Patient preparation for medical imaging

Abstract

A computer-implemented method for preparing an object in medical imaging, comprising: obtaining a series of images comprising a region of interest of at least part of the object, where the series of images comprises at least a first image and at least a subsequent second image (S10); determining a position of at least one landmark from the series of images, where the at least one landmark is anatomically related to a target anatomy (S20); determining a confidence level assigned to the position of the at least one landmark (S30); determining a position of the target anatomy based on the position of the at least one landmark and the confidence level (S40); the position of the target anatomy is provided for preparing the subject in medical imaging (S50).

Description

[0001] This application is a divisional application of patent application 202280004428.6 with the filing date of March 9, 2022 and the invention title of "Patient Preparation for Medical Imaging". Technical Field

[0003] The present invention relates to medical imaging, and in particular to a computer-implemented method for preparing an object in medical imaging, a device for preparing an object in medical imaging, an imaging system, and a computer program unit. Background Art

[0004] Medical imaging is a key technology in modern medicine. The medical imaging workflow requires trained personnel to operate the imaging unit, for example, a magnetic resonance imaging (MRI) system. Preparing the object and / or the imaging unit for medical imaging is time-consuming and crucial for the quality of the medical image. Therefore, the staff faces challenges in many different tasks. Such tasks include, for example, aligning the object with the medical imaging unit, positioning the medical imaging device adjacent to the object, adjusting the imaging unit, and / or documenting the workflow, etc.

[0005] Each step requires personnel attention and time. In addition, each step may lead to potential errors.

[0006] U.S. Patent Application US2018 / 0070904 A1 discloses a motion tracking system that superimposes tracking data on the imaging data of a patient and displays them together. The tracking data is generated by estimating patient motion based on the data.

[0007] U.S. Patent Application US2018 / 116518 A1 discloses providing preparatory data for an MRI process by using a depth map of a patient on a patient support obtained by using a time-of-flight camera. Summary of the Invention

[0008] Therefore, there may be a need for a method for preparing a patient in medical imaging, and in particular, an improved method for preparing a patient in medical imaging. The object of the present invention is solved by the subject matter of the independent claims, and further embodiments are included in the dependent claims.

[0009] According to a first aspect, there is provided a computer-implemented method for preparing an object in medical imaging, comprising: obtaining a series of images of a region of interest including at least a portion of the object, wherein the series of images includes at least a first image and at least a subsequent second image; determining the position of at least one landmark based on the series of images, wherein the at least one landmark is anatomically related to a target anatomical structure; determining a confidence level assigned to the position of the at least one landmark; determining the position of the target anatomical structure based on the position of the at least one landmark and the confidence level; and providing the position of the target anatomical structure for preparing the object in medical imaging.

[0010] The term "object" should be understood broadly in the present context and includes any human and any animal. The term "medical imaging" should be understood broadly in the present context and includes any imaging process configured to image a region of interest in terms of a medical image, which medical image will be further used, for example, for examination. Medical imaging may include CT imaging, MRI imaging, X-ray imaging. In the present context, medical imaging may particularly include MRI. In the present context, a series of images means multiple individual images. Individual images may be obtained at certain time intervals. Certain time intervals are related to the frame rate. In the present context, the frame rate means the number of images obtained within one second. In the present context, the frame rate may be 1, 10, 24, 30, 35 or 60. Images may be obtained by a camera arranged above the object and / or the medical imaging device such that the images obtained include the region of interest. The camera may also be arranged differently, for example, arranged adjacent to the object such that a side view of the region of interest is obtained. The camera may preferably be a digital optical camera or a digital optical video camera. Images may be obtained from one or more cameras. The term "region of interest" should be understood broadly in the present context and includes a part of the object on the surface of the object (e.g., torso, back, limbs, joints, etc.) or a part of the object inside the object (e.g., internal organs). For example, the region of interest may be the patella visible from the surface. In another example, the region of interest may be the liver inside the object and thus not visible from the perspective of the digital optical camera or the digital optical video camera. The region of interest may be blocked or obscured by, for example, a medical cover, imaging equipment (e.g., an MRI coil), medical personnel, the object itself (the leg above the desired patella), human tissue above an organ (e.g., the liver). The term "position" in the present context means the x, y, z position in the image, which indicates the x, y, z position in the imaging unit and / or the imaging system (e.g., MRI or CT). The position may also mean, in the present context, the content or extent of the region of interest in the image. For example, the position may be related to the volume dimension of the liver. The volume dimension in the image may relate to the volume dimension and / or position in the coordinate system of the imaging unit or the imaging system (e.g., MRI, CT or X-ray system). The term "landmark" should be understood broadly in the present context and means a reference point provided by the object itself. Preferably, the landmark may be a physical marker of the object itself, for example, a body part of the object, such as a bone, head, nose or fissure. The physical marker may be concise enough to be detected in the image. In other words, the physical marker can be easily detected if visible in the image. The term "anatomical connection" in the present context means that the movement or displacement of the landmark will affect the position of the target anatomical structure or act on it. For example, if the position of the patella changes, the position of the corresponding calf bone will also change due to their anatomical connection. In other words, the target anatomical structure and the landmark show a kinematic chain.For example, the position of an organ such as the liver as the target anatomical structure depends on the position of adjacent fissures that can serve as landmarks in this example. The term target anatomical structure means the desired anatomical structure whose position must be determined in the current context. Examples of target anatomical structures can be not only bones, joints, organs, tissue regions, blood vessels, but also tumors or irregular objects detected in a previous treatment of the object. The term confidence level means a measure of the certainty regarding the position of the landmark in the current context. In other words, the confidence level is related to the reliability of the position. The confidence level can range from 0 to 1, where 0 represents low reliability and 1 represents high reliability. The confidence level can be estimated based on the entropy or variance of the prediction of the position in at least one image. The term preparation of a medical object in medical imaging should be understood broadly in the current context and includes any task related to the medical imaging process, for example, positioning the object on a table, determining the scan position, placing the imaging equipment on or at the object, adjusting the control of the imaging unit or imaging system based on the determined position, or ensuring compliance with safety guidelines. Information about the position of the target anatomical structure can be sent to the controls of the imaging system or imaging unit. Information about the position of the target anatomical structure can be displayed on a screen to guide medical assistants.

[0011] In other words, the disclosed computer-implemented method for preparing an object in medical imaging is based on the finding that medical personnel have difficulty in precisely and rapidly determining the position of the target anatomical structure of the object during the preparation phase of medical imaging. Knowing the exact position is crucial for adjusting the medical imaging unit or medical imaging system. In addition, knowing the exact position of the target anatomical structure is crucial for preparing the object (e.g., placing imaging equipment on or at the object). However, the target anatomical structure (e.g., the liver) may be blocked by the camera, making it impossible to directly determine the position of the target anatomical structure (e.g., the liver), but rather to indirectly determine the position of the liver based on adjacent landmarks that are anatomically related to the target anatomical structure. For example, three landmarks (i.e., the head, the fissure, and the buttocks) are tracked in order to determine their respective positions and to determine the position of the liver based on their position information. Regarding the positioning of these three landmarks, some reliability issues may arise. For example, medical personnel may block the line of sight between the camera and the buttocks of the object. Detecting such reliability issues is crucial for determining the confidence level of the landmarks. The confidence level of the landmarks blocked in the determined image will be reduced. To solve this reliability problem, a previous image with a high confidence level for the landmark is used. Thus, even in the presence of obstacles in the camera's line of sight, the position of the target anatomical structure can be derived with high accuracy and high reliability. This improves the productivity of the medical imaging process because medical personnel do not have to manually perform the task of determining the position of the target anatomical structure. In addition, since the position of the target anatomical structure is precisely calculated by the disclosed method, the quality of medical imaging is improved. The determination of the position should be understood broadly and in the current context means positioning, particularly the positioning of landmarks and / or the target anatomical structure.

[0012] According to an embodiment, in a case where a confidence level of the at least one landmark in the second image is lower than a predetermined threshold, a position of the target anatomical structure is determined based on the at least one landmark in the first image. The predetermined threshold may be 0.85, preferably 0.9, and particularly preferably 0.95. In other words, if the current image, which is the latest available image, provides a poor confidence level only for the position of at least one landmark, the method does not use the position of the landmark in the current image to determine the position of the target anatomical structure. Instead, the method uses the position of the landmark in the previous image. When the method considers a series of images, it is clear that in the case of several consecutive images having a low confidence level for at least one landmark, the last image having a high confidence level for at least one landmark is used to determine the position of the target anatomical structure. In other words, the first image and the second image are not necessarily obtained directly consecutively, and there may be several images between them. This can be advantageous because for a case where the object is at least partially occluded, the method can correct the landmark and obtain a more appropriate determination of the position of the target anatomical structure.

[0013] According to an embodiment, determining the position of the target anatomical structure may include further determining a shift of the position of the at least one landmark between the first image and the second image. The term shift in the current context means a change in the position of at least one landmark between the first image and the second image. The shift of the position of at least one landmark can serve as a verification possibility for a landmark whose confidence level determined in the second image is close to the threshold. In addition, the shift can result in a more accurate determination of the position of the target anatomical structure because possible ambiguities in the position of the landmark in the second image can be excluded based on the landmark position in the first image. The shift can also be determined by calculating a mean or average of two positions in the first image and the second image. This can be advantageous in order to reduce errors occurring during image acquisition because they may smooth out outliers. In a case where an additional landmark (e.g., a sixth landmark) is occluded, a shift of the position of the additional landmark can be determined using a shift of more than one landmark (e.g., five landmarks), where the determination is based on an average of the shifts of the five landmarks. This can be beneficial for improving the positioning accuracy of the position of the target anatomical structure.

[0014] According to an embodiment, in a case where a confidence level of the at least one landmark in the second image is higher than the predetermined threshold, a position of the target anatomical structure is determined based on: either a shift of the position of the at least one landmark between the first image and the second image, or the at least one landmark in the second image. In other words, based on the determined confidence level of the at least one landmark in the second image, two possibilities for determining the position of the target anatomical structure can be obtained. In the first possibility, the position of the at least one landmark is determined based on the shift of the position of the at least one landmark between the first image and the second image, which can facilitate correcting / adjusting the position of the target anatomical structure from the first image to the second image. The second possibility is to determine the anatomical structure based on the landmark determined in the second image. This can be advantageous in terms of computational efficiency because only one image is required. The method may further include determining a confidence level of the target anatomical structure for the position of the target anatomical structure. The above possibilities of determining the position of the landmark in a case where the confidence level is higher than the predetermined threshold may also take into account the confidence level of the position of the target anatomical structure. Both possibilities can be performed, and as a result, the position of the target anatomical structure with the highest confidence level can be selected. This can facilitate improving the positioning accuracy of the position of the target anatomical structure.

[0015] In an embodiment, a weighting factor for the shift of the position of the at least one landmark can be determined based on the determined confidence level, and wherein the weighting factor is used to determine the position of the target anatomical structure. Determining the shift of the position of the landmark may introduce uncertainty. Therefore, it is useful to consider this uncertainty through a weighting factor. One possibility is to use the confidence level of the position of the at least one landmark in the corresponding image (e.g., according to the second image by the shift from the first image to the second image). This can be advantageous in improving the accuracy of determining the position of the target anatomical structure.

[0016] In an embodiment, the at least one landmark is selected based on the presence of the landmark in the series of images. The term presence in the current context means that the landmark is visible to, for example, a camera implemented above the region of interest. The presence may be adversely affected by obstacles (e.g., medical support personnel). The position of the target anatomical structure can be described by a quantity x from 1 to n landmarks, where n is a finite number. The method can select only the visible landmarks for the target anatomical structure from the quantity x at the start of the preparation phase in order to determine the position of the target anatomical structure. During the preparation phase, the landmarks can be selected multiple times.

[0017] In an embodiment, the target anatomical structure may be obscured by an obstacle. The term obstacle should be understood broadly in the present context and includes any element configured to obscure the target anatomical structure. The term may include parts of the object itself, medical personnel, medical equipment, etc. In an embodiment, at least one landmark may be obscured by an obstacle.

[0018] In an embodiment, the position of the target anatomical structure is determined based on a plurality of landmarks (i.e., two or more landmarks). This can be advantageous in cases where one or more landmarks are obscured by an obstacle while one or more additional landmarks remain visible to the camera. This can be further advantageous because more landmarks can lead to a more precise determination of the position of the target anatomical structure. It should be noted that in the case of several landmarks, a corresponding confidence level can also be determined for each landmark, and this confidence level can be considered for determining the target anatomical structure. Additionally, the above possibilities including displacement, weighting factors can also be applied to the case of multiple landmarks.

[0019] In an embodiment, determining the position of the at least one landmark and determining the position of the target anatomical structure can be based on an image analysis algorithm. The analysis algorithm can include, for example, a segmentation algorithm, an artificial neural network, a deep convolutional neural network, an image acquisition model, and / or a model for determining a confidence level. In an embodiment, the analysis algorithm can use an image acquisition model that describes the process by which the acquired images in this series of images have been obtained. The image acquisition model can describe irregularities that occur during the imaging process (e.g., alignment errors, rotation errors). As a result, the image acquisition model can generate a number of output images indicating the irregularities based on an input image (e.g., the first image in a series of images). Based on these number of output images, a deep convolutional neural network can determine the landmarks in each of the number of output images. Then, the uncertainty can be determined by calculating the entropy between the input image and the number of output images. A description of an image analysis algorithm that can be used in the embodiments described in this disclosure is described in the article “Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks” by Guotai Wang, Wenqi Li, Michael Aertsen, Jan Deprest, Sebastian Ourselin, Tom Vercauteren (published in Neurocomputing, 2019, https: / / doi.org / 10.1016 / j.neucom.2019.01.103). The contents of this article are incorporated herein by reference in their entirety.

[0020] In an embodiment, the control signal for controlling the imaging unit can be derived based on the determined position of the target anatomical structure. The position of the target anatomical structure can be part of the control information for an imaging unit such as an MRI. This can be advantageous in terms of workflow efficiency because medical support personnel do not have to perform calculations or related tasks. The method can further automate part of the workflow of medical support personnel.

[0021] In an embodiment, guidance data for preparing the object in medical imaging is derived based on the position of the target anatomical structure, and wherein the guidance data includes a target alignment of the object relative to the imaging unit. The term guidance data in the present context means any data configured to guide a medical assistant in preparing the object and / or the medical imaging unit for imaging. The guidance data may include a visual representation of a target alignment of medical equipment (e.g., a coil) that must be aligned with the object.

[0022] In an embodiment, the determination of the position of the target anatomical structure is based on one or more degrees of freedom of one or more joints of the object. The object may include joints, bones, tissues, organs, etc., which cannot move completely freely and independently of each other. The method may use a kinematic model that includes constraints on the movement of the various parts of the object. In particular, the method may consider the degrees of freedom of the joints. For example, the knee joint can only move within a range of 180°, otherwise the knee joint will break. As a result, this can improve the accuracy of determining the position of the target anatomy by excluding unrealistic results (e.g., a knee joint angle of 230°).

[0023] Another aspect of the present disclosure relates to an apparatus for preparing an object in medical imaging, comprising: an acquisition unit configured to acquire a series of images including a region of interest of at least a portion of the object, wherein the series of images includes at least a first image and a subsequent second image; a first determination unit configured to determine the position of at least one landmark based on the series of images, wherein the at least one landmark is anatomically related to a target anatomical structure; a second determination unit configured to determine a confidence level assigned to the position of the at least one landmark; a third determination unit configured to determine the position of the target anatomical structure based on the position of the at least one landmark and the confidence level; and a provision unit configured to provide the position of the target anatomical structure for preparing the object in medical imaging.

[0024] The acquisition unit and / or the determination unit and / or the provision unit may be distributed across different hardware units or combined in a single hardware. The first determination unit, the second determination unit, and the third determination unit may be one hardware unit. Additionally, the acquisition unit and / or the determination unit and / or the provision unit may be virtual units (i.e., software units).

[0025] Optionally, the apparatus may be configured to perform the method according to the first aspect.

[0026] Another aspect of the present disclosure relates to an imaging system, comprising: the above-mentioned apparatus; an imaging unit; and an imaging control unit. The imaging unit may be a CT, MRI, and X-ray imaging unit.

[0027] Another aspect of the present disclosure relates to a computer program unit which, when run by a processor, is configured to: perform the above-described method, and / or control the above-described device, and / or control the above-described system.

[0028] The computer program unit may be stored on a computer unit, which may also be part of an embodiment. The computing unit may be configured to perform the steps of the above-described method or cause the steps of the above-described method to be performed. In addition, the computing unit may be configured to operate the components of the above-described device. The computing unit can be configured to automatically operate and / or execute the commands of a user. The computer program may be loaded into the working memory of a data processor. The data processor may thus be equipped to perform the method according to one of the foregoing embodiments. This exemplary embodiment of the invention covers both computer programs that use the invention from the start and computer programs that turn existing programs into programs using the invention by means of an update. Additionally, the computer program unit may be able to provide all the necessary steps to complete the process of the exemplary embodiment of the above-described method. According to another exemplary embodiment of the invention, a computer-readable medium is proposed, for example, a CD-ROM, a USB stick, etc., on which a computer program unit is stored as described in the previous section. The computer program may be stored and / or distributed on a suitable medium, for example, an optical storage medium or a solid-state medium provided together with other hardware or as part of other hardware, but may also be distributed in other forms, for example, via the Internet or other wired or wireless telecommunication systems. However, the computer program may also be presented on a network such as the World Wide Web and can be downloaded from such a network into the working memory of a data processor. According to another exemplary embodiment of the invention, a medium for making a computer program unit available for download is provided, the computer program unit being arranged to perform the method according to one of the foregoing embodiments of the invention.

[0029] Note that, regardless of the aspects involved, the above embodiments can be combined with each other. Thus, the method can be combined with the structural features of the devices and / or systems of other aspects, and likewise, the devices and systems can be combined with the features of each other and also with the features described above with respect to the method.

[0030] These and other aspects of the invention will become apparent and be elucidated with reference to the embodiments described hereinafter. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Exemplary embodiments of the invention will be described in the following drawings.

[0032] Figure 1 Shows an exemplary image of a series of images according to a first embodiment of the present disclosure;

[0033] Figure 2 Shows a graph of confidence levels over a series of images;

[0034] Figure 3 Is a schematic diagram of a method according to a first aspect of the present disclosure; and

[0035] Figure 4 Is a schematic diagram of a device according to a further aspect of the present disclosure.

[0036] List of reference numerals:

[0037] 1, 2, 3, 4 Images

[0038] 5 Patient support

[0039] 6 Imaging unit

[0040] 7, 10 Patients

[0041] 8 Left foot

[0042] 9 Left knee

[0043] 11 Medical auxiliary staff

[0044] 12 Coil

[0045] 13 Right ankle of the foot, landmark

[0046] 14 Right knee, target anatomical structure

[0047] 15 Right hip, landmark

[0048] 16 Bore

[0049] 17 Cable

[0050] 20 Graph

[0051] 21 Vertical axis

[0052] 22 Horizontal axis

[0053] 23, 24, 25, 26 Vertical dotted lines

[0054] 40 Device

[0055] 41 Acquisition unit

[0056] 42 First determination unit

[0057] 43 Second determination unit

[0058] 44 Second determination unit

[0059] 45 Providing Unit Detailed Implementation Manner

[0060] Figure 1 An exemplary image of a series of images according to a first embodiment of the present disclosure is shown. Four images 1, 2, 3, and 4 all show the same scene, that is, the patient preparation at different time stages for the medical imaging of the patient's right knee. The image is obtained from a digital optical camera implemented above the scene (see, for example Figure 4 , which is referred to as the acquisition unit 41). In particular, image 1 shows patient 7 who has just boarded the patient support 5. The patient support 5 is part of the magnetic resonance imaging unit 6. The patient support 5 is used for the preparation of the patient outside the bore 7 of the MRI unit 6. Once patient 7 is ready, the patient support 5 moves with the patient lying on it into the bore of the MRI unit 6. After that, the medical imaging process for obtaining medical MRI images begins. As can be seen in image 1, not all parts of patient 7 are visible to the camera. For example, the patient's left foot 8 is blocked by the patient's left knee 9. Figure 2 Another stage of this preparation is shown. Patient 10 is already lying on the patient support. The medical assistant 11 appears in image 2, holding the coil 12 to be placed on patient 10. As can be seen in image 2, at this stage of the preparation, no part of patient 10 is blocked. There is no obstacle between patient 10 and the camera implemented above. Therefore, the field of view of the camera projected onto potential landmarks (e.g., the right ankle 13, the right hip 15) and the target anatomical structure 14 (i.e., the knee 14) is free. The landmarks (the right ankle 13 and the right hip 15) are anatomically connected to the target anatomical structure 15 (the right knee 15). Therefore, any movement of the landmarks will affect the position of the target anatomical structure. For example, if the right ankle 13 moves horizontally to the left, the right knee 15 must also move horizontally to the left. In Figure 3 , the medical assistant has almost completed the placement process of placing the coil on the patient's right knee. Therefore, the right knee is blocked by the coil, so there is no free field of view of the camera to the right knee. The right hip is clearly visible, but the right ankle is also slightly blocked by the cable 17 of the coil. In Figure 4 , the patient support moves automatically with the patient on it into the bore 16 of the MRI. Therefore, the landmarks and the target anatomical structure are no longer visible to the camera either.

[0061] Figure 2 A chart of the confidence level on a series of images is shown. Figure 2 Corresponding to Figure 1 , where Figure 1 relates to four images, and Figure 2 relates to 300 images, and where Figure 1The four images are Figure 2 part of 300 images of Figure 2 . The confidence level is plotted on the vertical axis 21 of the graph. The confidence level represents the reliability of the determined position. In this example, the confidence level includes values from 0 to 1, where decimal values similar to 0.81 are possible. How to determine the confidence level will be explained in Figure 3 . The numbers of the corresponding pictures are shown on the horizontal axis 22 of the graph 21. For each image, the confidence levels of the right knee or the right ankle of the foot or the right hip are plotted as points in the graph respectively. The vertical dashed lines 23, 24, 25, and 26 with the numbers 1 to 4 in parentheses refer to the respective points in the graph related to the Figure 1 images 1 to 4 in Figure 1 . It can be seen that due to the lack of landmarks or target anatomical structures of the images, the confidence levels in the first 50 images have a value of approximately 0. In the range between 50 and 100 images, the confidence level rises to approximately 0.9. This is because from the first stage of the preparation (where the patient climbs onto the patient support) to the second stage (where the patient lies on the patient support), the visibility of the landmarks and target anatomical structures increases. In the range of 120 to 150 images, due to the medical assistant placing the coil on the patient's right knee, the confidence levels of the right knee and the right ankle of the foot decrease. The coil, the corresponding cable, and the medical assistant cover the right ankle of the foot and the right knee, which reduces the confidence level of each item. During this process, the right hip is not covered by any obstacles, so that the camera has a free field of view on the right hip. The free field of view on the right hip results in a high confidence level of approximately 0.9. After the positioning process of positioning the coil on the patient is completed, the confidence levels of the right ankle of the foot and the right knee slightly increase in the range of 150 to 200 images, but do not reach the previous high confidence level (i.e., 0.9 reached by the right hip). This is because the coil and the corresponding cable become obstacles. In the range of 200 to 300 images, as the patient moves into the bore with the patient support and is thus invisible to the camera, the confidence levels of both the landmarks and the target anatomical structures decrease. It can be seen from this example that the confidence level of the region of interest, the position of the right knee, decreases during the preparation, and the confidence level of one landmark (i.e., the right ankle of the foot) decreases. However, the second landmark (the right hip) remains visible, so the confidence level of the right hip has a high value until the patient moves into the bore.

[0062] Figure 3 is a schematic diagram of a method according to a first aspect of the present disclosure. This computer-implemented method is used to prepare an object in medical imaging. The object is a patient in the current situation. The right knee of the patient should be prepared for an MRI procedure. The right knee is the region of interest in the current situation. The method includes the following steps:

[0063] In a first step S10, a series of images of an area of interest including at least a portion of an object is obtained, where the series of images includes at least a first image and at least a subsequent second image. The images are obtained by a digital optical camera implemented above a patient support on which a patient is positioned and prepared for an imaging procedure. In step S20, the position of at least one landmark is obtained from the series of images, where the at least one landmark is anatomically related to a target anatomical structure. The position is obtained by an image analysis algorithm (in particular an algorithm based on a deep neural network). The algorithm also includes an image acquisition model that simulates the process of obtaining the images. The model describes irregularities (e.g., alignment errors, rotation errors) that occur during the imaging process. As a result, the image acquisition model generates a number of output images indicating the irregularities based on an acquired input image (e.g., the first image in the series of images). Based on the number of output images, a deep convolutional neural network determines the position of the landmark in each of the number of output images. The average result of the positions of the landmarks in the number of output images then serves as the position of the landmark in the image. In step S30, a confidence level assigned to the position of at least one landmark is obtained. The confidence level is determined by calculating the entropy between the input image and the number of output images. Alternatively, the confidence level is determined by calculating the deviation of the position of the landmark in the number of output images. A description of an image analysis algorithm that can be used for steps S10 to S30 described in the present disclosure is described in the article “Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks” by Guotai Wang, Wenqi Li, Michael Aertsen, Jan Deprest, Sebastian Ourselin, Tom Vercauteren (published in Neurocomputing, 2019,

[0064] https: / / doi.org / 10.1016 / j.neucom.2019.01.103). The content of the article is incorporated herein by reference in its entirety.

[0065] In step S40, the position of the target anatomical structure is determined based on the position and confidence level of at least one landmark. For example, in Figure 2In Image 3, the position and confidence level of the ankle and the position and confidence level of the right hip are used to determine the position of the right knee. Since the confidence level of the ankle is lower than a predetermined threshold value of 0.85, the position of the ankle in this image is not used to determine the position of the right knee. Instead, the position of the ankle in Image 2 of Figure 2 and the position of the right hip in Image 3 of Figure 2 are used because the confidence level of the position of the right hip in Image 3 is higher than the predetermined threshold. In other words, when the confidence levels of all landmarks are not higher than the predetermined threshold, more than one image is used to determine the position of the right knee. In Figure 2 Image 2, all confidence levels of the positions are higher than the predetermined threshold, so only one image is used to determine the position of the right knee. Alternatively, additional previous images can also be used to determine the position of the right knee. This may be useful to improve reliability. Additionally, it should be noted that in this example, the position of the right knee is also obtained using a neural network, so the confidence level is also determined as described above. In other examples or applications, the target anatomical structure (e.g., an organ such as the liver) is not visible. The position of the liver must be determined by landmarks; otherwise, it cannot be determined. In step S50, the position of the target anatomical structure is provided for preparing the object in medical imaging. The information about the position of the target anatomical structure can be transmitted to the control of the imaging system or imaging unit. The information about the position of the target anatomical structure can be displayed on the screen to guide medical assistants.

[0066] Figure 4 is a schematic diagram of a device according to an embodiment of the present disclosure. A device 40 for preparing an object in medical imaging includes: an acquisition unit 41 configured to acquire a series of images of an area of interest including at least a part of the object, where the series of images includes at least a first image and a subsequent second image; a first determination unit 42 configured to determine the position of at least one landmark based on the series of images, where the at least one landmark is anatomically related to the target anatomical structure; a second determination unit 43 configured to determine the confidence level assigned to the position of the at least one landmark; a third determination unit 44 configured to determine the position of the target anatomical structure based on the position and confidence level of the at least one landmark; and a provision unit 45 configured to provide the position of the target anatomical structure for preparing the object in medical imaging.

Claims

1. A computer-implemented method for preparing an object in medical imaging, comprising: obtaining a series of images of a region of interest including at least a portion of the object, wherein the series of images includes at least a first image and at least a subsequent second image; determining the position of at least one landmark based on the series of images, the position of the at least one landmark being detected and tracked in the series of images, wherein the at least one landmark is anatomically related to a target anatomical structure; determining a confidence level assigned to the position of the at least one landmark; determining the position of the target anatomical structure based on the position of the at least one landmark in the series of images and the confidence level on the series of images, wherein a weighting factor for a shift of the position of the at least one landmark is determined based on the determined confidence level, and wherein the weighting factor is used to determine the position of the target anatomical structure; providing the position of the target anatomical structure for preparing the object in medical imaging.

2. The method according to claim 1, wherein In the case where the confidence level of the at least one landmark in the second image is lower than a predetermined threshold, the position of the target anatomical structure is determined based on the at least one landmark in the first image.

3. The method according to claim 1 or 2, wherein, Determining the position of the target anatomical structure further includes determining a shift of the position of the at least one landmark between the first image and the second image.

4. The method according to any one of the preceding claims, wherein, In the case where the confidence level of the at least one landmark in the second image is higher than the predetermined threshold, the position of the target anatomical structure is determined based on: either a shift of the position of the at least one landmark between the first image and the second image, or the at least one landmark in the second image.

5. The method according to any one of the preceding claims, wherein, The at least one landmark is selected based on the presence of the landmark in the series of images.

6. The method according to any one of the preceding claims, wherein, The target anatomical structure is obscured by an obstacle.

7. The method according to any one of the preceding claims, wherein, The position of the target anatomical structure is derived from a plurality of landmarks.

8. The method according to any one of the preceding claims, wherein Determining the position of the at least one landmark and determining the position of the target anatomical structure are based on an image analysis algorithm.

9. The method according to any one of the preceding claims, wherein, A control signal for controlling an imaging unit is derived based on the determined position of the target anatomical structure.

10. The method according to any one of the preceding claims, wherein, Guiding data for preparing the object in medical imaging is derived based on the position of the target anatomical structure, and wherein the guiding data includes a target alignment of the object relative to the imaging unit.

11. The method according to any one of the preceding claims, wherein, The determination of the position of the target anatomical structure is based on one or more degrees of freedom of one or more joints of the object.

12. An apparatus for preparing an object in medical imaging, comprising: an obtaining unit configured to obtain a series of images of a region of interest including at least a portion of the object, wherein the series of images includes at least a first image and a subsequent second image; A first determination unit configured to determine the positions of at least one landmark based on the series of images, the positions of the at least one landmark being detected and tracked in the series of images, wherein the at least one landmark is anatomically related to a target anatomical structure; A second determination unit configured to determine a confidence level assigned to the positions of the at least one landmark; A third determination unit configured to determine the position of the target anatomical structure based on the positions of the at least one landmark in the series of images and the confidence levels on the series of images, wherein a weighting factor for a shift of the positions of the at least one landmark is determined based on the determined confidence levels, and wherein the weighting factor is used to determine the position of the target anatomical structure; A providing unit configured to provide the position of the target anatomical structure for preparing the subject in medical imaging.

13. An imaging system comprising: The device according to claim 12; An imaging unit; An imaging control unit.

14. A computer-readable medium storing a computer program, the computer program being configured, when run by a processor, to: perform the method according to any one of claims 1 to 11, and / or control the device according to claim 12, and / or control the system according to claim 13.

Images