Method for tracking lesion in medical images, and medical image analysis device and computer program for providing same

Abstract

This medical image analysis method performed by a medical image analysis device comprises the steps of: acquiring medical images of a patient captured at different time points; analyzing at least one medical image captured at each time point, so as to detect a lesion, and recognizing location identification information inserted in the corresponding medical image so as to identify an anatomical location of the lesion; clustering, on the basis of the anatomical location of each lesion detected from the medical images, lesions detected from the medical images, so as to generate lesion clusters; and identifying lesions included in each lesion cluster as the same lesion, and comparing images at different time points including the same lesion so as to track changes in the lesions over time.

Description

Method for tracking lesions in medical images, and a medical image analysis device and computer program providing the same

[0001] The present disclosure relates to medical image analysis technology.

[0002] Breast cancer and thyroid cancer are among the most common cancers. For breast cancer, early detection significantly impacts treatment outcomes and can greatly increase survival rates. While thyroid cancer is curable in most cases, treatment can become complicated if not detected early. Therefore, accurate diagnosis and continuous follow-up are essential.

[0003] Ultrasound is a widely used modality due to its ability to provide real-time imaging, its safety, and its cost-effectiveness. In this regard, ultrasonography is crucial for the early detection of breast and thyroid cancers, and ultrasound-assisted mass tracking plays an important role in monitoring and analyzing changes in the mass over time to determine if the nature of the tumor is changing.

[0004] However, comparing previous and current examinations to consistently track the location and changes of a mass requires significant time, effort, and advanced technology. Furthermore, the consistency of results may be compromised because the process relies on the subjective judgment of medical professionals. Additionally, since the interpretation of ultrasound images often depends on the examiner's experience, it is difficult to consistently track the same mass over time. Therefore, technology capable of automatically analyzing changes in a mass over time is required.

[0005] The present disclosure relates to a method for tracking lesions in medical images, and a medical image analysis device and computer program for providing the same.

[0006] A method for analyzing medical images using a medical image analysis device according to some embodiments comprises: acquiring medical images of a patient taken at different times; analyzing at least one medical image taken at each time point to detect a lesion and recognizing location identification information inserted in the medical image to identify the anatomical location of the lesion; clustering the lesions detected in the medical images based on the anatomical location of each lesion detected in the medical images to generate lesion clusters; and identifying the lesions included in each lesion cluster as the same lesion and comparing images at different times point containing the same lesion to track changes in the lesion over time.

[0007] The above location identification information may include body markers or comments entered by an inspector.

[0008] The step of identifying the anatomical location of the above lesion may further specify or fine-tune the anatomical location by utilizing location-related information extracted from the relevant medical image.

[0009] The above medical image analysis method may further include the step of providing tracked lesion change information for each identified lesion to a user interface.

[0010] The above medical image analysis method may further include the step of generating a report containing lesion change information for at least one lesion detected in the medical images and providing it to a user interface.

[0011] A method for analyzing medical images using a medical image analysis device according to some embodiments comprises: acquiring ultrasound images of a patient taken at time intervals; recognizing body markers inserted in the ultrasound images or comments entered by an examiner to identify the anatomical location of a lesion detected in the corresponding images; extracting follow-up images containing the same lesion from the ultrasound images through lesion location-based clustering; and comparing the follow-up images to track changes in the lesion over time.

[0012] The step of identifying the anatomical location of the above lesion may further specify or fine-tune the anatomical location by utilizing location-related information extracted from the corresponding image.

[0013] The above medical image analysis method may further include the step of providing tracked lesion change information for each identified lesion to a user interface.

[0014] The above medical image analysis method may further include the step of generating a report containing lesion information and lesion change information for at least one lesion detected in the ultrasound images and providing it to a user interface.

[0015] A medical image analysis device according to some embodiments comprises at least one memory and at least one processor that executes instructions stored in the at least one memory, wherein the processor, by executing the instructions, acquires medical images of a patient taken at different times, analyzes at least one medical image taken at each time point to detect a lesion, recognizes location identification information inserted in the medical image to identify the anatomical location of the lesion, clusters the lesions detected in the medical images based on the anatomical location of each lesion detected in the medical images to generate lesion clusters, identifies the lesions included in each lesion cluster as the same lesion, and compares images at different times point containing the same lesion to track changes in the lesion over time.

[0016] The above processor can be implemented to provide tracked lesion change information for each identified lesion to the user interface.

[0017] The processor may be implemented to generate a report including lesion information and lesion change information for at least one lesion detected in the ultrasound images and provide it to a user interface.

[0018] A computer program stored on a computer-readable recording medium according to some embodiments comprises instructions for executing the steps of: acquiring medical images of a patient taken at different times; analyzing at least one medical image taken at each time point to detect a lesion and recognizing location identification information inserted in the medical image to identify the anatomical location of the lesion; clustering the lesions detected in the medical images based on the anatomical location of each lesion detected in the medical images to generate lesion clusters; and identifying lesions included in each lesion cluster as the same lesion and comparing images at different times point containing the same lesion to track changes in the lesion over time.

[0019] The above computer program may further include instructions that enable the execution of a step to provide tracked lesion change information for each identified lesion to a user interface.

[0020] The above computer program may further include instructions that enable the execution of the step of generating a report containing lesion information and lesion change information for at least one lesion detected in the ultrasound images and providing it to a user interface.

[0021] According to the embodiment, changes in lesions over time can be automatically analyzed and consistently tracked, thereby reducing the time and cost for follow-up observation and providing reliable information to medical staff.

[0022] According to the embodiment, even in cases where it is difficult to determine the identity of a lesion based solely on the location of the lesion within an image, such as with ultrasound images, the same lesion can be identified in follow-up images through clustering based on the anatomical location of the lesion, thereby enabling the automation of changes in the lesion over time.

[0023] According to the embodiment, it can be utilized as a diagnostic assistance solution to support the decision-making of medical staff.

[0024] FIG. 1 is a drawing illustrating a medical image analysis device according to one embodiment.

[0025] FIG. 2 is a diagram illustrating a method for identifying the location of a lesion according to one embodiment.

[0026] FIG. 3 is a diagram illustrating lesion location-based clustering according to one embodiment.

[0027] FIGS. 4 through 6 are examples of user interfaces provided by a medical image analysis device according to one embodiment.

[0028] FIG. 7 is a flowchart of a method for tracking lesions in medical images according to one embodiment.

[0029] FIG. 8 is a flowchart of a method for tracking lesions in ultrasound images according to one embodiment.

[0030] Embodiments of the present disclosure are described below with reference to the attached drawings so that those skilled in the art can easily implement them. However, the present disclosure may be embodied in various different forms and is not limited to the embodiments described herein. Furthermore, in order to clearly explain the present disclosure in the drawings, parts unrelated to the explanation have been omitted, and similar parts throughout the specification are denoted by similar reference numerals.

[0031] Throughout the specification, when a part is described as "including" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components. Furthermore, terms such as "…part," "…unit," and "module" as used in the specification refer to a unit that processes at least one function or operation, and this may be implemented in hardware, software, or a combination of hardware and software.

[0032] The device of the present disclosure is a computing device configured and connected so as to perform the operation of the present disclosure by having at least one processor execute instructions. The computing device may include one or more processors, memory for loading a computer program executed by the processor, and storage for storing the computer program and various data.

[0033] A computer program may include instructions that cause a processor to perform a method / operation according to various embodiments of the present disclosure, and may be stored on a non-transitory computer-readable storage medium. The computer program may be downloaded over a network or sold in the form of a product.

[0034] The processor can perform methods / operations according to various embodiments of the present disclosure by executing instructions. The processor controls the overall operation of each component of the computing device. The processor may be configured to include at least one of a Central Processing Unit (CPU), a Micro Processor Unit (MPU), a Micro Controller Unit (MCU), a Graphic Processing Unit (GPU), or any type of processor well known in the art of the present disclosure.

[0035] The model, neural network, or network of the present disclosure may be implemented as software / computer program executed on a computing device as an artificial intelligence model (AI model) that learns at least one task.

[0036] The medical images of the present disclosure may be images of various body parts captured by various modalities, for example, the modalities may be ultrasound, X-ray, MRI (magnetic resonance imaging), CT (computed tomography), mammography, DBT (digital breast tomosynthesis), etc. In the description, breast ultrasound images are used as examples, but are not limited thereto.

[0037] In the medical field, follow-up is performed for various purposes. For example, through follow-up, it is possible to check whether lesions grow larger or change shape over time, and to assess the potential for malignancy in lesions that were unclear as benign or malignant during the initial examination. Furthermore, follow-up allows for determining the appropriate timing for treatment and verifying the effectiveness of radiation therapy or surgery.

[0038] It is common practice for medical professionals to compare changes in lesions found in the same area within follow-up images. However, when multiple lesions are present in an image, it can be difficult to identify changes specific to each lesion. Furthermore, it can be challenging to determine whether lesions present in images from different time points are the same lesion. For example, in the case of ultrasound imaging, since the examiner acquires multiple images using an ultrasound probe from arbitrary points and directions within the examination site, it is difficult to identify whether lesions detected in follow-up images are the same lesion, located in different positions, or are new. Consequently, consistently tracking changes in lesions by comparing previous and current examinations requires significant time and effort. Additionally, because tracking lesions relies on the medical professional's subjective judgment, the consistency of the analysis results regarding lesion changes may be compromised.

[0039] In the following, the present disclosure is described in detail for identifying the same lesion in medical images acquired at different times and tracking the lesion.

[0040] FIG. 1 is a drawing illustrating a medical image analysis device according to one embodiment, FIG. 2 is a drawing illustrating a lesion location identification method according to one embodiment, and FIG. 3 is a drawing illustrating lesion location-based clustering according to one embodiment.

[0041] Referring to FIG. 1, a medical image analysis device (simply referred to as an ‘analysis device’) (100) may be a computing device that executes a computer program to provide the lesion tracking method of the present disclosure. A computer program stored on a computer-readable recording medium includes instructions that cause at least one processor to execute the present disclosure. The computing device includes a processor that executes instructions stored in memory, and the processor is implemented to execute instructions to provide the present disclosure.

[0042] The analysis device (100) can acquire medical images of patients by linking with various data systems (10) of a medical institution. The various data systems (10) of a medical institution may be, for example, PACS (Picture Archiving and Communication System), EMR / EHR (Electronic Medical Record / Electronic Health Record), etc.

[0043] Medical images may be images taken with a single modality (e.g., breast ultrasound) or images taken with various modalities (e.g., breast ultrasound and mammography). In this explanation, breast ultrasound images are used as an example.

[0044] The analysis device (100) may be implemented to provide an analysis result including lesion information through medical image analysis. The analysis device (100) may additionally acquire patient data including patient information, various test results, reports, etc., in addition to medical images, and may provide an analysis result by using the additional data together with the images. The lesion information may include the location of the lesion, the size of the lesion, the shape of the lesion, the lesion classification result (e.g., benign, malignant), the probability of malignancy (score), the characteristics of the lesion, and the classified lesion indicators, and the information to be analyzed may be determined according to the type of medical image and the body part. Here, the lesion information may be analyzed according to the criteria required in medical image analysis. For example, in the case of breast images, the lesion information may be analyzed according to the Breast Imaging Reporting and Data System (BI-RADS).

[0045] When analyzing images of a patient with an examination history, the analysis device (100) may be implemented to acquire medical images taken at different times and to provide an analysis result including information on changes in lesions through medical image analysis. In the present disclosure, medical images taken at different times are referred to as follow-up images for convenience. Here, follow-up does not refer only to examinations based on follow-up diagnosis, but includes various types of examinations such as regular examinations and additional examinations (re-examinations). Follow-up images may include, for example, images taken at time t, and images taken at time (t-1) and (t-2) which are earlier.

[0046] Meanwhile, in order to compare follow-up images, it is necessary to determine whether the lesions detected in each image are the same lesion, a different lesion, or a new lesion.

[0047] The analysis device (100) can identify the same lesion in medical images taken at different times by clustering edges in a position space based on the lesion location guided in the medical image to identify the same lesion in medical images acquired at different times. The analysis device (100) can analyze changes in imaging characteristics over time on an identified lesion basis.

[0048] The analysis device (100) can identify the anatomical location of a lesion by recognizing location identification information inserted into each image. Here, anatomical location is a term used to distinguish it from a location within the image, and refers to the location of the lesion in the patient's body. For example, the anatomical location of a lesion included in a breast ultrasound image can be extracted as the left 11 o'clock direction (Left 11H), etc.

[0049] The analysis device (100) can track changes in a specific lesion over time by analyzing follow-up observation images containing a specific lesion. The analysis device (100) can obtain lesion information regarding a specific lesion included in the corresponding image through image analysis, and track changes in the lesion by comparing the lesion information regarding the specific lesion in the follow-up observation images in chronological order. The analysis device (100) can determine whether the lesion is benign or malignant by tracking changes in the lesion. The analysis device (100) can determine the grade of the lesion by tracking changes in the lesion.

[0050] The analysis device (100) can analyze the change trend of each lesion to predict the future state of the lesion and recommend follow-up measures for the lesion. For example, if the change in lesion size is below a standard, it may recommend follow-up observation after a certain period, and if the shape of the lesion changes irregularly, it may recommend a biopsy.

[0051] Changes in the lesion may include changes in lesion size, changes in lesion shape, changes in the probability of lesion malignancy, changes in lesion characteristics, changes in lesion index values, etc. Lesion characteristics may include, for example, imaging characteristics such as ultrasound echo patterns, homogeneity within the lesion, and quantitative variable values ​​such as attenuation coefficient / sound velocity. For example, the analysis device (100) can track changes in the size of the lesion. The analysis device (100) can track changes in the shape of the lesion and track changes toward a malignant lesion based on changes such as the shape becoming irregular or the boundary becoming indistinct. The analysis device (100) can track changes such as ultrasound echo patterns and homogeneity within the lesion, and track changes toward a malignant lesion based on this.

[0052] The analysis device (100) may include an artificial intelligence-based analysis model trained to analyze medical images and infer lesion information. The analysis model may be implemented in various network structures. The training data and training method of the analysis model may be selected according to medical images and analysis indicators. The analysis model infers lesion information detected in input medical images and can extract quantitative information such as the attenuation coefficient (AC), speed of sound (SoS), effective scatterer concentration (ESC) representing density distribution within tissue, and effective scatterer diameter (ESD) representing the size of cells within tissue from ultrasound images.

[0053] The analysis device (100) can automatically generate a report based on the analysis results of medical images. The report may include information on lesions detected in the images, information on changes in lesions analyzed through comparison of follow-up images, etc. The report may include information on changes in lesions detected in past examinations up to the recent examination. Meanwhile, if lesions identified in the past are not detected in the recent examination images, a finding of "lesion not detected" may be recorded in the report.

[0054] Analysis results and reports regarding medical images may be displayed through a user interface. The analysis results may include lesion information detected in medical images acquired at a specific point in time, and information on changes in lesions in follow-up images. Lesion information may include lesion location, lesion size, lesion shape, lesion classification result (e.g., benign, malignant), malignancy probability (score), lesion characteristics, and classified lesion indices. Changes in lesions may include changes in lesion size, lesion shape, lesion malignancy probability, lesion characteristics, and lesion indices. Lesion characteristics may include imaging characteristics such as ultrasound echo patterns, homogeneity within the lesion, and quantitative variable values ​​such as attenuation coefficients / sound velocities.

[0055] The medical image analysis results analyzed by the analysis device (100) can be visually provided through a user interface.

[0056] The user interface can provide lesion information identified in follow-up images, distinguished by lesion unit. Users can select a malignant lesion to view changes in the malignant lesion over time.

[0057] The user interface can visually display the anatomical location of lesions. For example, it can display the anatomical locations of all lesions identified in follow-up images and distinguish the location of a lesion selected by the user from other lesions.

[0058] The user interface can visually display changes by lesion and display follow-up images containing the lesion, allowing the user to view the images along with the changes in the lesion. Follow-up images to be compared from the entire examination history can be freely selected. For example, if a patient is examined at intervals of 6 months or 3 months, the user can select and compare only the images taken at 6-month intervals.

[0059] Referring to FIG. 2, the analysis device (100) can identify the anatomical location of a lesion by recognizing a body marker (30A) or an examiner comment (30B) inserted into an image (20A, 20B). The body marker (30A) is location identification information inserted into the image by an imaging device. The examiner comment (30B) can be inserted into the image by an examiner input (e.g., 'Left Breast 1:00 a Rad'). The analysis device (100) can identify the location of a lesion by recognizing the lesion location comment left by the examiner through Optical Character Recognition (OCR).

[0060] The analysis device (100) can recognize at least one of a body marker or an examiner comment, and can identify the lesion location by selecting either the body marker or the examiner comment, or can identify the lesion location by combining the body marker and the examiner comment.

[0061] Even in a single examination, the examiner may photograph the same lesion multiple times. Accordingly, the analysis device (100) can extract images of the same lesion from the images taken in a single examination based on the anatomical location of the lesion included in each image, and designate a representative image of the lesion among them. The representative image can be used as a comparison target in follow-up observation. Based on the anatomical location of the lesion within the image, lesion location-based clustering can be performed.

[0062] Referring to Fig. 3, clusters of lesions #1, #2, #3, and #4 can be identified in follow-up images through clustering based on the anatomical location of the lesions within the images. At this time, images containing the lesion at the corresponding location can be mapped to each lesion cluster. For example, the lesion #2 cluster can be mapped to image #2 at time (t-2), image #3 at time (t-1), image #1 at time t, etc., containing lesion #2. Each medical image can be tagged with an identifier (e.g., #2) of the associated lesion cluster.

[0063] For lesion clustering, location-related information extracted from the image can be additionally utilized along with the anatomical location of the lesion. For example, in the case of a breast ultrasound image, the nipple can be used as a reference point, so the analysis device (100) can extract the distance and direction of the lesion from the nipple in the breast ultrasound image and specify the anatomical location of the lesion based on this. The anatomical location of the lesion included in the breast ultrasound image can be specified as the left breast, 11 o'clock direction, 2 cm from the nipple (Left 11H, 2cmN), etc. Alternatively, the lesion location estimated by body markers or examiner comments can be fine-tuned using the location information extracted from the image.

[0064] The analysis device (100) can extract morphological characteristics (size, shape, boundaries, etc.) of lesions from images and cluster similar lesions in images using the anatomical location and morphological characteristics of the lesions within the images. Based on the morphological characteristics of the lesions, mild lesions and benign lesions can be distinguished.

[0065] The analysis device (100) may assign an identifier to the lesion if a lesion at a new location is found in a recent examination, and may start tracking the newly found lesion in a subsequent examination. If the lesion detected in the recent examination is not clustered with other lesions, it may be identified as a new lesion.

[0066] Lesion clustering can be repeated whenever a recent examination image is acquired. Alternatively, if lesion clusters were generated in a previous examination, similarity to existing lesion clusters can be determined based on the location of lesions detected in the recent examination image, and identifiers of similar lesion clusters can be assigned to the lesions detected in the recent examination image.

[0067] FIGS. 4 through 6 are examples of user interfaces provided by a medical image analysis device according to one embodiment.

[0068] Referring to FIG. 4, the medical image analysis results analyzed by the analysis device (100) can be visually provided through a user interface (200). The user interface (200) can be displayed on a display device.

[0069] The user interface (200) may basically include a patient information area (210), a lesion information area (220, 230, 240), and a viewer area (250).

[0070] The patient information area (210) can display various patient information stored in the data system (10), and can display information necessary for diagnosis, such as having a family history, using hormone medication, having chest pain, having a birth history, etc.

[0071] The lesion information area (220, 230, 240) can provide lesion information analyzed by the analysis device (100) in various forms. For example, if the analysis device (100) analyzes a breast ultrasound image, it may consist of a lesion location visualization area (220), a lesion malignancy score visualization area (230), and a lesion detailed information area (240). A component (241) capable of selecting at least one lesion to generate a report may be provided.

[0072] The lesion location visualization area (220) can display the locations of lesion #1, lesion #2, lesion #3, and lesion #4 on the body part.

[0073] The lesion malignancy score visualization area (230) can display the malignancy scores of lesions #1, #2, #3, and #4 on a score bar. The higher the malignancy score, the higher the probability that the lesion is malignant, and if the malignancy score is below a threshold (e.g., 0), it may correspond to a benign lesion. The user can identify the lesion information at a glance by looking at the lesion locations displayed on the score bar.

[0074] The lesion detail information area (240) can display detailed information for each of the lesions #1, #2, #3, and #4 analyzed in the image.

[0075] Detailed information may include, for example, lesion location (e.g., Left 11H, 2cmN), lesion size (e.g., 0.42cm x 0.27cm), lesion findings (e.g., Oval, Parallel, circumscribed, hypoechoic, No posterior features), malignancy score (e.g., -99), and whether it is benign or malignant (e.g., Benign). The lesion location and lesion findings may follow, for example, the Breast Imaging Reporting Data System (BI-RADS).

[0076] Findings for benign lesions may include, for example, oval, parallel, circumscribed, hypoechoic, and no posterior features. Findings for malignant lesions may include, for example, irregular, not parallel, not circumscribed, indistinct, angular, microlobulated, and complex cystic / solid combined pattern.

[0077] The viewer area (250) can display images taken on a specific inspection date (e.g., 2025-08-01), and can be implemented to display selected images enlarged or in a separate window.

[0078] Referring to FIG. 5, when a lesion of interest is selected in the lesion information area (220, 230, 240), the user interface (200-1) can display lesion change information for the lesion of interest (e.g., #4) in the lesion tracking area (260). To make the lesion of interest easily recognizable, the lesion location visualization area (220) can display the lesion of interest separately from other lesions.

[0079] The lesion tracking area (260) can visually display information regarding changes in lesions of interest among the lesions detected in images acquired over a certain period, such as changes in size, changes in shape, changes in malignancy probability, changes in characteristics, and changes in index values. The information to be tracked in the lesion tracking area (260) can be selected in various ways.

[0080] Additionally, the lesion tracking area (260) can display a representative image and analysis results at each time point in which the lesion of interest is detected. Through this, the user can receive tracking images containing the lesion of interest simply by selecting the lesion of interest, without needing to search for the image of the lesion of interest among the tracking images, and can compare them at a glance.

[0081] Referring to FIG. 6, when a lesion of interest (e.g., #2, #3) for which a report is to be generated is selected, the user interface (200-2) can display a report containing lesion information and lesion change information for the lesion of interest. A report containing information on changes in the lesion of interest in the follow-up image is automatically generated and can be stored in a designated repository.

[0082] The user interface (200-2) can be implemented to allow modification of the automatically generated report through user input.

[0083] Through this, users can receive a report containing information on changes in the lesion of interest simply by selecting the lesion of interest, and complete the creation of the image report by verifying and saving it.

[0084] FIG. 7 is a flowchart of a method for tracking lesions in medical images according to one embodiment.

[0085] Referring to FIG. 7, the analysis device (100) acquires medical images of a patient taken at different times (S110). The analysis device (100) can acquire medical images of a patient by linking with a data system (10), such as a PACS of a medical institution.

[0086] The analysis device (100) analyzes at least one medical image taken at each time point to detect a lesion and identifies the anatomical location of the lesion by recognizing location identification information inserted in the medical image (S120). The location identification information may include, for example, body markers or examiner comments. The analysis device (100) may further use location-related information extracted from the image to specify or fine-tune the location of the lesion. For example, in the case of a breast ultrasound image using the nipple as a reference point, the analysis device (100) extracts the distance and direction of the lesion from the nipple in the breast ultrasound image and, based on this, can specify or fine-tune the anatomical location of the lesion.

[0087] The analysis device (100) generates lesion clusters by clustering the lesions detected in medical images based on the anatomical location of each lesion detected in the medical images (S130). The lesion clustering can be repeated whenever a recent examination image is acquired. Alternatively, if lesion clusters were generated in a previous examination, the similarity to the existing lesion clusters can be determined based on the location of the lesions detected in the recent examination image, and an identifier of the similar lesion cluster can be assigned to the lesions detected in the recent examination image.

[0088] The analysis device (100) identifies the lesions included in each lesion cluster as the same lesion and tracks changes in the lesion over time by comparing images taken at different times that include the same lesion (S140). The analysis device (100) can analyze changes in each lesion by comparing follow-up images that include the lesion for each identified lesion. Changes in the lesion may include changes in lesion size, changes in lesion shape, changes in the probability of malignancy, changes in lesion characteristics, changes in lesion index values, etc. Lesion characteristics may include, for example, imaging characteristics such as ultrasound echo patterns, homogeneity within the lesion, and quantitative variable values ​​such as attenuation coefficient / sound velocity. The analysis device (100) can analyze the trend of change of each lesion to predict the future state of the lesion and recommend follow-up measures for the lesion. For example, if the change in lesion size is below a standard, follow-up observation after a certain period may be recommended, and if the shape of the lesion changes irregularly, a biopsy may be recommended.

[0089] The analysis device (100) provides information on tracked lesion changes for each identified lesion to the user interface (S150).

[0090] The analysis device (100) generates a report containing change information for at least one lesion and provides it to a user interface (S160). A report including all lesions detected in medical images may be generated. Alternatively, the lesions to be included in the report may be determined by user selection, or a report may be automatically generated for lesions classified as malignant or lesions with a malignancy score above a certain threshold.

[0091] FIG. 8 is a flowchart of a method for tracking lesions in ultrasound images according to one embodiment.

[0092] Referring to FIG. 8, the analysis device (100) acquires ultrasound images of a patient taken at different times (S210).

[0093] The analysis device (100) recognizes body markers inserted into ultrasound images or comments entered by an examiner to identify the anatomical location of a lesion detected in the image (S220).

[0094] The analysis device (100) extracts follow-up images containing the same lesion from ultrasound images through lesion location-based clustering (S230).

[0095] The analysis device (100) tracks changes in the lesion over time by comparing follow-up observation images containing the same lesion (S240).

[0096] The analysis device (100) provides information on tracked lesion changes for each identified lesion to the user interface (S250).

[0097] The analysis device (100) generates a report containing change information for at least one lesion and provides it to the user interface (S260).

[0098] Some of the steps described with reference to FIGS. 7 and FIGS. 8 may be omitted or the order may be changed.

[0099] An analysis device (100) comprises at least one memory and at least one processor that executes instructions stored in at least one memory, and the processor provides the present disclosure by executing the instructions. A computer program stored on a computer-readable recording medium includes instructions that cause at least one processor to execute the present disclosure.

[0100] As such, according to the embodiment, changes in lesions over time can be automatically analyzed and changes in lesions can be consistently tracked, thereby reducing the time and cost for follow-up observation and providing reliable information to medical staff.

[0101] According to the embodiment, even in cases where it is difficult to determine the identity of a lesion based solely on the location of the lesion within an image, such as with ultrasound images, the same lesion can be identified in follow-up images through clustering based on the anatomical location of the lesion, thereby enabling the automation of changes in the lesion over time.

[0102] According to the embodiment, it can be utilized as a diagnostic assistance solution to support the decision-making of medical staff.

[0103] The embodiments of the present disclosure described above are not implemented only through devices and methods, but may also be implemented through a program that realizes a function corresponding to the configuration of the embodiments of the present disclosure or a recording medium on which such program is recorded.

[0104] Although embodiments of the present disclosure have been described in detail above, the scope of the present disclosure is not limited thereto, and various modifications and improvements by those skilled in the art using the basic concepts of the present disclosure as defined in the following claims also fall within the scope of the present disclosure.

Claims

1. A method for analyzing medical images using a medical image analysis device, A step of acquiring medical images of a patient taken at different times, A step of detecting a lesion by analyzing at least one medical image captured at each time point, and identifying the anatomical location of the lesion by recognizing location identification information inserted in the medical image, A step of generating lesion clusters by clustering the lesions detected in the medical images based on the anatomical location of each lesion detected in the medical images, and A step of identifying lesions included in each lesion cluster as identical lesions, and tracking changes in lesions over time by comparing images from different time points containing identical lesions. A medical image analysis method including 2. In Paragraph 1, The above location identification information A medical image analysis method including body markers or comments entered by an examiner.

3. In Paragraph 1, The step of identifying the anatomical location of the above lesion is A medical image analysis method that further utilizes location-related information extracted from the medical image to specify or fine-tune the anatomical location.

4. In Paragraph 1, Step of providing tracked lesion change information for each identified lesion to the user interface A medical image analysis method including further 5. In Paragraph 1, A step of generating a report containing lesion change information for at least one lesion detected in the medical images and providing it to a user interface A medical image analysis method including further 6. A method for analyzing medical images using a medical image analysis device, A step of acquiring ultrasound images of a patient taken at different times, A step of recognizing body markers inserted into the above ultrasound images or comments entered by an examiner to identify the anatomical location of a lesion detected in the corresponding image, A step of extracting follow-up images containing the same lesion from the ultrasound images through lesion location-based clustering, and A step of tracking changes in lesions over time by comparing the above tracking observation images. A medical image analysis method including 7. In Paragraph 6, The step of identifying the anatomical location of the above lesion is A medical image analysis method that further utilizes location-related information extracted from the image to specify or fine-tune the anatomical location.

8. In Paragraph 6, Step of providing tracked lesion change information for each identified lesion to the user interface A medical image analysis method including further 9. In Paragraph 6, A step of generating a report including lesion information and lesion change information for at least one lesion detected in the above ultrasound images and providing it to a user interface A medical image analysis method including further 10. At least one memory, and It includes at least one processor that executes instructions stored in at least one memory, and The above processor executes the above instructions, Acquire medical images of the patient taken at different times, and Detecting a lesion by analyzing at least one medical image captured at each time point, and identifying the anatomical location of the lesion by recognizing location identification information embedded in the medical image, and Based on the anatomical location of each lesion detected in the medical images, the lesions detected in the medical images are clustered to generate lesion clusters, and A medical image analysis device implemented to identify lesions included in each lesion cluster as the same lesion, and to track changes in lesions over time by comparing images from different time points containing the same lesion.

11. In Paragraph 10, The above processor A medical image analysis device implemented to provide information on tracked lesion changes for each identified lesion to a user interface.

12. In Paragraph 10, The above processor A medical image analysis device implemented to generate a report containing lesion information and lesion change information for at least one lesion detected in the above ultrasound images and provide it to a user interface.

13. A computer program stored on a computer-readable recording medium, A step of acquiring medical images of a patient taken at different times, A step of detecting a lesion by analyzing at least one medical image captured at each time point, and identifying the anatomical location of the lesion by recognizing location identification information inserted in the medical image, A step of generating lesion clusters by clustering the lesions detected in the medical images based on the anatomical location of each lesion detected in the medical images, and A step of identifying lesions included in each lesion cluster as identical lesions, and tracking changes in lesions over time by comparing images from different time points containing identical lesions. A computer program containing instructions that cause to execute.

14. In Paragraph 13, Step of providing tracked lesion change information for each identified lesion to the user interface A computer program that includes additional instructions to cause to execute.

15. In Paragraph 13, A step of generating a report including lesion information and lesion change information for at least one lesion detected in the above ultrasound images and providing it to a user interface A computer program that includes additional instructions to cause to execute.

Images