An image processing method, device and readable storage medium

Abstract

The specification discloses an image processing method, device and readable storage medium. From each image included in a medical image sequence, a plurality of candidate images are selected according to a preset rule, and a positioning image is determined according to a difference between the remaining images in the medical image sequence except the candidate images and the candidate images, so as to eliminate the positioning image from the medical image sequence, and determine a re-determined medical image sequence according to the medical image sequence after the positioning image is eliminated. It can be seen that, by eliminating the positioning image from the medical image sequence, the interference information included in the re-determined medical image sequence is reduced, and the accuracy of a clinical index of a patient obtained based on the re-determined medical image sequence is improved.

Description

Technical Field

[0001] This specification relates to the field of computer technology, and in particular to an image processing method, apparatus, and readable storage medium. Background Technology

[0002] With the development of computer-aided medical diagnostic technology, multiple medical images obtained from scanning a patient can now be arranged in the order of acquisition to form a medical image sequence. This sequence can then be used to obtain the patient's clinical indicators, thereby assisting doctors in making diagnoses.

[0003] However, medical image sequences often contain errors, leading to low accuracy in obtaining clinical indicators for patients based on these erroneous sequences.

[0004] Based on this, this specification provides an image processing method. Summary of the Invention

[0005] This specification provides an image processing method, apparatus, device, and readable storage medium to partially solve the aforementioned problems existing in the prior art.

[0006] The following technical solution is adopted in this specification:

[0007] This specification provides an image processing method, including:

[0008] Acquire a medical image sequence to be identified, the medical image sequence comprising multiple images of body parts acquired in a specified order;

[0009] Select several candidate images from the medical image sequence according to preset rules;

[0010] The localization image is determined based on the differences between the remaining images in the medical image sequence (excluding the candidate image) and the candidate image.

[0011] The localization image is removed from the medical image sequence to obtain a newly determined medical image sequence.

[0012] Optionally, several candidate images are selected from the medical image sequence according to preset rules, specifically including:

[0013] Obtain the sequence number of each image in the medical image sequence;

[0014] Several images whose serial numbers are greater than a first preset threshold and / or whose serial numbers are less than a second preset threshold are selected as candidate images.

[0015] Optionally, the localization image is determined based on the differences between the remaining images in the medical image sequence (excluding the candidate image) and the candidate image, specifically including:

[0016] Images in the medical image sequence whose sequence number is less than the first preset threshold and whose sequence number is greater than the second preset threshold are selected as the remaining images other than the candidate images.

[0017] Select a reference image from the remaining images;

[0018] For each candidate image, determine the angle between the scanning direction of the candidate image and the scanning direction of the reference image;

[0019] Candidate images with an included angle greater than a preset angle threshold are used as positioning images.

[0020] Optionally, the method further includes:

[0021] In response to the recognition request, the redefined medical image sequence is input into the pre-trained body part recognition model to obtain the patient's body part recognition results.

[0022] Optionally, before inputting the redefined medical image sequence into the pre-trained body part recognition model in response to a recognition request, the method further includes:

[0023] Based on the scanning position and scanning direction of each image contained in the redefined medical image sequence, the medical image sequence is split into sub-sequences.

[0024] Optionally, in response to the recognition request, the redefined medical image sequence is input into a pre-trained body part recognition model to obtain the recognition results of the patient's body parts, specifically including:

[0025] For each subsequence, the images contained in the subsequence are used as input to a pre-trained body part recognition model to obtain the body part corresponding to the subsequence as the recognition result of the subsequence;

[0026] Based on the recognition results corresponding to each sub-sequence, the recognition results of the patient's body parts are determined.

[0027] Optionally, the medical image sequence is split into sub-sequences, specifically including:

[0028] For each image in the medical image sequence, determine whether the image meets the following conditions:

[0029] The scan position of this image is different from the scan position of the next image.

[0030] Furthermore, the scanning direction of this image is the same as the scanning direction of the next image;

[0031] If all conditions are met, then the image is determined to be in the same subsequence as the next image.

[0032] If at least one condition is not met, then it is determined that the image is not in the same subsequence as the next image.

[0033] Optionally, the method further includes:

[0034] In response to an update of the parameters of the body part recognition model, a specified medical image sequence that has undergone recognition based on the body part recognition model before the update is acquired;

[0035] Based on the updated body part recognition model, the specified medical image sequence is recognized to obtain the specified recognition result;

[0036] Update the recognition result of the specified medical image sequence based on the specified recognition result.

[0037] Optionally, a specified medical image sequence that has undergone recognition based on the previous body part recognition model can be obtained, specifically including:

[0038] Determine the update cycle for the recognition results;

[0039] According to the update cycle, a preset number of the specified medical image sequences are obtained.

[0040] This specification provides an image processing apparatus, comprising:

[0041] A sequence determination module is used to acquire a medical image sequence to be identified, the medical image sequence including multiple images of body parts acquired in a specified order;

[0042] The selection module is used to select several candidate images from the medical image sequence according to preset rules;

[0043] The localization image determination module is used to determine the localization image based on the differences between the remaining images in the medical image sequence (excluding the candidate image) and the candidate image.

[0044] The removal module is used to remove the positioning image from the medical image sequence to obtain a redefined medical image sequence.

[0045] This specification provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described image processing method.

[0046] This specification provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described image processing method.

[0047] The above-mentioned technical solutions adopted in this specification can achieve the following beneficial effects:

[0048] The image processing method provided in this specification involves selecting several candidate images from the images included in a medical image sequence according to preset rules. Then, based on the differences between the candidate images and the remaining images in the medical image sequence, a localization image is determined and removed from the medical image sequence. The result is a newly determined medical image sequence after removing the localization image. Therefore, by removing the localization image from the medical image sequence, the interference information contained in the newly determined medical image sequence is reduced, thus improving the accuracy of the patient's clinical indicators obtained based on the newly determined medical image sequence. Attached Figure Description

[0049] The accompanying drawings, which are included to provide a further understanding of this specification and form part of this specification, illustrate exemplary embodiments and their descriptions, serving to explain this specification and do not constitute an undue limitation thereof.

[0050] In the picture:

[0051] Figure 1 This is a flowchart illustrating one image processing method described in this specification.

[0052] Figure 2 This is a flowchart illustrating one image processing method described in this specification.

[0053] Figure 3 This is a flowchart illustrating one image processing method described in this specification.

[0054] Figure 4 This is a schematic diagram of an image processing apparatus provided in this specification;

[0055] Figure 5 The corresponding information provided in this specification Figure 1 A schematic diagram of an electronic device. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this specification clearer, the technical solutions of this specification will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of them. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this specification.

[0057] Additionally, it should be noted that all actions involving the acquisition of signals, information, or data in this invention are carried out in compliance with the relevant data protection laws and regulations of the country where the invention is located, and with authorization from the owner of the corresponding device.

[0058] The technical solutions provided in the various embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0059] Figure 1 This is a flowchart illustrating an image processing method provided in this specification.

[0060] S100: Obtain a medical image sequence to be identified, the medical image sequence including multiple images of body parts acquired in a specified order.

[0061] This specification provides an image processing method in its embodiments. The image processing method can be executed by an electronic device such as a server used to identify body parts of a patient in an image.

[0062] In the embodiments of this specification, medical image sequences can be extracted from the Digital Imaging and Communications in Medicine (DICOM) file to be identified.

[0063] Optionally, in the embodiments of this specification, the patient's medical image sequence is taken as originating from a DICOM file. However, the medical image sequence to be identified can also originate from a volume data sequence. Furthermore, the medical image sequence and the volume data sequence in the DICOM file are identical in content, and the DICOM sequence can be converted into a volume data sequence.

[0064] Optionally, the DICOM file to be identified can be obtained from a Picture Archiving and Communication Systems (PACS) or in real time from a medical image acquisition device; this manual does not limit this to either method.

[0065] The medical image sequence in the medical digital imaging and communication files to be identified can be an image sequence obtained by computed tomography (CT), an image sequence obtained by magnetic resonance imaging (MRI), an image sequence obtained by low-dose positron emission tomography / magnetic resonance imaging (PET), or other modal image sequences. This specification does not specifically limit the modality of the medical image sequence to be identified. For ease of description, this specification uses a medical image sequence to be identified originating from a CT scan as an example to elaborate on the specific technical solution.

[0066] In practical applications, a medical image sequence contains multiple images obtained by scanning a patient's body parts. The order in which the images are arranged in the medical image sequence is related to the order in which they were acquired. They can be arranged from early to late according to the acquisition time of each image, or from late to early according to the acquisition time of each image. This manual does not limit this arrangement.

[0067] S102: Select several candidate images from the medical image sequence according to preset rules.

[0068] Furthermore, in the current process of obtaining patient clinical indicators based on medical image sequences from DICOM files, there may be interfering information in the medical image sequences that could affect the accuracy of clinical indicators. This interfering information may be localization images within the medical image sequences. To further improve the accuracy of clinical indicator results, in the embodiments of this specification, interfering information is removed by filtering localization images from the medical image sequences. The medical image sequences with reduced interference are then combined with existing clinical indicator determination schemes to obtain clinical indicators with higher accuracy.

[0069] In practice, patients' physiological parameters (such as height, weight, etc.) vary, making it difficult to standardize their initial position and limb posture on the scanning table. Furthermore, different patients require different areas to be scanned. Therefore, it is necessary to determine the approximate location of the areas to be scanned before or after the CT scan. Based on this, acquiring positioning images is an indispensable part of the CT scan process. By obtaining the approximate location of the areas to be scanned from the positioning images, the operator can not only determine the appropriate range of body parts to be scanned but also identify any unnecessary foreign objects on the patient and determine whether the patient's limb posture meets the requirements for normal scanning.

[0070] Generally, since the images in a medical image sequence are arranged sequentially according to the order of image acquisition, while the positioning images are acquired before or after the formal CT scan and are used to assist the operator in positioning the transverse scan, the positioning images can be selected according to the order of the positioning images in the medical image sequence and the preset rules.

[0071] Based on this, in the embodiments of this specification, a number of candidate images are first selected from the medical image sequence according to preset rules, wherein the preset rules can be determined with the scanning timing of the image location in the actual scanning process for the patient.

[0072] Optionally, the attribute information of each image in a medical image sequence includes an instance number. The instance number represents the image's position within the sequence; that is, the instance number is related to the image acquisition order. Generally, when images are arranged from earliest to latest acquisition time, the instance numbers can be arranged from smallest to largest, meaning the earlier the image was acquired, the smaller its instance number. Of course, the acquisition time and instance number can have other correspondences, which this specification does not limit. Optionally, since localization images are generally acquired before or after the formal CT scan, the instance numbers of these localization images are usually a few of the largest or smallest (usually no more than three) among all images in the sequence.

[0073] S104: Determine the localization image based on the differences between the remaining images in the medical image sequence (excluding the candidate image) and the candidate image.

[0074] Furthermore, based on the candidate images selected according to preset rules in step S102 above, it is necessary to further filter out the positioning images from the candidate images.

[0075] In practical applications, although the preset rules used to select candidate images can be dynamically adjusted, directly using each candidate image as a localization image may result in candidate images that are not actually localization images being used as localization images. Subsequently, to reduce interference information in the medical image sequence, localization images are directly removed from the sequence. If candidate images that are not localization images are removed, images containing information about the patient's body parts may be excluded, causing unnecessary information loss. Therefore, after selecting candidate images according to preset rules, localization images can be further filtered from these candidate images.

[0076] Specifically, the remaining images in the medical image sequence, excluding candidate images, are first identified. Then, based on the differences between the remaining images and the candidate images, the candidate images with the largest differences are selected as localization images. This is because localization images are acquired separately before or after the formal scan, and their purpose is to help the operator obtain the approximate location of the patient's body parts, not the exact anatomical structure of the body parts. Therefore, localization images differ significantly from the remaining images used to characterize the anatomical structure of the patient's body parts in terms of scanning position and direction.

[0077] S106: Remove the positioning image from the medical image sequence to obtain a newly determined medical image sequence.

[0078] Furthermore, removing the localization image from the medical image sequence can eliminate interfering information contained within it. Doctors can then obtain the patient's 3D medical image, the identification results of the patient's body parts, and clinical indicators such as lesions in the patient's medical images from the medical image sequence free of interfering information, thus assisting in clinical diagnosis. The scheme for obtaining the patient's clinical indicators based on the redefined medical image sequence can be any existing scheme that can obtain the patient's clinical indicators, such as existing 3D image reconstruction schemes, pre-trained body part recognition models, or lesion recognition models; this specification does not limit this approach.

[0079] Optionally, since the positioning images in the medical image sequence are removed, at least some of the attribute information of the DICOM file is modified. This attribute information may include the number of images in the medical image sequence and the sectional thickness used to describe the coordinate interval between two adjacent images in the medical image sequence. Of course, it may also include other types of attribute information, which are not limited in this specification.

[0080] The image processing method provided in this specification selects several candidate images from the images included in a medical image sequence according to preset rules. Based on the differences between the candidate images and the remaining images in the medical image sequence, a localization image is determined and removed from the medical image sequence. The resulting medical image sequence is then redefined. Therefore, by removing the localization image from the medical image sequence, the interference information contained in the redefined medical image sequence is reduced, thus improving the accuracy of the patient's clinical indicators obtained based on the redefined medical image sequence.

[0081] In one or more embodiments of this specification, such as Figure 1 Steps S102 to S104 involve selecting several candidate images from the medical image sequence according to preset rules and determining the localization images among the candidate images. Specifically, this can be achieved through the following methods: Figure 2 As shown.

[0082] S200: Obtain the sequence number of each image in the medical image sequence.

[0083] In practical applications, medical image sequences contain multiple images obtained from scanning body parts of a patient. The order in which these images are arranged in the sequence is related to their acquisition order. Furthermore, the attribute information of each image in the DICOM file includes an instance number, which represents the image's position in the medical image sequence. In other words, the instance number is related to the acquisition order. Generally, when images are arranged from earliest to latest acquisition time, their instance numbers can be arranged from smallest to largest; that is, the earlier the image was acquired, the smaller its instance number. Of course, the relationship between acquisition time and instance number can also be other, and this specification does not limit this.

[0084] S202: Select several images whose serial numbers are greater than a first preset threshold and / or whose serial numbers are less than a second preset threshold as candidate images.

[0085] Locating images are usually acquired before or after the formal CT scan. Therefore, the sequence number of the locating images is usually one of the largest or smallest numbers among all the images in the sequence.

[0086] Based on this, a first preset threshold and a second preset threshold are determined, wherein the sequence number corresponding to the first preset threshold is greater than the sequence number corresponding to the second preset threshold. Several images with sequence numbers greater than the first preset threshold and / or less than the second preset threshold are selected as candidate images, which is equivalent to filtering out several images at the beginning and / or the end of the medical image sequence as candidate images.

[0087] S204: Select the images in the medical image sequence whose sequence number is less than the first preset threshold and whose sequence number is greater than the second preset threshold as the remaining images other than the candidate images.

[0088] Furthermore, the localization images are acquired separately before or after the formal scan. Their purpose is to assist the operator in obtaining the approximate location of the patient's body parts, not the precise anatomical structure of those parts. Therefore, the localization images differ significantly from the remaining images used to characterize the patient's anatomical structure in terms of scanning position and direction. Thus, images with sequence numbers between a first and a second preset threshold are considered as the remaining images, excluding candidate images. Alternatively, images in the medical image sequence other than the candidate images determined in step S202 can be directly used as the remaining images.

[0089] S206: Select a reference image from the remaining images.

[0090] Generally, since the images used to represent specific patient body parts in a medical image sequence are obtained through continuous scanning, the scanning directions of the other images in the sequence, except for the positioning image, are not significantly different; that is, the angles between the scanning directions of the other images are small. Therefore, it is sufficient to select one of the remaining images for comparison with the candidate image. A reference image can be randomly selected from the remaining images, or the image with the middle sequence number can be used as the reference image. This specification does not limit the method of selecting a reference image from the remaining images.

[0091] S208: For each candidate image, determine the angle between the scanning direction of the candidate image and the scanning direction of the reference image.

[0092] Generally, the image attribute information in a DICOM file includes the image's scanning direction. By determining the angle between the scanning direction of the candidate image and the scanning direction of the reference image, the difference between the scanning directions of the candidate image and the reference image can be determined. The larger the angle, the greater the difference.

[0093] S210: The candidate image with the included angle greater than the preset angle threshold is used as the positioning image.

[0094] The purpose of filtering out the localization images in the above steps is to remove interference information in the medical image sequence that may affect the accuracy of the body part recognition model in recognizing the patient's body parts in the images. Therefore, by using candidate images with larger angles as localization images and removing them from the medical image sequence, we can remove localization images that serve as interference information as much as possible, thereby improving the accuracy of the body part recognition model.

[0095] In one or more embodiments of this specification, based on such Figure 1 The image processing method shown yields a redefined medical image sequence, which can be used by a pre-trained body part recognition model to identify the patient's body parts represented by the redefined medical image sequence. Specifically, in response to a recognition request, the redefined medical image sequence is input into the pre-trained body part recognition model to obtain the patient's body part recognition results.

[0096] The electronic device that performs the training process of the body part recognition model can be the same as or different from the electronic device that performs the image processing method; this specification does not limit this.

[0097] The model structure of the body part recognition model can be any existing body part recognition model. Its training samples can be pre-acquired images that can represent the patient's body parts. The annotations of the training samples can be the body parts or medical anatomical structures represented by the images. This specification does not limit the structure of the model, the composition of the training samples and their labels, or the specific training process.

[0098] In one or more embodiments of this specification, before inputting the pre-trained body part recognition model with the medical image sequence to be redefined, and obtaining the patient's body part recognition result, another type of interference information in the medical image sequence can be eliminated again to further improve the accuracy of the body part recognition model's recognition result. Specifically, this interference information can be a multi-phase sequence contained in the medical image sequence, such as... Figure 3 As shown, the specific solution is as follows:

[0099] S300: For each image, determine whether the scan position of the image is the same as the scan position of the next image. If yes, proceed to step S306; otherwise, proceed to step S302.

[0100] Furthermore, for medical image sequences that have eliminated interference from localization images, another aspect of information may affect the accuracy of the body part recognition model's results in identifying the patient's body parts: multi-phase sequence information. Multi-phase sequence refers to the image sequence obtained by contrast-enhanced CT scans, where contrast agent is injected and its flow into the arteries, capillaries, and veins of the examined organ is tracked in real time. Therefore, the resulting medical image sequence is also called a multi-phase image sequence. Specifically, a multi-phase sequence may include image sequences of the arterial phase, venous phase, and equilibrium phase. In the embodiments of this specification, "arterial phase," "venous phase," and "equilibrium phase" generally refer to different phases of blood vessels observed in CT scans or angiography. The arterial phase is the period when arterial vessels are filled and visualized, making arterial vessels relatively clear; the venous phase is the period when venous vessels are filled and visualized, making the portal vein relatively clear; and the equilibrium phase is when all blood vessels have filled and visualized, making the overall vascular system relatively clear.

[0101] In addition, there are also sequential relationships between subsequences of different phases in the same medical image sequence. These relationships are usually determined based on the different stages of different types of vascular imaging in a patient during a CT scan or angiography. The specific order is not limited in this instruction manual.

[0102] S302: Determine whether the scanning direction of the current image is the same as the scanning direction of the next image. If yes, proceed to step S304; otherwise, proceed to step S306.

[0103] Specifically, subsequences within the same medical image sequence can be understood as images obtained from multiple scans of a patient during the same CT scan or angiography examination. In other words, the patient needs to be scanned completely several times, such as three times, during a single examination. Therefore, while the scanning direction of the images within each subsequence is not significantly different, the scanning location may vary depending on the body part being scanned. On the other hand, images within different subsequences may share the same scanning location because they were acquired during different scan processes. Furthermore, the scanning direction may differ even when the scanning location is different.

[0104] When performing multi-phase segmentation on medical image sequences, determining whether the scanning direction of a given image is the same as the scanning direction of the next image can be achieved by determining the angle between their scanning directions. If the angle is less than a preset threshold, the image and its next image are considered to have the same scanning direction; conversely, if the angle is not less than the preset threshold, the image and its next image are considered to have different scanning directions. It is understandable that the angle of the scanning direction determined during multi-phase segmentation is related to... Figure 1 Steps S102 to S104 determine the angle of the scanning direction when determining the positioning image. Although both are angles determined based on the scanning direction, they are unrelated. Furthermore, the preset angle threshold used to determine the positioning image is different from the preset threshold used when splitting multiple phases.

[0105] Based on this, for each image in the medical image sequence, it is determined whether the image satisfies the following conditions: the scanning position of the image is different from the scanning position of the next image; and the scanning direction of the image is the same as the scanning direction of the next image. If both conditions are met, it is determined that the image and the next image are in the same subsequence. If at least one condition is not met, it is determined that the image and the next image are not in the same subsequence.

[0106] It's understandable that the reason for comparing each image with its next image to perform sequence splitting is because the images in a medical image sequence are arranged in a specified order, such as a specified acquisition order. Furthermore, in the above scheme, "same" can mean a difference less than a preset threshold, and "different" can mean a difference greater than a preset threshold.

[0107] S304: Determine that the image is in the same subsequence as the next image of the image.

[0108] S306: Determine that the image is not in the same subsequence as the next image.

[0109] For the medical image sequence processed through steps S300 to S306, if all the images in the medical image sequence are in the same subsequence, it indicates that the medical image sequence is not a multi-phase sequence and will not be split into subsequences. If at least one image in the medical image sequence is not in the same subsequence as the others, it indicates that the medical image sequence will be split into multiple subsequences. For a medical image sequence split into multiple subsequences, the attribute information of each subsequence changes, such as the number of images in each subsequence and the average thickness of the cross section used to characterize the coordinate interval between images in the subsequence. Of course, other types of attribute information may also be included, and this specification does not limit this.

[0110] S308: For each subsequence in the redefined medical image sequence, take each image contained in the subsequence as input and input it into the pre-trained body part recognition model to obtain the body part of the patient corresponding to the subsequence as the recognition result of the body part corresponding to the subsequence.

[0111] Specifically, all images contained in each subsequence are used as input to the body part recognition model. The recognition result of the patient's body part corresponding to each subsequence is determined by the output of the body part recognition model.

[0112] S310: Based on the recognition results corresponding to each sub-sequence, determine the recognition results of the patient's body parts.

[0113] Through the above steps, the images of each sub-sequence are input into the body part recognition model to obtain the recognition results of each sub-sequence. In the process of the body part recognition model recognizing the patient's body parts, the interference between images of different sub-sequences is eliminated, further improving the accuracy of recognition.

[0114] It is evident that by introducing sequence information error correction and medical image processing technology, the information in medical image sequences becomes more accurate, and the identification becomes more precise. This allows operators to more accurately identify the patient's body parts based on the patient's medical image sequences.

[0115] It should be noted that the above-mentioned methods for removing interference information, namely, eliminating the localization image and splitting the subsequence, can be implemented by first performing the following steps: Figure 1 The removal of the localization images shown in steps S102 to S104, followed by the execution of the following steps on the medical image sequence after removing the localization images: Figure 3 The subsequence splitting shown in steps S300 to S306; of course, it is also possible to first perform the following on the medical image sequence: Figure 3 Steps S300 to S306 show the splitting of subsequences, and then for each subsequence, the following steps are performed: Figure 1The removal of the positioning images shown in steps S102 to S104. This specification does not specify the execution order.

[0116] In an optional embodiment of this specification, since the trained body part recognition model can be retrained based on updated training samples and their annotations according to specific application scenarios, and its own model parameters can be updated, in response to the update of the body part recognition model's parameters, the specified medical image sequence identified by the body part recognition model before the update can be re-identified to update the recognition result corresponding to the specified medical image sequence. The specific scheme is as follows:

[0117] First, in response to an update of the parameters of the body part recognition model, a specified medical image sequence that has undergone recognition based on the body part recognition model before the update is acquired.

[0118] In the embodiments of this specification, for a specified medical image sequence that has undergone recognition based on the previous body part recognition model, the recognition result of the specified medical image sequence can be updated by using the updated body part recognition model. For medical image sequences that have not undergone recognition based on the previous body part recognition model, the updated body part recognition model can be used directly for recognition. This specification will not elaborate on this situation.

[0119] Specifically, each set of model parameters for a body part recognition model can correspond to a version. Due to the difficulty and lengthy acquisition process of training samples and their annotations in the field of medical imaging, the initial version of a body part recognition model may be trained using only a small number of training samples or with relatively inaccurate annotations. However, as the field of medical imaging develops, the number of training samples increases and the accuracy of annotations improves. Therefore, the body part recognition model can be retrained based on the original version to obtain more accurate model parameters, i.e., an updated version.

[0120] Therefore, for a given medical image sequence that has already been identified using the previous body part recognition model, the recognition results can be updated as the body part recognition model version iterates.

[0121] Secondly, based on the updated body part recognition model, the specified medical image sequence is recognized to obtain the specified recognition result.

[0122] Furthermore, the database used to store the recognition results of medical image sequences can store a large number of recognition results for medical image sequences and establish a correspondence between the recognition results and the medical image sequences. That is, the recognition results of a medical image sequence can be found in the database using that medical image sequence. For a given medical image sequence that has already been recognized using the previous body part recognition model, it can be re-recognized using the updated body part recognition model to obtain the specified recognition result. Because the model parameters of the body part recognition model have been updated, improving the accuracy and realism of the model's output results, the accuracy of the specified recognition result obtained through the updated body part recognition model will be higher than that of the original recognition result of the specified medical image sequence stored in the database.

[0123] Optionally, in practical applications, there may be too many recognition results for medical image sequences stored in the database, making it impossible to immediately update the recognition results of all medical image sequences after the model parameters of the body part recognition model are updated. Therefore, the update cycle of the recognition results can be determined according to the actual situation. According to the update cycle, a preset number of specified medical image sequences can be extracted, and the specified recognition results of the preset number of specified medical image sequences can be obtained through the updated body part recognition model.

[0124] Then, based on the specified recognition result, the recognition result of the specified medical image sequence is updated.

[0125] Furthermore, the specified recognition results obtained based on the updated body part recognition model are stored in the database. The storage method can be to update the original recognition results, i.e., overwrite the original recognition results, or to store the executed recognition results together with the original recognition results. This specification does not limit the method.

[0126] Figure 4 This specification provides a schematic diagram of an image processing apparatus, specifically including:

[0127] The sequence determination module 400 is used to acquire a medical image sequence to be identified, the medical image sequence including multiple images of body parts acquired in a specified order;

[0128] Selection module 402 is used to select several candidate images from the medical image sequence according to preset rules;

[0129] The localization image determination module 404 is used to determine the localization image based on the differences between the remaining images in the medical image sequence (excluding the candidate image) and the candidate image.

[0130] The removal module 406 is used to remove the positioning image from the medical image sequence to obtain a redefined medical image sequence.

[0131] Optionally, the selection module 402 is specifically used to obtain the sequence number of each image in the medical image sequence; and to select a number of images whose sequence number is greater than a first preset threshold and / or whose sequence number is less than a second preset threshold as candidate images.

[0132] Optionally, the positioning image determination module 404 is specifically used to: select images in the medical image sequence whose sequence number is less than the first preset threshold and whose sequence number is greater than the second preset threshold as all remaining images other than the candidate images; select a reference image from all remaining images; for each candidate image, determine the angle between the scanning direction of the candidate image and the scanning direction of the reference image; and select candidate images whose angle is greater than a preset angle threshold as positioning images.

[0133] Optionally, the device further includes:

[0134] The recognition module 408 is used to respond to the recognition request by inputting the redefined medical image sequence into the pre-trained body part recognition model to obtain the patient's body part recognition results.

[0135] Optionally, the device further includes:

[0136] The splitting module 410 is specifically used to split the medical image sequence into sub-sequences based on the scanning position and scanning direction of each image contained in the newly determined medical image sequence.

[0137] Optionally, the recognition module 408 is specifically used to: for each subsequence, take each image contained in the subsequence as input, input it into a pre-trained body part recognition model, and obtain the body part corresponding to the subsequence as the recognition result of the subsequence; and determine the recognition result of the patient's body part based on the recognition results corresponding to each subsequence.

[0138] Optionally, the splitting module 410 is specifically used to determine, for each image in the medical image sequence, whether the image satisfies the following conditions: the scanning position of the image is different from the scanning position of the next image; and the scanning direction of the image is the same as the scanning direction of the next image. If both conditions are met, the image and the next image are determined to be in the same subsequence. If at least one condition is not met, the image and the next image are determined to be in different subsequences.

[0139] Optionally, the device further includes:

[0140] The update module 412 is specifically used to, in response to the update of the parameters of the body part recognition model, acquire a specified medical image sequence that has undergone recognition based on the body part recognition model before the update; recognize the specified medical image sequence according to the body part recognition model after the parameter update to obtain a specified recognition result; and update the recognition result of the specified medical image sequence according to the specified recognition result.

[0141] Optionally, the update module 412 is specifically used to determine the update cycle of the recognition results; and to obtain a preset number of the specified medical image sequences according to the update cycle.

[0142] This specification also provides a computer-readable storage medium storing a computer program that can be used to execute the above-described... Figure 1 The image processing method shown.

[0143] This instruction manual also provides Figure 5 The diagram shows a schematic structural representation of the electronic device. Figure 5 At the hardware level, the electronic device includes a processor, internal bus, network interface, memory, and non-volatile memory, and may also include other hardware required for the business operations. The processor reads the corresponding computer program from the non-volatile memory into memory and then runs it to achieve the above-mentioned functions. Figure 1 The image processing method shown is not limited to software implementation. This specification does not exclude other implementation methods besides software implementation, such as logic devices or a combination of hardware and software. In other words, the execution entity of the following processing flow is not limited to individual logic units, but can also be hardware or logic devices.

[0144] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many methodological improvements today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that a methodological improvement cannot be implemented using hardware physical modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program and "integrate" a digital system onto a PLD themselves, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should understand that by simply performing some logic programming on the method flow using one of these hardware description languages ​​and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.

[0145] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, ASICs, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0146] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0147] For ease of description, the above devices are described in terms of function, divided into various units. Of course, in implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0148] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0149] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0150] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0151] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0152] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0153] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0154] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0155] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0156] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this specification may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0157] This specification can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0158] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0159] The above description is merely an embodiment of this specification and is not intended to limit this specification. Various modifications and variations can be made to this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this specification should be included within the scope of the claims of this specification.

Claims

1. An image processing method, characterized in that, The method includes: Acquire a medical image sequence to be identified, the medical image sequence comprising multiple images of body parts acquired in a specified order; Obtain the sequence number of each image in the medical image sequence, and select several images whose sequence numbers are greater than a first preset threshold and / or whose sequence numbers are less than a second preset threshold as candidate images; Images in the medical image sequence whose sequence number is less than the first preset threshold and whose sequence number is greater than the second preset threshold are selected as the remaining images other than the candidate images; a reference image is selected from the remaining images; for each candidate image, the angle between the scanning direction of the candidate image and the scanning direction of the reference image is determined; candidate images whose angle is greater than a preset angle threshold are selected as positioning images; the positioning images are acquired before or after the formal scan and are used to assist in the positioning of the formal scan. The positioning images differ significantly from the remaining images used to characterize the anatomical structure of the patient's body parts in terms of scanning position and scanning direction. The localization image is removed from the medical image sequence to obtain a newly determined medical image sequence.

2. The method as described in claim 1, characterized in that, The method further includes: In response to the recognition request, the redefined medical image sequence is input into the pre-trained body part recognition model to obtain the recognition results of the patient's body parts.

3. The method as described in claim 2, characterized in that, In response to a recognition request, before inputting the redefined medical image sequence into a pre-trained body part recognition model, the method further includes: Based on the scanning position and scanning direction of each image contained in the redefined medical image sequence, the medical image sequence is split into sub-sequences.

4. The method as described in claim 3, characterized in that, In response to the recognition request, the redefined medical image sequence is input into a pre-trained body part recognition model to obtain the recognition results of the patient's body parts, specifically including: For each subsequence, the images contained in the subsequence are used as input to a pre-trained body part recognition model to obtain the body part corresponding to the subsequence as the recognition result of the subsequence; Based on the recognition results corresponding to each sub-sequence, the recognition results of the patient's body parts are determined.

5. The method as described in claim 3, characterized in that, The medical image sequence is split into sub-sequences, specifically including: For each image in the medical image sequence, determine whether the image meets the following conditions: The scan position of this image is different from the scan position of the next image. Furthermore, the scanning direction of this image is the same as the scanning direction of the next image; If all conditions are met, then the image is determined to be in the same subsequence as the next image. If at least one condition is not met, then it is determined that the image is not in the same subsequence as the next image.

6. The method as described in claim 2, characterized in that, The method further includes: In response to an update of the parameters of the body part recognition model, a specified medical image sequence that has undergone recognition based on the body part recognition model before the update is acquired; Based on the updated body part recognition model, the specified medical image sequence is recognized to obtain the specified recognition result; Update the recognition result of the specified medical image sequence based on the specified recognition result.

7. The method as described in claim 6, characterized in that, Obtain a specified medical image sequence that has undergone recognition based on the previous body part recognition model, specifically including: Determine the update cycle for the recognition results; According to the update cycle, a preset number of the specified medical image sequences are obtained.

8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the method described in any one of claims 1 to 7.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method described in any one of claims 1 to 7.

Images