..n Based on the intensity at one or more locations within, each medical image 110 in the time series 1..n The operation to adjust the intensity is performed on the time-series medical image 110 1..n Normalized medical image 110' with the same intensity value as one or more locations within it. 1..nAdjusted intensity values ​​can be provided for one or more corresponding locations within the reference image. Similarly, based on the intensity at one or more locations within the reference image, each medical image 110 in time series can be provided. 1..n The operation to adjust the intensity is a normalized medical image 110' where the intensity value is the same as that at one or more locations in the reference image. 1..n It can provide adjusted intensity values ​​at one or more corresponding locations within the system.

[0037] To normalize intensity values ​​according to this embodiment, techniques such as windowing or histogram normalization can be used. The location within the atlas image 130 can be defined manually or automatically. For example, if the region of interest is the lung, the location can be defined as within the lung field or within the mediastinum, etc. The location can be predefined within the atlas image or defined based on user input provided from a user input device in combination with the displayed image. This embodiment is also described with reference to Figure 7. Figure 7 shows one embodiment of warping two time-series 2D medical images 1101, 1102 into an atlas image and adjusting the intensity of the medical images according to several aspects of this disclosure. The intensity of the image on the right in Figure 7 has been adjusted in the manner described above, making it easier to more reliably identify changes in the region of interest between images.

[0038] Alternatively, an image style transfer algorithm can be used to normalize the medical image 110' 1..n It is also possible to provide adjusted intensity values ​​within. For this purpose, various image style transfer transformations are known. These are based on, for example, Laplace pyramids and (convolutional) neural networks. An example of an image style transfer algorithm is disclosed in International Patent Publication WO2013 / 042018A1. This document discloses a technique for transforming a slave image, which includes generating a color or grayscale transformation based on a master image and a slave image. This transformation is used to optically fit the slave image to the master image. Using this technique, a normalized medical image 110' 1..nThe adjusted intensity values ​​within the time series of medical images 110 have a similar appearance or style. 1..n The intensity can be adjusted. (Time-series medical images 110) 1..n The image is used as a slave image, and the atlas image and time-series medical images 110 1..n The techniques disclosed in this document can be applied by using one or another reference image as a master image.

[0039] In related embodiments, each medical image 110 in a time series is determined based on the intensity at one or more locations within the atlas image 130. 1..n The operation to adjust the intensity is, To calculate the average intensity value within a portion of atlas image 130, Using the average intensity value, each medical image 110 in the time series 1..n This includes normalizing the intensity values ​​within the expression.

[0040] In this embodiment, each medical image 110 in time series 1..n The operation of normalizing the intensity values ​​within each medical image 110 in the time series using the mean intensity value 1..n This is done by mapping the intensity values ​​within the corresponding portion to the average intensity values ​​within the portion of the atlas image 130. In this embodiment, each medical image 110 1..n A portion of the atlas image 130 used to normalize the intensity values ​​within is a time-series 2D medical image 110. 1..n During this period, it is selected as part of an atlas image where the image intensity is expected to be constant over time. For example, if the region of interest is the lungs, the selected portion of the atlas image corresponds to a portion of the spine, because the attenuation of the spine is expected not to change over time. Alternatively, if the region of interest is the lungs, the selected portion corresponds to a time-series 2D medical image 110 1..nThis corresponds to a portion of the lung in the atlas image where disease is not expected. For example, if the subject under examination has pulmonary edema, the collection of fluid over time is expected to result in a change in the image intensity of the lower part of the lung, but the upper part of the lung is expected to remain disease-free and consequently have a stable image intensity. In this case, the upper part of the lung in atlas image 130 corresponds to each medical image 110. 1..n It can be used as part of the atlas image 130, which is used to normalize the intensity values ​​within it.

[0041] In embodiments where the region of interest is the lung, these operations are performed by identifying a portion of the lung in an atlas image. The portion of the lung may be identified automatically or based on user input received via a user input device in conjunction with a display device. The portion of the atlas image may represent a combination of air within the lung and the ribs surrounding it. Next, the average intensity of the pixel values ​​for this portion of the atlas image is calculated. Then, this average is used to obtain a time-series medical image 110 1..n The intensity values ​​within are normalized. Referring to Figure 5, this includes, for example, identifying a corresponding portion of the lung that is free from disease (e.g., towards the upper end of the lung in medical image 1101), calculating the mean intensity values ​​within that corresponding portion free from disease, calculating a scaling factor as the ratio of the mean intensity values ​​within the disease-free region of medical image 1101 to the mean intensity values ​​of the portion of the lung in the atlas image, and using the scaling factor to scale the image intensity values ​​within medical image 1101 to provide adjusted intensity values ​​within the normalized medical image 110'1. These operations are then performed on other medical images 110 in the time series. 1..n This is repeated for each medical image in time series using a portion of the lung in the atlas image. 1..n To normalize the image intensity, medical images are scaled to a common reference intensity, facilitating more reliable comparisons between image intensities in other parts of the image.

[0042] The intensity normalization described above can also be performed using additional portions of the atlas image. For example, it can be used to identify a second region of interest within the mediastinum or bone region of the atlas image and perform intensity normalization. Using two regions of interest with different attenuation values, a time-series medical image 110 1..n Normalizing the intensity within an image facilitates more accurate mapping of intensity values, resulting in more reliable identification of changes in the region of interest.

[0043] In this example, the normalized medical image 110' 1..n It is also possible to calculate and output quality metric values ​​for the adjusted intensity values ​​within. The quality metric values ​​can be, for example, a portion of an atlas image or a medical image 110'. 1..n It can be calculated based on the statistical values ​​of the image intensity values ​​within the corresponding portion, or based on the quality of the mapping between landmarks in operation S130. If the variance of image intensity values ​​in a portion of the atlas image is large, the value of the quality metric will be relatively low, and vice versa. Outputting the value of the quality metric allows the clinician conducting the review to reconfirm the accuracy of the method.

[0044] In one embodiment, the operation S130 to warp the medical image to the atlas image 130 is performed on the medical image 110 1..n This is performed before the operation to adjust the intensity of the medical images. In this embodiment, the adjustment of the intensity of the medical images is performed according to the embodiment described above. Thus, the adjustment is performed based on the intensity at one or more locations in the atlas image 130 of the time-series medical images 110. 1..n This is done based on the intensity at one or more locations within the image, or based on the intensity at one or more locations within the reference image, or using an image style transfer algorithm. As explained above with reference to Figure 5, warping has the effect of making the warped region of interest have a more similar shape. Therefore, warping each medical image into an atlas image before adjusting the intensity provides a more reliable basis for adjusting the intensity values. Thus, the normalized medical image 110' 1..nThe reliability of the adjusted intensity values within is improved. In contrast, if the medical image was normalized prior to warping, the adjusted intensity values do not take into account the adjustment of pixel intensity values provided by the warping. For example, if the normalization is provided based on the lung region of the atlas image, the normalization should be performed on the warped image because the warped lungs have an equivalent shape. As a result, the image intensity values within the lungs more accurately represent the warped shape of the lungs.

[0045] Also, the inventors have observed that following the warping operation S130, some regions outside the region of interest of the normalized medical image 110’ 1..n may be warped into unnatural shapes. This can be confusing to the physician performing the review. In another embodiment, the method described with reference to FIG. 1 also the normalized medical image 110’ 1..n includes suppressing one or more image features outside the region of interest 120’ 1..n within.

[0046] By suppressing one or more image features outside the region of interest 120’ 1..n it is avoided that distracting information is presented to the clinician performing the review. This embodiment will be described with reference to FIG. 8. FIG. 8 shows an example of warping two 2D medical images 1101, 1102 of a time series into an atlas image, adjusting the intensity of the medical image, and suppressing image features outside the lung region of interest, according to some aspects of the present disclosure. In the images on the right side of FIG. 8, the normalized medical image 110’ 1..nPortions outside the lung region of interest are suppressed. In the embodiment of FIG. 8, image features outside the lung region of interest are suppressed by mapping their pixel values to pixel intensity values at the black level. However, these features can also be suppressed by different methods. For example, their pixel values can be mapped to pixel intensity values different from the pixel intensity values at the black level, or gradually adjusted to a predetermined value in order to "feather out" the boundary of the region of interest. This operation of suppressing image features outside the region of interest is performed using an atlas image. For example, in this embodiment, the method described with reference to FIG. 1 is also identifying the region of interest 120 within the atlas image 130, mapping the outer shape of the region of interest 120 from the atlas image 130 to each of the normalized medical images 110’ 1..n and adjusting the image intensity values within the normalized medical images 110’ outside the region of interest 120’ 1..n to suppress one or more image features outside the mapped outer shape. 1..n This includes.

[0047] In this embodiment, various examples of regions of interest are pre-defined in the atlas image and can be used to suppress one or more image features outside the region of interest 120’ 1..n Alternatively, the region of interest can be defined by segmenting the atlas image. In this embodiment, a reliable outer shape is used to suppress one or more image features outside the region of interest 120’ 1..n from the outer shape of the atlas image. In operation S130, since the normalized medical image 110’ 1..n is warped to the atlas image using landmarks within the atlas image, the outer shape is reliable, and thus the shape of the region of interest in the normalized image is similar to the shape of the region of interest in the atlas image.

[0048] ​​​1..n The corresponding normalized image 110' is warped into the atlas image and adjusted so that its intensity provides adjusted intensity values. 1..n Figure 9 shows an example of the temporal sequence (center) and the corresponding normalized temporal sequence (bottom) in which image features outside the lung region of interest are suppressed. Figure 9 shows the aforementioned operation on more images than the two images shown in Figures 5, 7, and 8, and also shows 2D medical images 110 from long-term studies of subjects. 1..n The amount of variation that can be observed and the normalized medical image 110'1 provided by the method described herein ..n Further details will be provided regarding its advantages.

[0049] In another embodiment, the method described with reference to Figure 1 is also, Normalized medical image 110' 1..n Internal area of ​​interest 120' 1..n This includes performing a statistical analysis of the image intensity values ​​within the image. Normalized medical image 110' 1..n Warped region of interest between 120' 1..n Outputting the magnitude of the change (S140b) Normalized medical image 110' 1..n This includes outputting the results of the statistical analysis.

[0050] Normalized medical image 110' 1..n Internal area of ​​interest 120' 1..n Various statistical analysis methods can be applied to the image intensity values ​​within the image. One exemplary implementation of such statistical analysis is normalized medical image 110' 1..n This is based on local patch-based analysis of pixel values ​​within the image. In this example, the normalized medical image 110' 1..n Statistical values ​​are generated for each patch in the image. This includes calculating the mean and standard deviation of the pixel values ​​within the patch. For example, a Gaussian distribution is assumed. Normalized medical image 110' 1..nFor example, when comparing corresponding patches in two images, A and B, the Gaussian statistics of the patch in image A can be used to estimate the likelihood of the measured intensity value in the patch in image B. This likelihood can be thresholded to provide predictions of changes within the patch. This likelihood can also be visualized as an overlay of images A and B. This allows the reviewing clinician to analyze changes detected in the images over time. Consequently, this visualization allows the reviewing clinician to assess whether the detected changes are statistically related or attributable to an imperfect normalization procedure.

[0051] In the related examples, the normalized medical image 110' 1..n The operation to output the results of the statistical analysis involves normalized medical images 110' 1..n This includes displaying a spatial map of the statistical analysis as an overlay on one or more of the other elements. By providing the results of the statistical analysis in this way, clinicians can efficiently analyze the progression of the patient's disease.

[0052] In another embodiment, a computer program product is provided. When the computer program product is executed by one or more processors, it generates a normalized medical image 110' representing a region of interest 120 within a subject on that one or more processors. 1..n This includes instructions that cause a method to be performed that provides a method. Time-series 2D medical images 110 including the region of interest 120 within the subject. 1..n Receiving image data including (S110), Receiving an atlas image 130 representing the region of interest 120 (S120), For each medical image in time series, Warped region of interest 120' for comparison with region of interest 120 in atlas image 130. 1..n Normalized medical image 110' 1.. n To provide this, the medical image is warped to the atlas image 130 (S130), This method includes normalized medical images 110' 1..n Output (S140a) and / or normalized medical image 110' 1..n Warped region of interest between 120' 1..n This further includes outputting the magnitude of the change (S140b).

[0053] In another embodiment, a normalized medical image 110' representing the region of interest 120 within the subject is used. 1..n A system 200 is provided which provides the following. This system includes one or more processors 210, and one or more processors 210 are Time-series 2D medical images 110 including the region of interest 120 within the subject. 1..n Receiving image data including (S110), Receiving an atlas image 130 representing the region of interest 120 (S120), For each medical image in time series, Warped region of interest 120' for comparison with region of interest 120 in atlas image 130. 1..n Normalized medical image 110' 1.. n To provide this, the medical image is warped to the atlas image 130 (S130), One or more processors perform the normalized medical image 110'. 1..n Output (S140a) and / or normalized medical image 110' 1..n Warped region of interest between 120' 1..n Further, output the magnitude of the change (S140b).

[0054] Figure 2 shows an embodiment of system 200. System 200 also includes a 2D medical imaging system (for example, the projection X-ray imaging system 220 shown in Figure 2) for generating image data received in operation S110, and the output normalized medical image 110'. 1..n Display and / or normalized medical image 110' 1..n Warped region of interest between 120' 1..nThe system includes a display device (monitor 250, tablet, etc.) for outputting the magnitude of changes, the results of statistical analysis of image intensity values, etc., a patient bed 260, and one or more user input devices (not shown in Figure 2) (keyboard, mouse, touchscreen, etc.) for receiving user input in relation to operations performed by the system.

[0055] It should be understood that the embodiments described above are illustrative and not limiting to the present disclosure. Other embodiments are also contemplated. For example, embodiments described in relation to computer implementations are also provided in corresponding embodiments by computer program products, computer-readable storage media, or System 200. It should be understood that features described in relation to any one embodiment may be used alone, in combination with other described features, in combination with one or more features of another embodiment, or in combination with other embodiments. Furthermore, equivalents and modifications not described above may be used without departing from the scope of the invention as defined in the appended claims. In the claims, the word “includes” does not exclude other elements or operations, and singular elements do not exclude plural elements. The mere fact that certain features are described in different dependent claims does not mean that combinations of these features cannot be used advantageously. Any reference numerals in the claims should not be construed as limiting the scope.

Claims

1. A computer-aided method for providing normalized medical images representing regions of interest within a subject, wherein the computer-aided method is: The steps include receiving image data including a time-series 2D medical image containing a region of interest within the subject, The steps include receiving a 2D atlas image representing the region of interest, For each 2D medical image in the aforementioned time series, in order to provide a normalized medical image having a warped region of interest for comparison with the region of interest in the 2D atlas image, the steps include: warping the 2D medical image to the 2D atlas image; Includes, The steps of outputting the normalized medical image and / or, A step of outputting the magnitude of the change in the warped region of interest between the normalized medical images, A computer implementation method further including the following.

2. The computer implementation method according to claim 1, wherein the 2D atlas image represents a preferred viewpoint of the region of interest for evaluating the condition in the time-series 2D medical images.

3. The computer-aided method according to claim 1, wherein the warping step is based on mapping between a plurality of corresponding landmarks represented in both the 2D medical image and the 2D atlas image.

4. The aforementioned 2D atlas image includes multiple labeled anatomical landmarks, The aforementioned computer implementation method is: The process further includes applying a feature detector to each of the 2D medical images in the time series in order to identify a plurality of landmarks in the 2D medical images that correspond to the labeled anatomical landmarks, The computer implementation method according to claim 3, wherein the mapping is determined using the identified corresponding landmarks.

5. The computer implementation method according to claim 4, wherein the step of applying the feature detector is performed using an edge detector, model-based segmentation, or a neural network.

6. To provide the adjusted intensity values ​​in the normalized medical images, the steps include adjusting the intensity of each 2D medical image in the time series based on the intensity at one or more locations in the 2D atlas image, based on the intensity at one or more locations in the time series medical images, or based on the intensity at one or more locations in the reference image, and / or The computer implementation method according to claim 1, further comprising the step of adjusting the intensity of each 2D medical image in the time series using an image style transfer algorithm to provide adjusted intensity values ​​in the normalized medical image.

7. The computer implementation method according to claim 6, wherein the step of warping the 2D medical image to the 2D atlas image is performed before the step of adjusting the intensity of the 2D medical image.

8. The process includes the step of calculating the magnitude of the change in the warped region of interest between the normalized medical images based on the adjusted intensity values ​​in the normalized medical images, The computer implementation method according to claim 6, wherein the step of outputting the magnitude of the change in the warped region of interest between the normalized medical images includes the step of outputting the calculated magnitude of the change.

9. The step of adjusting the intensity of each 2D medical image in the time series based on the intensity at one or more locations in the 2D atlas image provides adjusted intensity values ​​at the corresponding one or more locations in the normalized medical image that are the same as the intensity values ​​at the one or more locations in the 2D atlas image. The step of adjusting the intensity of each 2D medical image in the time series based on the intensity at one or more locations in the 2D medical images in the time series provides adjusted intensity values ​​at the corresponding one or more locations in the normalized medical image which are the same as the intensity values ​​at the one or more locations in the 2D medical images in the time series. The computer-aided method according to claim 6, wherein the step of adjusting the intensity of each 2D medical image in the time series based on the intensity at one or more locations in the reference image provides adjusted intensity values ​​at the corresponding one or more locations in the normalized medical image which are the same as the intensity values ​​at the one or more locations in the reference image.

10. The step of adjusting the intensity of each 2D medical image in the time series based on the intensity at one or more locations within the 2D atlas image is: The steps include: calculating the average intensity value within a portion of the aforementioned 2D atlas image; The steps include: normalizing the intensity values ​​in each 2D medical image of the time series using the average intensity value; The computer implementation method according to claim 6, including the method described in claim 6.

11. The computer implementation method according to claim 10, wherein the step of normalizing the intensity values ​​in each 2D medical image of the time series using the average intensity values ​​includes the step of mapping the intensity values ​​in a corresponding portion of each 2D medical image of the time series to the average intensity values ​​in the portion of the 2D atlas image.

12. The computer-aided method according to claim 1, further comprising the step of suppressing one or more image features outside the region of interest within the normalized medical image.

13. The steps include identifying the region of interest within the 2D atlas image, The steps include mapping the outline of the region of interest from the 2D atlas image to each of the normalized medical images, The steps include adjusting the image intensity values ​​in the mapped, out-of-bounds, normalized medical image in order to suppress one or more image features outside the region of interest, The computer implementation method according to claim 12, further comprising:

14. The method further includes the step of performing a statistical analysis of the image intensity values ​​within the region of interest in the normalized medical image, The step of outputting the magnitude of the change in the warped region of interest between the normalized medical images is: The computer implementation method according to claim 1, further comprising the step of outputting the results of the statistical analysis of the normalized medical image.

15. The computer implementation method according to claim 14, wherein the step of outputting the results of the statistical analysis of the normalized medical images includes the step of displaying a spatial map of the statistical analysis as an overlay on one or more of the normalized medical images.

16. The computer implementation method according to claim 1, wherein the aforementioned time-series 2D medical images include X-ray images or ultrasound images.

17. The computer-aided method according to claim 1, wherein the aforementioned time-series 2D medical images include X-ray images generated by a mobile X-ray imaging system.

18. The computer implementation method according to claim 1, wherein the aforementioned time-series 2D medical images represent a long-term study of the subject.

19. A computer program, when executed by one or more processors, includes instructions that cause the one or more processors to perform a method for providing a normalized medical image representing a region of interest within a subject, The aforementioned method, The steps include receiving image data including a time-series 2D medical image containing a region of interest within the subject, The steps include receiving a 2D atlas image representing the region of interest, For each 2D medical image in the aforementioned time series, in order to provide a normalized medical image having a warped region of interest for comparison with the region of interest in the 2D atlas image, the steps include: warping the 2D medical image to the 2D atlas image; Includes, The aforementioned method, The steps of outputting the normalized medical image and / or, The step of outputting the magnitude of the change in the warped region of interest between the normalized medical images is as follows: Furthermore, it includes computer programs.

20. A system that provides normalized medical images representing regions of interest within a subject, comprising one or more processors, the one or more processors Receiving image data including a time-series 2D medical image containing a region of interest within the subject, Receiving a 2D atlas image representing the aforementioned region of interest, For each 2D medical image in the aforementioned time series, in order to provide a normalized medical image having a warped region of interest for comparison with the region of interest in the 2D atlas image, the 2D medical image is warped into the 2D atlas image. Perform The one or more processors described above are: Outputting the normalized medical image, and / or, Output the magnitude of the change in the warped region of interest between the normalized medical images. A system that further performs this task.

Images