Annotating medical images

Abstract

Manual placement of indicia in medical imaging is destructive and error-prone to workflow. Accordingly, a computer-implemented method for annotating a medical image is provided. The method includes obtaining a medical image to be annotated, and applying an annotation algorithm to the medical image. An annotation algorithm derives an entropy map from the obtained medical image to be annotated, identifies at least one region of the medical image, the at least one region being a low entropy region within the entropy map, and determines values of one or more parameters to be applied to post-acquisition annotations of the medical image. The one or more parameters include image coordinates annotating a position in the medical image after acquisition, the image coordinates being within the at least one low entropy region. In addition, the method includes annotating the medical image by applying the post-acquisition annotation to the medical image using the determined value, and outputting the annotated medical image. The method improves the workflow by automating marker placement.

Description

Technical Field

[0001] This invention relates to methods and systems for annotating medical images. Background Technology

[0002] In medical imaging, a patient's anatomical structures can be positioned in different orientations, and these structures can exhibit different lateralities. For example, in anterior-to-posterior (AP) radiographs, posterior structures are closer to the detector; while in posterior-to-anterior (PA) images, anterior structures are closer to the detector. Laterality refers to which side of the body is considered (i.e., left or right). Without prior knowledge, a person viewing X-ray images may not be able to identify the orientation or laterality of a patient's anatomical structures. Therefore, images should be annotated as part of the examination. Typically, radiographers annotate images manually using pre-acquisition markers (e.g., guide markers placed in the field of view indicating "L = Left" or "R = Right") or post-acquisition markers. In either case, marker placement is disruptive to the radiographic workflow and prone to error. Furthermore, marker placement may be subject to hospital-specific guidelines governing the location and / or size of the markers.

[0003] According to US2021 / 383565A1, a computer-implemented method is available for training a probability-based computational model to determine the location of an image representation of an annotated anatomical structure in a two-dimensional X-ray image. Summary of the Invention

[0004] To better address one or more of these concerns, in a first aspect of the invention, a computer-implemented method for annotating medical images is provided. The method includes: obtaining a medical image to be annotated; and applying an annotation algorithm to the medical image, wherein the annotation algorithm derives an entropy map from the obtained medical image to be annotated, identifies at least one region of the medical image, the at least one region being a low-entropy region within the entropy map, and determines values ​​for one or more parameters to be applied to a post-acquisition annotation of the medical image. The one or more parameters include image coordinates of the location of the post-acquisition annotation in the medical image, the image coordinates being within at least one of the low-entropy regions. Additionally, the method includes: annotating the medical image by applying the post-acquisition annotation to the medical image using the determined values, and outputting the annotated medical image.

[0005] In contrast to the possible methods disclosed in US2021 / 0383565A1, the annotation method proposed in this invention allows the use of post-collection information.

[0006] The one or more parameters may also include a scaling factor for the post-capture annotation. The one or more parameters may also include the brightness of the post-capture annotation. The one or more parameters may also include the orientation or rotation of the post-capture annotation, for example, horizontal or vertical orientation. The one or more parameters may also include the text direction of the annotation, for example, left-to-right script or right-to-left script.

[0007] An annotation algorithm can determine the values ​​of one or more parameters based on predetermined criteria. For example, the annotation algorithm can determine the values ​​of the one or more parameters at least in part based on the type of the post-acquisition annotation. The type of annotation can correspond to one or more of the following: lateralization marker; view position (orientation) marker; operator ID marker. For example, the annotation algorithm can determine the size of the post-acquisition annotation based on the type and / or diagnostic relevance of the post-acquisition annotation, wherein, for example, the lateralization or orientation marker is larger than the operator ID marker. The annotation algorithm can determine the values ​​of the one or more parameters at least in part based on image content. For example, the annotation algorithm can determine the values ​​of the one or more parameters to avoid the post-acquisition annotation overlapping with at least one pre-identified region in the medical image depicting diagnostically relevant structures and / or foreign bodies. In another example, the annotation algorithm can determine the values ​​of the one or more parameters to place the post-acquisition annotation in at least one direct radiation area of ​​the medical image. Different priorities can be assigned to different criteria, wherein, for example, overlap with diagnostically relevant structures assumed to have the highest priority is avoided. When brightness is used as a parameter, the annotation algorithm can determine the brightness of the post-acquisition annotation at least in part based on the background brightness of the medical image. In this context, predetermined criteria allow the brightness to be set so that the annotation is not too bright but still visually distinguishable from the background. In the example, by using orientation as a parameter, the annotation algorithm can be configured to rotate normally horizontal annotations to a vertical orientation to avoid overlapping with at least one pre-identified region, such as in the case of a chest examination, where the relatively small direct radiation area leaves little space for horizontal annotation.

[0008] Applying post-acquisition annotations to medical images using the determined values ​​can include any suitable techniques for applying annotations directly and / or indirectly to the image. In one example, the values ​​determined by the annotation algorithm can be stored together with the post-acquisition annotations in the metadata of the medical image for later use in displaying the annotations as an overlay, for example, on a viewing station. Alternatively or additionally, the annotation algorithm can use the determined values ​​of one or more parameters to directly plot the post-acquisition annotations onto the medical image, which may be useful in the case of printing an annotated image. Different categories can be assigned to different types of annotations, with some annotations burned into the image while others are provided as part of the metadata for on-demand availability.

[0009] Annotation algorithms can utilize one or more conventional and / or AI-based techniques to determine values. Annotation algorithms can utilize trained machine learning models to determine the values ​​of one or more parameters. Additionally or alternatively, annotation algorithms can utilize models of anatomical structures depicted in medical images to determine the image coordinates of the post-acquisition annotation location, at least in part, based on one or more landmarks included in the model. In this case, the annotation algorithm can also utilize prior knowledge about the orientation and / or laterality of the anatomical structures to determine the image coordinates. Additionally or alternatively, annotation algorithms can utilize image processing algorithms to select values ​​for one or more parameters that prevent the post-acquisition annotation from overlapping with at least one pre-identified region in the medical image depicting diagnostically relevant structures and / or foreign bodies. Annotation algorithms can use any combination of such techniques to determine the values ​​of one or more parameters.

[0010] Annotation algorithms utilize entropy when determining one or more values. Annotation algorithms can identify at least one region in a medical image that depicts diagnostically relevant structures based on entropy within the image. Annotation algorithms can also identify at least one direct radiation region in a medical image based on entropy within the image, wherein the at least one direct radiation region includes appropriate image coordinates for the location of post-acquisition annotation.

[0011] Obtaining medical images for annotation may include acquiring images using a medical imaging system or receiving pre-acquired images. Any suitable medical imaging modality (but especially those requiring standard annotation) can be used to acquire medical images. For example, a medical image may be an X-ray image acquired using an X-ray imaging system (e.g., a moving or stationary radiography system or a fluoroscopy system). In other examples, the medical imaging system used to acquire the image may be a magnetic resonance imaging (MRI) system, a positron emission tomography (PET) system, an image-guided hyperthermia (IGHT) system, a single-photon emission computed tomography (SPECT) system, a computed tomography (CT) system, an ultrasound (US) imaging system, or a computed tomography medical imaging system. In other examples, multiple imaging modalities may be used to acquire medical images. Medical images may be two-dimensional or three-dimensional.

[0012] Outputting annotated medical images may include displaying images on a user interface, storing images for later display, or transmitting images for display or storage.

[0013] Image coordinates can define the location within a medical image (e.g., X-axis, Y-axis, and optionally Z-axis values) based on axis values ​​that define the position within the medical image relative to a predetermined origin (e.g., a corner of the medical image). Axis values ​​can be expressed in pixels or any other suitable unit for defining spatial location.

[0014] Post-acquisition annotations may include any graphic and / or textual content used to describe the content of the images, such as markers used to indicate the orientation and / or laterality of anatomical structures.

[0015] The methods and systems described herein can improve workflows by automating marker placement and / or resizing. Automated marker placement, as described, ensures that markers are placed in configurable, view- and anatomically relevant locations, avoiding overlap with relevant anatomical structures and / or foreign bodies, while simultaneously ensuring readability. Automated marker resizing, as described herein, ensures that markers are not displayed as too large (which can be distracting, especially in pediatric examinations or examinations of small body parts such as fingers or toes) or too small (especially in the case of stitched images).

[0016] According to the second aspect, a computing system for performing the method of the first aspect is provided.

[0017] According to a third aspect, a computer program (product) including instructions is provided, which, when executed by a computing system, enable the computing system to perform the method of the first aspect.

[0018] According to a fourth aspect, a computer-readable (storage) medium is provided, comprising instructions that, when executed by a computing system, enable the computing system to perform the method of the first aspect. The computer-readable medium may be transient or non-transient, volatile or non-volatile.

[0019] As used herein, the term “acquire” can include, for example, receiving from another system, device, or process; receiving via interaction with a user; loading or retrieving from a storage device or memory; or measuring or capturing using a sensor or other data acquisition device.

[0020] As used herein, the term "determine" encompasses a wide variety of actions and may include, for example, calculation, operation, processing, derivation, investigation, searching (e.g., searching in a table, database, or other data structure), ascertainment, etc. Furthermore, "determine" may include receiving (e.g., receiving information), accessing (e.g., accessing data in memory), etc. Moreover, "determine" may include parsing, selecting, choosing, building, etc.

[0021] The quantifiers “one” or “a” do not exclude multiple. Furthermore, as used herein, the quantifiers “one” and “a” should generally be interpreted as meaning “one or more” unless otherwise stated or the context clearly indicates a single form.

[0022] Unless otherwise stated or clearly indicated by the context, the phrases “one or more of A, B, and C,” “at least one of A, B, and C,” and “A, B, and / or C” as used herein are intended to refer to all possible permutations of one or more of the listed items. That is, the phrase “A and / or B” means (A), (B), or (A and B), while the phrase “A, B, and / or C” means (A), (B), (C), (A and B), (A and C), (B and C), or (A, B, and C).

[0023] The term "comprising" does not exclude other elements or steps. Furthermore, the terms "comprising," "including," and "having" are used interchangeably herein.

[0024] This invention may include one or more aspects, examples, or features used alone or in combination, whether disclosed in the combination or individually. Any optional feature or sub-aspect of one of the foregoing aspects is applicable to any other aspect.

[0025] The above aspects will become apparent and clarified by referring to and utilizing the detailed description provided below. Attached Figure Description

[0026] A detailed description will now be given by way of example only, with reference to the accompanying drawings, in which: Figure 1The illustration shows manually placed markers in a medical image; Figure 2 Two medical images are shown, viewed at different zoom levels; Figure 3 The diagram illustrates the steps of an algorithm for automatically placing and scaling markers in medical images; and Figure 4 The diagram illustrates a computing system that can be used according to the systems and methods disclosed herein. Detailed Implementation

[0027] Figure 1 The illustration shows two medical images 100 with manually placed markers applied. In the left image 100-A, the radiologist has manually placed a post-acquisition marker 102 using a software package to indicate left-sided deviation of the patient's hand. In the right image 100-B, the radiologist has placed a pre-acquisition marker (guide marker) 104 on the detector next to the patient's hand to indicate left-sided deviation. Because lead is a radiopaque material, marker 104 appears in the acquired image.

[0028] Figure 2 Two medical images 200 are shown viewed at different zoom factors. The left image 200-A shows a finger examination, while the right image 200-B shows a hand examination. In each image 200, markers 202 are placed to indicate left-side lateralization. Medical images are typically viewed using a zoom adapted to the screen. Therefore, marker 202-A in the finger image 200-A is larger than marker 202-B in the hand image 202-B, even though the same marker size (relative to the detector pixel matrix) is chosen. This effect becomes more pronounced when the same marker configuration is used for pediatric finger examinations and for large splicing examinations spanning several detector heights.

[0029] Figure 3 The illustration shows the steps of an algorithm for automatically placing and scaling markers in medical images according to this disclosure, which is used to improve radiological workflows and reduce the total time for patient examinations.

[0030] In step S1, a medical image 300 to be annotated is obtained. In the non-limiting example shown, image 300 is an X-ray image of a patient (left hand).

[0031] In step S2, the medical image 300 may optionally be preprocessed. Figure 3 In the non-limiting example shown, preprocessing includes rotation of image 300. It should be understood that other preprocessing techniques (e.g., other geometric transformations; noise reduction, contrast enhancement, image resizing, segmentation, or feature extraction) may be applied to image 300.

[0032] In subsequent steps, the annotation algorithm is applied to the medical image 300 to determine the values ​​of one or more parameters to be applied to the post-acquisition annotation of the image.

[0033] In step S3, the annotation algorithm determines the image coordinates of the position of the annotation in the medical image 300 after acquisition. Figure 3 A medical image 300 is shown that is annotated by applying a left lateralization marker 302 to the image 300 using image coordinates where the marker 302 is placed at the upper right corner of the image 300. Step S3 may include manually or automatically selecting the correct marker. For example, in cases where lateralization is unknown, one or more techniques known in the art can be used to perform automatic detection of lateralization, thereby performing automatic detection of the correct marker.

[0034] In step S4, the annotation algorithm may optionally use a determined scaling factor to adjust the size of marker 302 to achieve the desired appearance. For example, returning to... Figure 2 The scaling factor can be determined so that marker 302 appears at an appropriate size (i.e., within predefined constraints) under multiple different zoom factors. In a variant of the algorithm executed at the viewing station where the image is viewed, the annotation algorithm can use the zoom factor selected for display at the viewing station as additional input for marker size determination, for example, so that marker 302 appears at a consistent size relative to the screen size under multiple different zoom factors. Figure 3 In the non-limiting example shown, the marker size is automatically selected from four different available marker sizes. The marker size can be selected at least in part based on the type of marker (e.g., laterality, orientation, etc.). Furthermore, the annotation algorithm can optionally modify the brightness of the marker 302 according to the background brightness of the medical image 300.

[0035] The algorithm terminates by outputting an annotated medical image 300 for display or storage. The determined values, along with a marker 302 (e.g., "L" or "R"), can be stored in X-ray image metadata (e.g., DICOM (Digital Imaging and Communications in Medicine) presentation status) for later display. Alternatively, the determined values ​​can be used to directly draw (i.e., "burn") the marker 302 into the X-ray image 300.

[0036] Annotation algorithms can be implemented in different ways. This disclosure proposes using AI-based methods and / or conventional image processing methods.

[0037] In one example of an end-to-end approach, end-to-end training is applied by training a machine learning model to output image coordinates using a training dataset that includes X-ray images and (e.g., based on image metadata) known or annotated marker locations. The DICOM rendering state can include information about marker locations and / or sizes. This information can be utilized using any appropriate model training technique. The model can be trained for each hospital, allowing it to learn hospital-specific annotation conventions.

[0038] In one example of a conventional image processing method, the annotation algorithm is configured to select a default initial position for the annotation after manual or automatic rotation of the anatomical structure. Depending on the anatomical structure, examination type, and collimation dimensions, the annotation algorithm selects a default marker size (e.g., font size) and adjusts the position and / or font size to avoid overlap with diagnostically relevant information and / or foreign bodies appearing in the medical image 300. The annotation algorithm can also utilize additional image content information, such as direct radiation areas, soft tissue areas, and bone contours, to further adjust the marker position and / or size.

[0039] In one example of a model-based approach, model-based quality assurance software (which uses 2D atlases or 3D joint models to determine pose, alignment, and / or posture issues) is extended to perform marker placement with reference to landmarks appearing in the atlas or model. An annotation algorithm determines appropriate marker locations within the X-ray image by using a forward projection of an adaptive model. Prior knowledge about body parts, orientation, and lateralization can be considered.

[0040] In any of the examples described herein, the annotation algorithm checks the entropy within the X-ray image before placing the annotation to prevent the annotation from being placed on top of diagnostically relevant structures and / or to promote the placement of the annotation in low-entropy regions that indicate flat areas (e.g., areas within the direct radiation zone of the image).

[0041] More specifically, the method of using image entropy in the annotation algorithm can be described in four steps, one or more of which may be implemented in a practical embodiment.

[0042] The first step involves entropy calculation and mapping. Entropy is calculated for each pixel or region within the medical image. This calculation can be performed using a predefined entropy calculation algorithm (e.g., the Shannon entropy algorithm). Calculating entropy at the pixel level and analyzing pixel intensity values ​​and their distribution yields an entropy map. In this entropy map, high-entropy regions and low-entropy regions are associated with high-complexity regions and high-homogeneity regions, respectively.

[0043] The second step involves annotation placement strategies. Low-entropy regions indicate areas with homogeneous and less complex information, which are unlikely to contain diagnostically important details. These regions are suitable for annotation placement. High-entropy regions, on the other hand, are less suitable for annotation placement and will be avoided to ensure that annotations do not obscure key diagnostic information, as they may correspond to important anatomical structures or regions of interest.

[0044] The third step involves dynamic adjustment of the annotations. Initially, annotations are placed based on a heuristic or pre-trained model. This placement is then refined using an entropy map. For example, annotations initially placed near or within high-entropy regions by the heuristic or pre-trained model are dynamically adjusted toward adjacent low-entropy regions. This adjustment may use gradient descent methods, moving the annotations toward the lowest entropy gradient in the local vicinity.

[0045] The fourth step involves cross-validating the annotations against a benchmark dataset to ensure they do not obscure diagnostically relevant regions. Feedback from this validation step is used to further optimize the annotation algorithm. Finally, the entropy calculation and annotation adjustment process may be iteratively repeated to fine-tune the annotation positions to consistently avoid highly informative regions.

[0046] Figure 4 An exemplary computing system 800 capable of being used according to the systems and methods disclosed herein is illustrated. The computing system 800 may be part of or include any desktop computer, laptop computer, server, or cloud-based computing system. The computing system 800 includes at least one processor 802 that executes instructions stored in memory 804. Instructions may be, for example, instructions for implementing functions described as being performed by one or more components described herein, or instructions for implementing one or more methods described herein. The processor 802 can access memory 804 via system bus 806. In addition to storing executable instructions, memory 804 may also store session inputs, scores assigned to session inputs, etc.

[0047] The computing system 800 additionally includes a data storage unit 808 accessible by the processor 802 via the system bus 806. The data storage unit 808 may include executable instructions, log data, etc. The computing system 800 also includes an input interface 810 that allows external devices to communicate with the computing system 800. For example, the input interface 810 can be used to receive instructions from external computer devices, users, etc. The computing system 800 also includes an output interface 812 that interfaces the computing system 800 with one or more external devices. For example, the computing system 800 can display text, images, etc., through the output interface 812.

[0048] It is conceivable that external devices capable of communicating with the computing system 800 via input interface 810 and output interface 812 are included in an environment that provides a user interface with which users can interact in virtually any type. Examples of user interface types include graphical user interfaces (GUIs), natural user interfaces (NUMAs), and the like. For example, a GUI can accept input from a user using one or more input devices (e.g., a keyboard, mouse, remote control, etc.) and provide output on an output device (e.g., a display). Conversely, a NUMA allows a user to interact with the computing system 800 in a manner unconstrained by input devices (e.g., a keyboard, mouse, remote control, etc.). In practice, NUMAs can rely on speech recognition, touch and stylus recognition, on-screen and near-screen gesture recognition, air gestures, head and eye tracking, voice and speech, vision, touch, gestures, machine intelligence, and the like.

[0049] Furthermore, although the computing system 800 is illustrated as a single system, it should be understood that the computing system 800 can be a distributed system. Therefore, for example, several devices can communicate via a network connection and collaboratively perform tasks described as being performed by the computing system 800.

[0050] The various functions described herein can be implemented using hardware, software, or any combination thereof. If a function is implemented in software, the function can be stored as one or more instructions or code on or transmitted through a computer-readable medium. Computer-readable media include computer-readable storage media. A computer-readable storage medium can be any available storage medium accessible by a computer. By way of example and not limitation, such computer-readable storage media can include flash memory storage media, RAM, ROM, EEPROM, CD-ROM or other optical disc storage devices, magnetic disk storage devices or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer. As used herein, “disk” and “optical disc” include compact discs (CDs), laser discs, optical discs, digital versatile discs (DVDs), floppy disks, and Blu-ray discs (BDs), wherein disks typically reproduce data magnetically, and optical discs typically reproduce data optically using lasers. Additionally, the propagation of signals can be included within the scope of computer-readable storage media. Computer-readable media also include communication media, which includes any medium that facilitates the transfer of a computer program from one place to another. For example, a connection can be a communication medium. For example, if software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of communication media. Combinations of the above items should also be included within the scope of computer-readable media.

[0051] Alternatively or additionally, the functions described herein can be performed at least in part by one or more hardware logic components. For example, and not as a limitation, illustrative types of hardware logic components that can be used include field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), etc.

[0052] The applicant hereby discloses individually each individual feature described herein, as well as any combination of two or more such features, provided that such features or combinations can be performed based on this specification as a whole, in accordance with common general knowledge of those skilled in the art, regardless of whether such features or combinations of features solve any problem disclosed herein, and not to limit the scope of the claims. The applicant notes that aspects of the invention can consist of any such individual features or combinations of features.

[0053] It should be noted that embodiments of the invention are described with reference to different categories. In particular, some examples are described with reference to methods, while other examples are described with reference to apparatus. However, those skilled in the art will appreciate from the description that, unless otherwise stated, this application discloses any combination of features relating to different categories, in addition to any combination of features belonging to one category. However, it is possible to combine all features to provide synergistic effects beyond a simple summation of features.

[0054] While the invention has been illustrated and described in detail in the accompanying drawings and the foregoing description, such illustrations and descriptions are to be considered exemplary and not restrictive. The invention is not limited to the disclosed embodiments. Other variations of the disclosed embodiments will be understood and implemented by those skilled in the art through study of the drawings, the disclosure, and the appended claims.

[0055] The mere fact that certain measures are described in different dependent claims does not imply that combinations of these measures cannot be used advantageously.

[0056] No reference numerals in the claims should be construed as limiting the scope.

Claims

1. A computer-implemented method for annotating medical images, the method comprising: Obtain the medical image to be annotated (300); The annotation algorithm is applied to the medical image, wherein the annotation algorithm performs the following operations: Derive the entropy map from the obtained medical images (300) to be annotated. Identify at least one region of the medical image (300), said at least one region being a low-entropy region within the entropy map. Determine the values ​​of one or more parameters to be applied to the post-acquisition annotation (302) of the medical images. Wherein, the one or more parameters include the image coordinates of the location annotated in the medical image after acquisition, the image coordinates being within at least one of the low-entropy regions, and The medical image is annotated by applying the post-acquisition annotation to the medical image using determined values; and Output annotated medical images.

2. The method according to claim 1, wherein, The one or more parameters also include a scaling factor for the post-acquisition annotation (302).

3. The method according to claim 1 or 2, wherein, The annotation algorithm determines the value of one or more parameters based on the type of the post-collection annotation (302).

4. The method according to claim 3, wherein, The type of the post-collection annotation (302) corresponds to one or more of the following: lateralization marker; viewing position marker; operator ID marker.

5. The method according to any of the preceding claims, wherein, The one or more parameters also include the brightness of the post-acquisition annotation (302), and wherein the annotation algorithm determines the brightness at least in part based on the background brightness of the medical image (300).

6. The method according to any of the preceding claims, wherein, The value determined by the annotation algorithm, together with the post-acquisition annotation (302), is stored in the metadata of the medical image (300).

7. The method according to any one of claims 1-5, wherein, The annotation algorithm uses the determined values ​​of one or more parameters to directly plot the acquired annotations (302) onto the medical image (300).

8. The method according to any of the preceding claims, wherein, The annotation algorithm uses a trained machine learning model to determine the values ​​of the one or more parameters.

9. The method according to any of the preceding claims, wherein, The annotation algorithm utilizes a model of the anatomical structures depicted in the medical image (300) to determine the image coordinates of the location to be annotated after acquisition, based at least in part on one or more landmarks included in the model.

10. The method according to claim 9, wherein, The annotation algorithm also utilizes prior knowledge about the orientation and / or laterality of the anatomical structures to determine the image coordinates.

11. The method according to any of the preceding claims, wherein, The annotation algorithm uses an image processing algorithm to select the values ​​of one or more parameters, the values ​​of which prevent the post-acquisition annotation (302) from overlapping with at least one pre-identified region in the medical image (300) that depicts a diagnosis-related structure and / or foreign body.

12. The method according to any of the preceding claims, wherein, The annotation algorithm identifies at least one direct radiation area of ​​the medical image (300) based on the entropy within the medical image, wherein the at least one direct radiation area includes the appropriate image coordinates of the location of the post-acquisition annotation (302).

13. A computing system (800) configured to perform the method according to any of the preceding claims.

14. A computer-readable medium (804, 808) including instructions that, when executed by a computing system (800), cause the computing system to perform the method according to any one of claims 1-12.

Images