Method, device and storage medium for verifying medical image metal artifacts

Abstract

The application discloses a medical image metal artifact verification method and device, an electronic device and a storage medium, and the method comprises the following steps: identifying a visit image sequence with a metal artifact label, wherein the visit image sequence comprises a plurality of image sequences; obtaining a plain scan image sequence with a metal artifact label from the visit image sequence; and verifying a metal artifact image of a corresponding target image sequence in the visit image sequence based on the serial number range of the metal artifact image in the plain scan image sequence, wherein the layer thickness of the target image sequence is greater than the plain scan image sequence, and / or the target image sequence is an enhanced image sequence. The method can realize metal artifact verification on each image sequence in the visit image sequence, and provides the possibility for intelligent image auditing.

Description

Technical Field

[0001] This application belongs to the field of computer data processing technology, and specifically relates to a method, apparatus, device and storage medium for verifying metal artifacts in medical images. Background Technology

[0002] The Medical Imaging Reading System (MIRS) provides a comprehensive multi-scenario solution for medical image reading, realizing the informatization and intelligentization of the entire business process from image management to reading management. In this process, the intelligent reading platform, as the image reading tool, can intelligently manage the uploading, review, and reading of images, and track image status and reading progress in real time. It is a core component of MIRS, and a key requirement for the intelligent image review function is the verification of potential metal artifacts in the images.

[0003] The information disclosed in this background section is intended only to enhance the understanding of the overall background of this application and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0004] The purpose of this application is to provide a method for verifying metal artifacts in medical images, which solves the problem of verifying possible metal artifacts in images in intelligent image review functions.

[0005] To achieve the above objectives, this application provides a method for verifying metal artifacts in medical images, the method comprising:

[0006] Identify visit image sequences with metal artifact labels, wherein the visit image sequence includes multiple image sequences;

[0007] Obtain a sequence of plain scan images with metal artifact labels from the access image sequence;

[0008] Based on the sequence number range of the metal artifact images in the flat scan image sequence, the metal artifact images of the corresponding target image sequence in the visited image sequence are verified, wherein the layer thickness of the target image sequence is greater than that of the flat scan image sequence, and / or the target image sequence is an enhanced image sequence.

[0009] In one embodiment, when the layer thickness of the target image sequence is greater than that of the flat scan image sequence, the method specifically includes:

[0010] Calculate the layer thickness ratio between the flat scan image sequence and the target image sequence;

[0011] Based on the layer thickness ratio, the sequence range of metal artifact images in the flat scan image sequence is mapped to the target image sequence to determine the real metal artifact images in the target image sequence.

[0012] In one embodiment, the method further includes:

[0013] Non-metal artifact images that belong to the real metal artifact images in the target image sequence are identified as false negative metal artifact images.

[0014] In one embodiment, the method specifically includes:

[0015] The sequence range of metal artifact images in the flat scan image sequence is mapped to the target image sequence using a tail removal method.

[0016] In one embodiment, when the target image sequence is an enhanced image sequence, the method specifically includes:

[0017] The sequence range of metal artifact images in the flat scan image sequence is mapped to the target image sequence to determine the real metal artifact images in the target image sequence;

[0018] Metal artifact images in the target image sequence that do not belong to the real metal artifact images are identified as high-density contrast agent artifact images.

[0019] In one embodiment, verifying the metal artifact images of the corresponding target image sequence in the visited image sequence based on the sequence number range of the metal artifact images in the flat scan image sequence specifically includes:

[0020] Obtain the part information and location information of multiple image sequences in the visited image sequence;

[0021] Based on the location information, query the image sequence in the visited image sequence that has the same location as the plain scan image sequence;

[0022] Based on the location information, it is determined whether the positions of the image sequence with the same location are the same as those of the plain scan image sequence; if so,

[0023] The image sequence with the same location is confirmed to be the target image sequence in the visited image sequence.

[0024] In one embodiment, before identifying a sequence of visited images with metal artifact tags, the method further includes:

[0025] Determine whether each image slice in the plurality of image sequences includes an image slice with a CT value greater than a first preset threshold; if so,

[0026] Determine whether the number of pixels in the image slice with a CT value greater than a first preset threshold is greater than a second preset threshold; if so,

[0027] The image slice is identified as a metal artifact image, and the image sequence to which the image slice belongs and the visited image sequence are labeled with metal artifact tags.

[0028] This application also provides a device for verifying metal artifacts in medical images, comprising:

[0029] A recognition module is used to recognize a sequence of visited images with metal artifact labels, wherein the sequence of visited images includes multiple image sequences;

[0030] The acquisition module is used to acquire a sequence of flat scan images with metal artifact labels from the accessed image sequence;

[0031] The verification module is used to verify the metal artifact images of the corresponding target image sequence in the visited image sequence based on the sequence number range of the metal artifact images in the flat scan image sequence, wherein the layer thickness of the target image sequence is greater than that of the flat scan image sequence, and / or the target image sequence is an enhanced image sequence.

[0032] This application also provides an electronic device, including:

[0033] At least one processor; and

[0034] A memory that stores instructions, when executed by the at least one processor, cause the at least one processor to perform the medical image metal artifact verification method as described above.

[0035] This application also provides a machine-readable storage medium storing executable instructions that, when executed, cause the machine to perform the medical image metal artifact verification method as described above.

[0036] Compared with existing technologies, the method for verifying metal artifacts in medical images according to this application uses a plain scan image sequence with a smaller slice thickness and a metal artifact label in the visit image sequence to verify a plain scan image sequence or enhanced image sequence with a larger slice thickness. On the one hand, it can avoid misidentification of metal artifacts caused by high-density contrast agents, and on the other hand, it can avoid false negatives caused by slice thickness. This assists quality control personnel in performing more accurate quality control, saves quality control time, and improves quality control efficiency. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the medical image quality control process;

[0038] Figure 2 This is a schematic diagram illustrating the application scenario of the method for verifying metal artifacts in medical images in this application;

[0039] Figure 3 This is a flowchart of a method for verifying metal artifacts in medical images according to an embodiment of this application;

[0040] Figure 4 This is a flowchart illustrating the identification of metal artifact image sequences and access image sequences in a medical image metal artifact verification method according to an embodiment of this application.

[0041] Figure 5 This is a flowchart illustrating the process of determining the scanning period of an image sequence in a method for verifying metal artifacts in medical images according to an embodiment of this application.

[0042] Figure 6 This is a schematic diagram of debone removal and bed removal in a medical image metal artifact verification method according to an embodiment of this application;

[0043] Figure 7 This is a schematic diagram of a two-dimensional tissue region image obtained by projection in a method for verifying metal artifacts in medical images according to an embodiment of this application.

[0044] Figure 8 This is a schematic diagram illustrating the verification of metal artifacts in medical images according to an embodiment of the present application, specifically for verifying metal artifacts in a target image sequence with a slice thickness greater than that of a plain scan image sequence.

[0045] Figure 9 This is a schematic diagram illustrating the metal artifact verification of an enhanced image sequence in a method for verifying metal artifacts in medical images according to an embodiment of this application.

[0046] Figure 10 A block diagram of a medical image metal artifact verification device according to an embodiment of this application;

[0047] Figure 11 This is a hardware structure diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0048] The present application will now be described in detail with reference to the embodiments shown in the accompanying drawings. However, these embodiments do not limit the present application, and any structural, methodological, or functional modifications made by those skilled in the art based on these embodiments are included within the protection scope of the present application.

[0049] In the description, claims and above-mentioned drawings of this application, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of this application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "corresponding to" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0050] With the increasingly fierce competition in the global pharmaceutical market, the strong demand of pharmaceutical companies for cost control and efficiency improvement in R & D, production and sales has promoted the emergence and development of the pharmaceutical outsourcing industry. The Site Management Organization (SMO) in the pharmaceutical outsourcing industry is an organization that provides professional services for the R & D clinical trials of pharmaceutical companies. The main professional of SMO, the Clinical Research Coordinator (CRC), will be assigned to the clinical trial site to support daily non-clinical work under the guidance of the Principal Investigator (PI). In the services provided by SMO, medical imaging quality control (QC) is an important part.

[0051] In the R & D process of new drugs, clinical data of subjects are required as the basis for evaluating the efficacy of drugs, and only drugs that pass clinical trials can be marketed subsequently. Taking the R & D of new cancer drugs as an example, Independent Radiological Review (IRC) has been designated by the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) as the recommended test method for evaluating the efficacy of new chemotherapy drugs.

[0052] See Figure 1 In a schematic medical imaging quality control process, the medical imaging specialist of the site management organization will perform quality control on the collected medical images. If they meet the project requirements and have no quality problems, they will continue to be submitted to IRC for film reading. Otherwise, a quality control opinion will be written to remind the CRC to re-collect the data or handle the problem in other appropriate ways.

[0053] The main reason for implementing medical image quality control is that images that do not meet project requirements and quality standards restrict IRC film reading and diagnosis, which may lead to a wrong understanding of the condition and thus fail to effectively reflect the true effect of the test drug. However, the content of medical image quality control is complex, and the verification of metal artifacts is an important part of it.

[0054] Artifacts in medical images refer to abnormal images that are irrelevant to tissue structures generated during the imaging process. Among them, metal artifacts are mainly caused by objects with extremely large density differences in the human body, such as metal implants in the human body: intrauterine devices, steel plates implanted in patients, and motors implanted in patients during epilepsy electrode implantation surgery. The absorption coefficient of metal objects is usually dozens of times or more than that of most human tissues, resulting in drastic and discontinuous changes in the projection data at the interface between metal and human tissues.

[0055] The method for verifying metal artifacts in medical images provided in this application is expected to be able to verify medical images with metal artifacts such as digital breast tomosynthesis (DBT) images, computed tomography (CT) images, single-photon emission computed tomography (SPECT) images, positron emission tomography (PET) images, etc.

[0056] See Figure 2 , which shows a schematic diagram of the implementation environment provided by an exemplary embodiment of this application. This implementation environment includes a terminal and a server. Among them, data communication is carried out between the terminal and the server through a communication network. Optionally, the communication network can be a wired network or a wireless network, and this communication network can be at least one of a local area network, a metropolitan area network, and a wide area network.

[0057] The terminal can be an electronic device for providing a sequence of visit images and / or displaying quality control images, and this electronic device can be a smart phone, a tablet computer, a personal computer, etc. Figure 2 In , the computer used by medical staff is taken as an example for the terminal for illustration.

[0058] After the terminal obtains the visit image sequence, it sends the visit image sequence to the server, and the server verifies the metal artifact images in each image sequence of the visit image sequence. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery network (Content Delivery Network, CDN), and big data and artificial intelligence platforms.

[0059] Schematically, as Figure 2 shown, after the server receives the visit image sequence sent by the terminal, it inputs the visit image sequence into a device that can run the verification method for metal artifacts in medical images to obtain a verification result.

[0060] In other possible implementation manners, the device that can run the verification method for metal artifacts in medical images can also be deployed on the terminal side. The terminal directly verifies the metal artifacts in the visit image sequence and can report the verification result to the server (to avoid the server directly accessing the visit image sequence).

[0061] Refer Figure 3 , an embodiment of the verification method for metal artifacts in medical images of the present application is introduced. In this embodiment, the method includes:

[0062] S11. Identify the visit image sequence with a metal artifact label.

[0063] The verification method for metal artifacts in medical images of the present application can be for an image sequence at the visit level. The "visit image sequence" refers to all image sequences collected for a subject during a single visit image collection.

[0064] Specifically, after a subject participates in a clinical trial study for a certain drug, they need to come to the trial site regularly or according to the trial requirements to communicate with doctors (or nurses, social workers, other researchers) so that they can monitor the health status of the subject. A "visit" can be understood as a subject coming to the trial site once during the process of taking a new drug or receiving a new treatment plan. During each visit, the subject needs to undergo some medical examinations (such as medical imaging examinations) or laboratory tests (such as blood routine, urine routine), and also needs to be examined and questioned by a doctor to receive further guidance from the doctor.

[0065] During the above visits, the image collection plan is determined in advance, and all image collection and scanning are completed based on this plan. Typically, the first image collection and scanning may correspond to the "baseline visit." At certain intervals after medication, such as four weeks or six weeks later, images can be acquired again using the same image collection plan as at the baseline visit. This repeated collection of images for a specific period of time, tailored to specific needs and for the same patient, is called a visit image sequence.

[0066] In this embodiment, the visit image sequence can be saved as DICOM images (i.e., DICOM files). DICOM files are saved as follows: one medical image from a CT scan sequence is saved as one DICOM file. If an image series is acquired, such as a brain image series or a whole-body image series, it will be saved as a corresponding number of DICOM files. Here, one DICOM file refers to a single file (e.g., a file with the *.dcm extension). Each DICOM file's image data corresponds to an image slice within the medical image sequence, and multiple image sequences can correspond to one study.

[0067] From the perspective of image storage hierarchy, a visit can include multiple studies (1-M), each study can include multiple series (1-N), and each series can include multiple specific DICOM images. In this embodiment, the verification of metal artifacts can be specific to the DICOM images contained at the series level.

[0068] In this embodiment, for a sequence of visited images, if it has a metal artifact label, it means that at least one image sequence in the sequence of visited images has a metal artifact label. (Refer to reference...) Figure 4 The following describes an example of an image sequence being labeled with a metal artifact tag.

[0069] S111. Determine whether each image slice in multiple image sequences includes an image slice with a CT value greater than a first set threshold.

[0070] For image slices, the CT value of normal tissue ranges from -1000 to 900, while the CT value of metal is higher than that of normal tissue. By setting a reasonable first threshold, metal artifacts in image slices can be identified.

[0071] For example, each image sequence can correspond to a series, and the first set threshold can be set to 3000 HU. That is, if an image slice is detected to contain pixels with more than 3000 HU, the image slice is considered to be a candidate image slice containing metal artifacts.

[0072] S112. Determine whether the number of pixels in the image slice whose CT value is greater than the first set threshold is greater than the second set threshold.

[0073] When an image slice in an image sequence has a CT value greater than a first preset threshold, it is further determined whether the number of pixels with a CT value greater than the first preset threshold is greater than a second preset threshold, in order to confirm whether there are metal artifacts in the aforementioned undetermined image slice. The second preset threshold can be set based on common metals implanted in the human body in the above embodiments.

[0074] S113. Identify the image slice as a metal artifact image and label the image sequence and the visited image sequence to which the image slice belongs with a metal artifact label.

[0075] For an image sequence, if it includes only one image slice with metal artifacts, the image sequence will be labeled with a metal artifact label, and the corresponding visited image sequence will also be labeled with a metal artifact label.

[0076] From the perspective of the image storage hierarchy, the above process can be to detect metal artifact images according to series, and to label each series of detected metal artifact images with a metal artifact label; finally, the series is restored into a visit-level sequence of visit images.

[0077] Correspondingly, in this embodiment, after receiving the accessed image sequence, it can also be parsed into study-level and series-level images based on the DICOM standard protocol. Exemplarily, the parsing can be based on libraries in computer languages ​​such as JAVA or PYTHON used for parsing the DICOM standard.

[0078] In some alternative embodiments, the identification of metal artifacts may be performed, for example, using a deep learning method. In such embodiments, the metal artifact verification method of this application can be regarded as a post-processing of the prior deep learning identification of metal artifacts, correcting for identification errors or omissions of metal artifacts.

[0079] S12. Obtain a sequence of flat scan images with metal artifact labels from the access image sequence.

[0080] Image sequences can be categorized into two types: plain CT scans and enhanced CT scans. Plain CT scans, also known as standard scans, refer to scans performed intravenously without iodine-containing contrast agents. Under plain CT scans, the absorption capacity of blood vessels, soft tissues, and common organs such as the liver, kidneys, and spleen is relatively similar. According to the principles of CT imaging, these tissues or organs have similar grayscale values. Therefore, plain CT scans do not provide good contrast for observing lesions.

[0081] Enhanced CT scans involve injecting a contrast agent (also called a contrast agent, a liquid that has a stronger ability to absorb X-rays than human tissue) intravenously. During the examination, the contrast agent circulates through the bloodstream, entering arteries, veins, hepatic artery, hepatic vein, renal artery, renal vein, and eventually the ureters. Because cancerous or diseased tissues have a very active blood supply, the contrast agent follows the blood circulation to reach the lesion area. Therefore, the use of contrast agents enhances the contrast of the lesion, making it easier to accurately observe the location and size of the lesion on CT images.

[0082] In contrast-enhanced CT scans, the scanning phase corresponds to the period when the contrast agent reaches different parts of the body, facilitating the observation of lesions in different locations. For example, in the arterial phase of the scanning phase, the contrast agent fills the arterial system, enhancing arterial vessels and organs with rich blood supply; in the venous phase, which is later than the arterial phase, the portal vein is enhanced by the contrast agent, making it very useful for observing lesions in the liver; in the equilibrium phase, the contrast agent redistributes to the portal vein, inferior vena cava, and abdominal aorta, allowing the filling of the contrast agent to be seen in both the venous and arterial systems; in the delayed phase, the contrast agent in the blood vessels decreases, but if there are highly vascularized tissues such as tumors, the contrast agent decays more slowly, making it beneficial for observing these highly vascularized tissues.

[0083] Plain CT scans and enhanced CT scans are widely used in various examination scenarios, and the identification of different scan phases in enhanced CT scans is crucial for selecting the appropriate timing to observe lesions in specific locations. In some embodiments of this application, it is at least desirable that each image sequence in the visit image sequence be distinguished as either a plain CT scan or an enhanced CT scan. Furthermore, in some embodiments, it is also desirable that the enhanced CT scan image sequences in the visit image sequence be distinguished as specific scan phases.

[0084] Reference Figure 5 As an example, the recognition of plain or enhanced CT scans can be achieved using machine learning methods.

[0085] S121. Segment the scanned tissue region images from each image in the image sequence.

[0086] In some application scenarios, considering the limitations of computing resources, each image in the image sequence can be resampled to a preset scanning interval in the scanning direction before segmenting the scanned tissue region image.

[0087] For example, if the spacing between images in the original image sequence is less than 5 mm, then 5 mm can be used as the preset scanning spacing to resample each image. Exemplarily, if the spacing between images in the original image sequence is 2.5 mm, then an average can be performed on every two adjacent image layers.

[0088] Understandably, in scenarios with ample computing resources, resampling of each image in the image sequence may be optional.

[0089] When using information from each image in an image sequence as a basis for recognition, it can be based on a variety of different image preprocessing methods and corresponding to training samples of different machine learning models.

[0090] For example, multiple images can be directly used as input to a machine learning model, in which case the machine learning model should also be trained based on multiple image slices from multiple sequences. Alternatively, multiple images can be projected into two-dimensional tissue region images and then used as input to a machine learning model, in which case the machine learning model should also be trained based on the two-dimensional tissue region images obtained from the projection of multiple image sequences.

[0091] Reference Figure 6 In this embodiment, two-dimensional tissue region images obtained by projecting each image in the image sequence can be selected for identification during the scanning period. Correspondingly, since the projection operation is performed along the column direction of the image, in order to enhance the projection results of soft tissue and contrast agent, other interfering factors with relatively large gray values ​​in the projection direction, such as bone and bed, can be suppressed, thereby achieving the ideal projection effect of improving the contrast of soft tissue and contrast agent.

[0092] As an example, each image in the image sequence can be input into a UNet network, a modified version of UNet called DOUBLE-UNET, NESTED-UNET, or UNET++ to obtain de-osteoscopic tissue images. Next, each de-osteoscopic tissue image is binarized to obtain a binarized image. Finally, the non-maximum connected regions in the binarized image are removed to obtain the scanned tissue region image; here, the non-maximum connected regions removed are primarily the bed image region.

[0093] The operation of removing the image region of the bone bed from the deostomized tissue image in the above embodiments can be understood as a morphological operation. It is understood that, in alternative embodiments, morphological operations can be performed on each image in the image sequence to remove the image region of the bone bed first, and then the image region of the bone can be removed, to obtain the same scanned tissue region image.

[0094] S122. Resample each scanned tissue region image at the same resolution in each direction to obtain a projection reference image.

[0095] For image sequences, the resolution of each image slice is usually not consistent in the scanning direction (i.e., the height direction) and inside the slice. Direct projection will cause the projected image to be distorted in scale, and will also affect the accuracy of the subsequent neural network in understanding image information.

[0096] For example, each pixel in an image slice is 10mm in the height direction, but each pixel is 1mm in the left-right and front-back directions within the slice. Thus, in a 3D image, the human body is effectively compressed in the height direction, and this compression also applies when projected into a 2D image. Therefore, it is necessary to resample each scanned tissue region image at the same resolution in all directions to ensure that each scanned tissue region image has the same resolution in all directions.

[0097] As an example, if the original resolution of an image slice in the three directions (height, left-right, and front-back) is [5mm, 0.5mm, 0.5mm], the resolution after resampling can be set to [1mm, 1mm, 1mm]. In this case, resampling in the height direction is equivalent to upsampling (requiring a denser image slice), while in the left-right or front-back directions it is downsampling (averaging out overly dense image slices).

[0098] S123. Project based on each projection reference image to obtain a two-dimensional tissue region image.

[0099] Reference Figure 7 The two-dimensional tissue region image obtained by projection here can be a coronal image, that is, a projection along the height direction of the image sequence. A specific projection algorithm can be, for example, the Maximum Intensity Projection (MaxIP) algorithm. The MaxIP algorithm uses the maximum value of samples along the viewing direction of the scan sequence as the pixel value at the corresponding position in the drawn image, thus enabling visualization of high-grayscale structures in volumetric data.

[0100] In an alternative embodiment, depending on the application scenario, the two-dimensional tissue region image can also be obtained, for example, based on the Average Intensity Projection (AvgIP) algorithm, the Median Intensity Projection (MedIP) algorithm, etc.

[0101] S124. Based on two-dimensional tissue region images and machine learning models, determine the scanning period of the image sequence.

[0102] In various embodiments of this application, the machine learning model can be trained and constructed using various convolutional neural networks. Exemplarily, a typical network structure includes a series of interconnected layers: input layer - convolutional layer - pooling layer - convolutional layer - pooling layer - fully connected layer - classifier. The input layer, convolutional layer, and the final classifier are general structures similar to neural networks, and the architecture of the intermediate layers can be adaptively adjusted according to the specific use case, i.e., the classification effect.

[0103] During the training of a machine learning model, image samples with areas such as bones and beds removed are input, and observable plain and enhanced features are highlighted. The neural network obtains these features through iterative parameter processing of convolutional layers and outputs the corresponding classification. In this process, a mapping relationship from image to classification is generated. A large amount of training data will make this mapping relationship have high accuracy and general applicability, thus providing reliable prediction accuracy on new image data.

[0104] Based on the above machine learning methods, each image sequence in the access image sequence can be labeled to represent the scanning period. Combined with the metal artifact label, the desired flat scan image sequence with metal artifact label can be located in this step.

[0105] S13. Based on the sequence number range of the metal artifact images in the flat scan image sequence, verify the metal artifact images of the corresponding target image sequence in the visited image sequence.

[0106] For different image sequences, various factors can lead to errors in the identification of metal artifacts. For example, in contrast-enhanced CT scan sequences, high-density contrast agents may form "metal-like artifact features" in the image, which may be misidentified as metal artifacts. In image sequences with large slice thickness, the features of metal artifacts may not reach the detection threshold due to information averaging, resulting in false negatives. The metal-like artifact features formed by high-density contrast agents may have partial similarities or identicalities to actual metal artifacts in image appearance. In some embodiments of this application, it is desirable to distinguish these image sequences with metal-like artifact features formed by high-density contrast agents from image sequences containing actual metal artifacts.

[0107] The embodiments of this application propose to use a plain scan image sequence with a relatively small slice thickness as a benchmark to verify the metal artifact images of the target image sequence and / or enhanced image sequence with a slice thickness greater than that in the access image sequence. On the one hand, this can avoid the misidentification of metal artifacts caused by high-density contrast agents, and on the other hand, it can avoid false negatives caused by slice thickness.

[0108] It should be noted that "correspondence" here means that the target image sequence and the plain scan image sequence are consistent in terms of part and position.

[0109] Specifically, the system can first obtain the part information and position information of multiple image sequences in the visited image sequence, and then, based on the part information, query the image sequences in the visited image sequence that have the same part as the plain scan image sequence; then, based on the position information, confirm whether the image sequence with the same part has the same position as the plain scan image sequence; if so, confirm that the image sequence with the same part is the target image sequence in the visited image sequence.

[0110] In this embodiment, the site information can be obtained through machine learning methods. For example, each image in the image sequence can be projected into a two-dimensional tissue region image. Depending on the different categories of the scanned sites output, a neural network model or a modified neural network model can be used for identification.

[0111] As an example, if the goal is to identify the scanned body parts "head," "neck," "chest," "abdomen," and "pelvis," a trained neural network model can receive image input and output using its built-in 5-class classifier. Of course, it's understandable that if more types of scanned body parts need to be identified, the neural network model can be modified and a corresponding number of classifiers added; this will not be elaborated upon here.

[0112] In such location information, the level of detail in location classification may lead to two not-so-corresponding image sequences being misidentified as "corresponding to each other." For example, image sequence A includes scanned image slices starting from the patient's "nose" and ending at the chest; image sequence B includes scanned image slices starting from the patient's "brain" and ending at the chest. However, regardless of whether it starts from the "nose" or the "brain," the location information could be "head," meaning that the location information for both image sequence A and image sequence B is "head, neck, and chest."

[0113] To address the above challenges, this embodiment uses location information for further confirmation. The location information can be obtained from the two tags "image orientation patient" and "image position patient" in the image sequence.

[0114] The header data of a DICOM file includes multiple data elements, and each data element contains four items: Tag, Value Representation (VR), Value Length (VL), and Value Field (VF). For example, the tag "Modality" represents the modality, its value representation is "CS", the value length is "2", and the value can be CT, MR, etc. Correspondingly, by comparing the values of the two tags "image orientation patient" and "image position patient", it can be determined that the positions and locations of the target image sequence and the non-contrast image sequence are consistent.

[0115] As previously described in this application, in the hierarchical structure of image storage, a visit can include multiple studies, and each study can include multiple series. In some cases, the location information and position information between multiple series under the same study are usually consistent, while there may be inconsistent position information as described above between multiple studies under the same visit. Therefore, in some embodiments, the step of confirming the corresponding target image sequence based on the location information and position information may be only for the series between different studies; for the series between the same study, the confirmation of the above position information may not be performed either.

[0116] When the slice thickness of the target image sequence is greater than that of the non-contrast image sequence and the target image sequence is a contrast-enhanced image sequence, this embodiment provides a corresponding verification method.

[0117] ① The slice thickness of the target image sequence is greater than that of the non-contrast image sequence

[0118] Calculate the slice thickness ratio of the non-contrast image sequence and the target image sequence, and based on the slice thickness ratio, map the serial number range of the metal artifact images in the non-contrast image sequence to the target image sequence to determine the true metal artifact images of the target image sequence; further, the non-metal artifact images belonging to the true metal artifact images in the target image sequence can also be determined as false negative metal artifact images.

[0119] See Figure 8, taking the slice thickness of the non-contrast image sequence as 1 mm and the slice thickness of the target image sequence as 4 mm as an example, metal artifacts are detected in the non-contrast image sequence with a slice thickness of 1 mm, but no metal artifacts are detected in the target image sequence with a slice thickness of 4 mm. At this time, the slice thickness ratio is 1 / 4, and the serial number range "255 - 257" of the metal artifact images in the non-contrast images will be proportionally mapped to the serial numbers "63.75, 64, 64.25".

[0120] In such a case, the serial numbers of the metal artifact images corresponding to "63.75, 64, 64.25" can be determined based on different strategies. In this embodiment, the floor function can be used to map the serial number range of the metal artifact images in the non-contrast image sequence to the target image sequence to prevent the serial number range from exceeding the range of the 4-mm slice thickness image sequence. Correspondingly, the serial numbers "63.75, 64, 64.25" are mapped to the serial numbers "63 - 64", that is, the images within the serial number range "63 - 64" in the target image sequence are identified as real metal artifact images. Here, since the images within the serial number range "63 - 64" in the target image sequence are non-metal artifact images before verification, these images are also determined to be false negative metal artifact images.

[0121] Correspondingly, if the image with serial number 63 has been identified as a metal artifact image before verification, only the image with serial number 64 will be determined to be a false negative metal artifact image at this time.

[0122] It can be understood that in some alternative embodiments, the round-off method, ceiling function, etc. can also be used to perform the above serial number range mapping, which will not be elaborated here.

[0123] ② The target image sequence is an enhanced image sequence

[0124] Map the serial number range of the metal artifact images in the non-contrast image sequence to the target image sequence to determine the real metal artifact images in the target image sequence; and determine the metal artifact images in the target image sequence that do not belong to the real metal artifact images as high-density contrast agent artifact images.

[0125] Refer Figure 9 , serial number 202 is the arterial phase, corresponding to the enhanced image sequence (target image sequence), and the serial number range "255 - 290" therein is metal artifact images. Serial number 201 is the non-contrast image sequence of the same part and position, and the serial number range "255 - 257" therein is metal artifact images. Through the mapping of the serial number range, it can be determined that the serial number range "255 - 257" in the target image sequence with serial number 202 is real metal artifact images, and the serial number range "258 - 290" is high-density contrast agent artifact images.

[0126] Similarly, in such a case, if the slice thickness of the target image sequence is also greater than that of the non-contrast image sequence, the slice thickness ratio can be calculated accordingly and then the mapping of the serial number range can be performed, which will not be elaborated here.

[0127] See Figure 10 , which introduces an embodiment of the verification device for metal artifacts in medical images of the present application. In this embodiment, the processing device for the medical image sequence includes an identification module 21, an acquisition module 22, and a verification module 23.

[0128] The identification module 21 is used to identify the visit image sequence with metal artifact labels, where the visit image sequence includes multiple image sequences; the acquisition module 22 is used to acquire the non-contrast image sequence with metal artifact labels from the visit image sequence; the verification module 23 is used to verify the metal artifact images of the corresponding target image sequence in the visit image sequence based on the serial number range of the metal artifact images in the non-contrast image sequence, where the slice thickness of the target image sequence is greater than that of the non-contrast image sequence, and / or the target image sequence is an enhanced image sequence.

[0129] In one embodiment, when the slice thickness of the target image sequence is greater than that of the non-contrast image sequence, the verification module 23 is specifically used to calculate the slice thickness ratio of the non-contrast image sequence and the target image sequence; based on the slice thickness ratio, map the serial number range of the metal artifact images in the non-contrast image sequence to the target image sequence to determine the true metal artifact images of the target image sequence.

[0130] In one embodiment, the verification module 23 is further used to determine the non-metal artifact images that do not belong to the true metal artifact images in the target image sequence as false negative metal artifact images.

[0131] In one embodiment, the verification module 23 is further used to map the serial number range of the metal artifact images in the non-contrast image sequence to the target image sequence by using the truncation method.

[0132] In one embodiment, when the target image sequence is an enhanced image sequence, the verification module 23 is specifically used to map the serial number range of the metal artifact images in the non-contrast image sequence to the target image sequence to determine the true metal artifact images of the target image sequence; determine the metal artifact images that do not belong to the true metal artifact images in the target image sequence as high-density contrast agent artifact images.

[0133] In one embodiment, the verification module 23 is specifically used to obtain the part information and position information of multiple image sequences in the visit image sequence; based on the part information, query the image sequence in the visit image sequence that has the same part as the non-contrast image sequence; based on the position information, confirm whether the position of the image sequence with the same part is the same as that of the non-contrast image sequence; if so, confirm that the image sequence with the same part is the corresponding target image sequence in the visit image sequence.

[0134] In one embodiment, the system further includes a determination module 24, which determines whether each image slice of the multiple image sequences includes an image slice with a CT value greater than a first set threshold; if so, determines whether the number of pixels in the image slice with a CT value greater than the first set threshold is greater than a second set threshold; if so, identifies the image slice as a metal artifact image and labels the image sequence to which the image slice belongs and the visited image sequence with a metal artifact label.

[0135] As per the above reference Figures 1 to 9 This specification describes a method for verifying metal artifacts in medical images according to embodiments thereof. The details mentioned in the above description of the method embodiments also apply to the apparatus for verifying metal artifacts in medical images according to embodiments thereof. The above-described apparatus for verifying metal artifacts in medical images can be implemented in hardware, software, or a combination of hardware and software.

[0136] Figure 11 A hardware structure diagram of an electronic device according to an embodiment of this specification is shown. Figure 11 As shown, the electronic device 30 may include at least one processor 31, a memory 32 (e.g., non-volatile memory), a memory 33, and a communication interface 34, and the at least one processor 31, memory 32, memory 33, and communication interface 34 are connected together via an internal bus 35. The at least one processor 31 executes at least one computer-readable instruction stored or encoded in the memory 32.

[0137] It should be understood that the computer-executable instructions stored in memory 32, when executed, cause at least one processor 31 to perform the above-described combinations in the various embodiments of this specification. Figures 1 to 9 The description includes various operations and functions.

[0138] In the embodiments of this specification, electronic device 30 may include, but is not limited to: personal computer, server computer, workstation, desktop computer, laptop computer, notebook computer, mobile electronic device, smartphone, tablet computer, cellular phone, personal digital assistant (PDA), handheld device, messaging device, wearable electronic device, consumer electronic device, etc.

[0139] According to one embodiment, a program product, such as a machine-readable medium, is provided. The machine-readable medium may have instructions (i.e., the elements implemented in software as described above), which, when executed by a machine, cause the machine to perform the above-described combinations of the various embodiments of this specification. Figures 1-3The various operations and functions described. Specifically, a system or apparatus equipped with a readable storage medium storing software program code that implements the functions of any of the embodiments described above, and enabling the computer or processor of the system or apparatus to read and execute the instructions stored in the readable storage medium.

[0140] In this case, the program code read from the readable medium itself can perform the functions of any of the above embodiments, and therefore the machine-readable code and the readable storage medium storing the machine-readable code constitute a part of this specification.

[0141] Examples of readable storage media include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD-RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer or the cloud via a communication network.

[0142] Those skilled in the art will understand that the various embodiments disclosed above can be modified and varied without departing from the spirit of the invention. Therefore, the scope of protection of this specification should be defined by the appended claims.

[0143] It should be noted that not all steps and units in the above process and system structure diagrams are mandatory; some steps or units can be omitted according to actual needs. The execution order of each step is not fixed and can be determined as needed. The device structure described in the above embodiments can be a physical structure or a logical structure. That is, some units may be implemented by the same physical client, or some units may be implemented by multiple physical clients, or they may be jointly implemented by certain components in multiple independent devices.

[0144] In the above embodiments, the hardware units or modules can be implemented mechanically or electrically. For example, a hardware unit, module, or processor may include permanent dedicated circuitry or logic (such as a dedicated processor, FPGA, or ASIC) to perform the corresponding operation. The hardware unit or processor may also include programmable logic or circuitry (such as a general-purpose processor or other programmable processor), which can be temporarily configured by software to perform the corresponding operation. The specific implementation method (mechanical, dedicated permanent circuitry, or temporarily configured circuitry) can be determined based on cost and time considerations.

[0145] The specific embodiments described above with reference to the accompanying drawings are exemplary embodiments, but do not represent all embodiments that can be implemented or fall within the scope of the claims. The term "exemplary" as used throughout this specification means "serving as an example, instance, or illustration" and does not imply that it is "preferred" or "advantageous" compared to other embodiments. Specific details are included to provide an understanding of the described techniques. However, these techniques can be practiced without these specific details. In some instances, well-known structures and apparatuses are shown in block diagram form to avoid obscuring the concepts of the described embodiments.

[0146] The foregoing description of this disclosure is provided to enable any person skilled in the art to implement or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles applicable herein can be applied to other variations without departing from the scope of this disclosure. Therefore, this disclosure is not limited to the examples and designs described herein, but is consistent with the widest scope of the principles and novel features disclosed herein.

Claims

1. A method for verifying metal artifacts in medical images, characterized in that, The method includes: Identify visit image sequences with metal artifact labels, wherein the visit image sequence includes multiple image sequences; Obtain a sequence of plain scan images with metal artifact labels from the access image sequence; Based on the sequence range of metal artifact images in the flat scan image sequence, the metal artifact images of the corresponding target image sequence in the visited image sequence are verified, wherein the layer thickness of the target image sequence is greater than that of the flat scan image sequence, and / or the target image sequence is an enhanced image sequence. When the layer thickness of the target image sequence is greater than that of the flat scan image sequence, the method specifically includes: calculating the layer thickness ratio between the flat scan image sequence and the target image sequence; based on the layer thickness ratio, mapping the sequence range of metal artifact images in the flat scan image sequence to the target image sequence, and identifying non-metal artifact images in the target image sequence that belong to real metal artifact images as false negative metal artifact images; When the target image sequence is an enhanced image sequence, the method specifically includes: mapping the sequence range of metal artifact images in the plain scan image sequence to the target image sequence to determine the real metal artifact images in the target image sequence; and determining the metal artifact images in the target image sequence that do not belong to the real metal artifact images as high-density contrast agent artifact images.

2. The method for verifying metal artifacts in medical images according to claim 1, characterized in that, The method specifically includes: The sequence range of metal artifact images in the flat scan image sequence is mapped to the target image sequence using a tail removal method.

3. The method for verifying metal artifacts in medical images according to any one of claims 1 to 2, characterized in that, Based on the sequence number range of metal artifact images in the scanned image sequence, the metal artifact images of the corresponding target image sequence in the visited image sequence are verified, specifically including: Obtain the part information and location information of multiple image sequences in the visited image sequence; Based on the location information, query the image sequence in the visited image sequence that has the same location as the plain scan image sequence; Based on the location information, it is determined whether the positions of the image sequence with the same location are the same as those of the plain scan image sequence; if so, The image sequence with the same location is confirmed to be the target image sequence in the visited image sequence.

4. The method for verifying metal artifacts in medical images according to any one of claims 1 to 2, characterized in that, Before identifying access image sequences with metal artifact labels, the method further includes: Determine whether each image slice in the plurality of image sequences includes an image slice with a CT value greater than a first preset threshold; if so, Determine whether the number of pixels in the image slice with a CT value greater than a first preset threshold is greater than a second preset threshold; if so, The image slice is identified as a metal artifact image, and the image sequence to which the image slice belongs and the visited image sequence are labeled with metal artifact tags.

5. A device for verifying metal artifacts in medical images, characterized in that, include: A recognition module is used to recognize a sequence of visited images with metal artifact labels, wherein the sequence of visited images includes multiple image sequences; The acquisition module is used to acquire a sequence of flat scan images with metal artifact labels from the accessed image sequence; The verification module is used to verify the metal artifact images of the corresponding target image sequence in the visited image sequence based on the sequence number range of the metal artifact images in the flat scan image sequence, wherein the layer thickness of the target image sequence is greater than that of the flat scan image sequence, and / or the target image sequence is an enhanced image sequence. When the layer thickness of the target image sequence is greater than that of the flat scan image sequence, the verification module is specifically used to: calculate the layer thickness ratio between the flat scan image sequence and the target image sequence; based on the layer thickness ratio, map the sequence range of metal artifact images in the flat scan image sequence to the target image sequence, and determine the non-metal artifact images in the target image sequence that belong to real metal artifact images as false negative metal artifact images; When the target image sequence is an enhanced image sequence, the verification module is specifically used to: map the sequence number range of metal artifact images in the flat scan image sequence to the target image sequence to determine the real metal artifact images in the target image sequence; and determine the metal artifact images in the target image sequence that do not belong to the real metal artifact images as high-density contrast agent artifact images.

6. An electronic device, comprising: At least one processor; as well as A memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform the medical image metal artifact verification method as described in any one of claims 1 to 4.

7. A machine-readable storage medium storing executable instructions that, when executed, cause the machine to perform the method for verifying metal artifacts in medical images as described in any one of claims 1 to 4.

Images