Method, device and storage medium for object analysis of medical images

Abstract

The present disclosure relates to a method, device and storage medium for object analysis of medical images, and the method may include the following steps. A 3D medical image containing an object may be obtained. Corresponding window widths and window levels may be set for each type of object, and each sub-image sequence may be adjusted based on each window width and window level to obtain a sub-image sequence of each channel. Based on the sub-image sequence of each channel, a sub-object analysis model corresponding to each part may be used for analysis to obtain a sub-object analysis result. The method and device set a plurality of window widths and window levels according to the object lesion category, and the sub-image sequence is adjusted according to the set window width and window level to obtain a multi-channel image, and the multi-channel image is used to replace the single-channel image as the input of the sub-object analysis model. This method can improve the recognition rate of the sub-object analysis model for different categories of lesions, and effectively solve the problem of CT value differences corresponding to different vascular lesion categories.

Description

[0001] This application is a divisional application of the Chinese invention patent application with application number 202111652073.5, application date December 31, 2021, and invention name “Method, device and storage medium for object analysis of medical images”. Technical Field

[0002] The present disclosure relates to the field of medical images, and more specifically, to a method, device and storage medium for performing object analysis on medical images. Background Art

[0003] Vascular disease has always been a major problem threatening human health. A considerable proportion of vascular diseases are caused by the accumulation of plaque lesions on the blood vessel walls, which causes vascular stenosis, and abnormal bulging on the blood vessel walls, which causes aneurysms. However, existing technologies have certain deficiencies in the detection and identification of vascular lesions.

[0004] Taking head and neck artery plaques as an example, head and neck artery disease usually refers to arterial stenosis or blockage caused by the accumulation of atherosclerotic plaques in the arterial wall. Patients with intracranial artery stenosis and blockage have limited blood supply to the brain, which can easily lead to ischemic stroke in patients. If the plaque ruptures, it is very easy to block and damage blood vessels, causing acute stroke in patients. According to the composition of atherosclerotic plaques, plaques can be further divided into calcified plaques, non-calcified plaques, and mixed plaques, among which mixed plaques have both calcified and non-calcified plaque components. Non-calcified and mixed plaques are prone to rupture.

[0005] Computed Tomography Angiography (CTA) or Magnetic Resonance Angiography (MRA) can image blood vessels and their lesions in various parts of the body and are commonly used vascular imaging techniques. Non-calcified plaques, mixed plaques, and aneurysms have low contrast with surrounding tissues on the images and are easily confused with surrounding tissues, resulting in missed detection, while surrounding tissues are very likely to lead to misdetection.

[0006] At present, the detection methods of vascular lesions in head and neck CTA generally include manual analysis and automatic analysis software. Manual plaque analysis relies heavily on the experience of radiologists and cardiovascular experts. Atherosclerotic plaques, aneurysms and other lesions are discretely distributed on the complex structure of the head and neck artery walls. Analyzing vascular lesions in massive CTA data is undoubtedly a very time-consuming task for doctors. The uncertainty of non-calcified and mixed plaques increases the difficulty of doctors' diagnosis.

[0007] Existing vascular lesion analysis software can reduce doctors' daily diagnostic workload to a certain extent, but it also has certain shortcomings. For example, the semi-automatic analysis software provided by CT equipment manufacturers such as Siemens requires a lot of manual interaction to complete vascular segmentation, diameter estimation, and wall morphology analysis, and this solution generally only targets local blood vessels.

[0008] Recently, deep learning technology has been gradually applied to vascular lesion detection and has achieved remarkable results. However, existing solutions generally use a single model to predict vascular lesion information in CTA sequences, which results in a long prediction process and many false positives. It is usually only applicable to a single part and the prediction results are not accurate when applied to medical images containing multiple parts. Summary of the invention

[0009] The present disclosure is provided to solve the above-mentioned problems existing in the prior art.

[0010] A method, device and storage medium for object analysis of 3D medical images are needed, which can accurately identify various objects of different types distributed in multiple parts in the entire 3D medical image with a reasonable time consumption, without being affected by or significantly suppressing the influence of differences in CT values ​​corresponding to different types of objects.

[0011] According to a first embodiment of the present disclosure, a method for object analysis of a medical image is provided, wherein the method may include acquiring a 3D medical image containing an object. The method may also divide the 3D medical image into a sub-image sequence of each part according to the part.

[0012] The method can also set corresponding window widths and window levels for each type of object, and adjust the windows of each sub-image sequence based on each window width and window level to obtain a sub-image sequence of each channel. The method can also analyze the sub-object analysis model corresponding to each part based on the sub-image sequence of each channel to obtain a sub-object analysis result. The method can also fuse the analysis results of each sub-object to obtain the object analysis result of the 3D medical image.

[0013] According to a second solution of the present disclosure, a device for performing object analysis on a medical image is provided, and the device for performing object analysis on a medical image may include an interface and a processor. The interface may be configured to acquire a 3D medical image containing an object. The processor may be configured (for example, via an interface) to include acquiring a 3D medical image containing an object. The processor may also be configured to divide the 3D medical image into sub-image sequences of each part according to the part. The processor may also be configured to set corresponding window widths and window levels for each type of object, and adjust the windows of each sub-image sequence based on each window width and window level to obtain a sub-image sequence of each channel. The processor may also be configured to analyze the sub-image sequence of each channel using a sub-object analysis model corresponding to each part to obtain a sub-object analysis result; the processor may also be configured to fuse the sub-object analysis results to obtain the object analysis result of the 3D medical image.

[0014] According to a third scheme of the present disclosure, a computer storage medium is provided, on which executable instructions are stored, and when the executable instructions are executed by a processor, the steps of a method for object analysis of a medical image are implemented. The method may include acquiring a 3D medical image containing an object. The method may also divide the 3D medical image into sub-image sequences of each part according to the part. The method may also set corresponding window widths and window levels for each type of object, and adjust the windows of each sub-image sequence based on each window width and window level to obtain a sub-image sequence of each channel. The method may also analyze the sub-object analysis model corresponding to each part based on the sub-image sequence of each channel to obtain a sub-object analysis result. The method may also fuse the analysis results of each sub-object to obtain the object analysis result of the 3D medical image.

[0015] According to the method, device and storage medium for object analysis of medical images according to various embodiments of the present disclosure, for example, for plaque detection of blood vessels (as an example of an object), the present application can have the following advantages over existing solutions:

[0016] 1. The present invention adopts a slice classification model to obtain a sub-image sequence of each part, sets a variety of window widths and window positions according to the type of object lesions, and obtains a multi-channel image after adjusting the sub-image sequence according to the set window widths and window positions. The multi-channel image replaces the single-channel image as the input of the sub-object analysis model, thereby improving the recognition rate of the sub-object analysis model for different types of lesions. For objects of different types distributed in multiple parts, accurate recognition of various objects can be achieved in the entire 3D medical image with a reasonable time consumption, without being affected by or significantly suppressing the influence of the differences in CT values ​​corresponding to different types of objects.

[0017] 2. The present disclosure does not rely on complex manual interactions and can accomplish accurate and efficient detection of vascular lesions in a series of images containing extended organs or tissues (e.g., blood vessels). This series of images, such as a head and neck CTA image sequence, contains the head, neck arteries, and aortic arch, each of which has a large number of branches. The object analysis method of the present disclosure can achieve accurate and efficient detection of vascular lesions in a complex vascular tree. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In the drawings, which are not necessarily drawn to scale, the same reference numerals may describe similar components in different views. The drawings generally illustrate various embodiments by way of example and not limitation, and together with the description and claims, serve to illustrate the disclosed embodiments. When appropriate, the same reference numerals are used throughout the drawings to refer to the same or similar parts. Such embodiments are illustrative and are not intended to be exhaustive or exclusive embodiments of the present apparatus or method.

[0019] Figure 1 A method for performing object analysis on a medical image according to an embodiment of the present disclosure is shown.

[0020] Figure 2 The part division process according to an embodiment of the present disclosure is shown.

[0021] Figure 3 The object analysis process based on a sub-image sequence according to an embodiment of the present disclosure is shown.

[0022] Figure 4 The process of intercepting a sliding window block according to an embodiment of the present disclosure is shown.

[0023] Figure 5 The detection result fusion process according to an embodiment of the present disclosure is shown.

[0024] Figure 6 The figure illustrates an illustrative block diagram of an exemplary apparatus for performing object analysis on a medical image according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the technical solution of the present disclosure, the present disclosure is described in detail below in conjunction with the accompanying drawings and specific implementation examples. The embodiments of the present disclosure are further described in detail below in conjunction with the accompanying drawings and specific implementation examples, but are not intended to limit the present disclosure. For the various steps described herein, if there is no necessity for a causal relationship between each other, the order in which they are described as examples herein should not be regarded as a limitation, and those skilled in the art should know that they can be adjusted in order as long as the logic between them is not destroyed, resulting in the inability to implement the entire process.

[0026] Figure 1The method for object analysis of a medical image according to an embodiment of the present disclosure is shown. Figure 1 As shown, a method for object analysis of medical images begins with step S1, obtaining a 3D medical image containing an object; wherein the object can be any organ or tissue extending a certain length, such as but not limited to at least one of a blood vessel, a digestive tract, a mammary duct, and a respiratory tract, or a lesion therein. The lesion is a lesion or abnormality such as an atherosclerotic plaque, an aneurysm, or a stent in a blood vessel. The 3D medical image is a CTA image containing a blood vessel, a CT image containing a rib, or a CT image containing a lung. The vascular lesion is at least one of a calcified plaque, a non-calcified plaque, a mixed plaque, an aneurysm, and a stent image.

[0027] In this embodiment, vascular lesions are used as an example of an object for description. The 3D medical image is a head and neck CTA image containing blood vessels. This embodiment is used to illustrate the detection of vascular atherosclerotic plaques.

[0028] 3D medical images need to conform to the medical imaging format of digital imaging and meet the Digital Imaging and Communications in Medicine (DICOM) protocol. 3D medical images also need to meet the basic requirements of CTA images, such as no contrast agent filling and no obvious motion artifacts.

[0029] In step S2, the processor may be used to divide the 3D medical image into sub-image sequences of each part according to the part (eg Figure 2 Step S21);

[0030] In this embodiment, step S2 may specifically include: based on the 3D medical image, using a slice classification model, identifying key slices in the 3D medical image as the junction of adjacent parts; using the identified key slices to achieve the division of sub-images according to parts. In some embodiments, the slice classification model is implemented using a two-dimensional learning network and is trained using training samples with classification information of slices corresponding to parts.

[0031] In this embodiment, the 3D medical image is a head and neck CTA image containing blood vessels. In order to distinguish the sub-image sequence of the three sub-parts of the head, neck and chest, two key slices are required to distinguish the head and neck, and the neck and chest. The sub-image sequence determination process is as follows: Figure 2As shown, the slice classification model is obtained by training according to the training samples with slice classification information of the parts of interest. In some embodiments, the slice classification model adopts a 2D ResNet network structure, and the training method is based on the labeling of two key slices in the training sample images by experienced radiologists, and the slices of the head, neck, and chest are collected according to the labeled key slice information, which is used as the gold standard classification information corresponding to the training sample. Then the training sample is input into the slice classification model to obtain the slice classification result of the training sample, and the loss between the slice classification result and the gold standard classification information is calculated. The slice classification model is adjusted according to the loss, which is not specifically limited here; when adjusting the network parameters, a stochastic gradient descent SGD optimizer or other types of optimizers can be used, which are not specifically limited here.

[0032] In step S3, corresponding window widths and window levels can be set for each type of object, and the windows of each sub-image sequence can be adjusted based on each window width and window level (e.g. Figure 3 Step S31) to obtain a sub-image sequence of each channel. In the present embodiment, taking atherosclerotic plaque as an example, its types may include calcified plaque, non-calcified plaque and mixed plaque, so 3 window widths and window positions are set. Based on these 3 window widths and window positions, each sub-image sequence is adjusted to obtain a sub-image sequence of 3-channel images. The CT values ​​of calcified plaque, non-calcified plaque and mixed plaque are different, and the difference in CT value is reflected in the difference in gray value. On the image, non-calcified plaque, mixed plaque and aneurysm have a low contrast with the surrounding tissue, which is easily confused with the surrounding tissue and causes missed detection. In the present embodiment, a variety of window widths and window positions are set according to the object lesion category, and the sub-image sequence is adjusted according to the set window width and window position to obtain a multi-channel image. Since the CT values ​​of different types of lesions are different, the window width and window position of the gray value are set and adjusted for each type of object. The sub-image sequence of each channel obtained after window adjustment can highlight the grayscale information of the corresponding type of objects. The multi-channel image obtained after window adjustment can replace the single-channel image as the input of the sub-object analysis model, thereby improving the recognition rate of the sub-object analysis model for different types of lesions and effectively solving the problem of different CT value differences corresponding to different types of vascular lesions.

[0033] In step S4, model parameters can be adjusted for each sub-object analysis model based on the prior information of each part and the object segmentation result after skeletonization. The prior information of the part may include at least one of the size (such as the diameter of blood vessels in different parts), shape and number of objects contained in the part.

[0034] In this embodiment, step S4 may specifically include: determining the size of the sliding window block based on prior information of each part, determining the internal representative points of the sliding window block based on the object segmentation results of each sub-image sequence after skeletonization, intercepting the sliding window block of the training sample containing the lesion annotation information according to the size of the sliding window block based on the internal representative points, and using the sliding window block as a training sample to train each sub-object analysis model.

[0035] The following is an example of using the center point of the sliding window block as an internal representative point, but it should be noted that the internal representative point is not limited to this, for example, the middle area of ​​the sliding window block can also be used as an example of the internal representative point, etc. The size of the sliding window block is determined based on the prior information of each part, the internal center point of the sliding window block is determined based on the object segmentation results of each sub-image sequence after skeletonization, the sliding window block of the training sample is intercepted according to the size of the sliding window block based on the internal center point, and each sub-object analysis model is trained using the sliding window block as a training sample, which can improve the prediction robustness of the model.

[0036] In this embodiment, the determining of the internal representative points of the sliding window block based on the object segmentation results of each sub-image sequence after skeletonization may specifically include: based on the sub-image sequence of each part, using the processor, using the corresponding segmentation model of each part to determine the corresponding object segmentation result (such as Figure 3 Step S32); performing skeletonization operation on the object segmentation results of each sub-image sequence (such as Figure 3 Step S33): Sparsely sample the skeletonized object segmentation result to obtain internal representative points of the sliding window block. In this embodiment, the sliding window prediction based on the segmentation result can improve the model prediction speed and reduce false positives by referring to the blood vessel segmentation result.

[0037] In this embodiment, each vascular segmentation model is trained separately using training samples with vascular information of corresponding parts. In this embodiment, the 3D medical image is a head and neck CTA image containing blood vessels. Therefore, this embodiment needs to apply a head vascular segmentation model, a neck vascular segmentation model and a chest vascular segmentation model. The training process of the three vascular segmentation models is similar. The head vascular segmentation model is taken as an example. The head vascular segmentation model is obtained by training based on training samples with vascular information of interest. In some embodiments, the vascular segmentation model generally adopts a 3D U-Net network structure. The training method of the vascular segmentation model may include: marking the head blood vessels in the training sample images based on experienced radiologists, and using this as the gold standard during training. Then the training sample image is input into the head vascular segmentation model to obtain the head vascular segmentation result, and the loss between the head vascular segmentation result and the gold standard is calculated. According to the loss, the network parameters of the head segmentation model are adjusted. When the loss is less than or equal to the preset threshold or convergence is reached, it indicates that the head vascular segmentation model training converges. Optionally, a Dice loss function, a cross entropy loss function or other types of loss functions may be used when calculating the loss, without specific limitation herein; when adjusting the network parameters, a stochastic gradient descent SGD optimizer or other types of optimizers may be used, without specific limitation herein.

[0038] In this embodiment, step S4 may further specifically include: using the sliding window block as a training sample to train each sub-object analysis model; using the false positive samples obtained by the training together with the training samples containing the lesion annotation information as new training samples to train each sub-object analysis model, thereby improving the sensitivity and accuracy of the prediction results of the lesion detection model.

[0039] Taking a head sub-image sequence as an example, the following describes how to adjust model parameters for a sub-object analysis model based on the prior information of the head and the object segmentation result after skeletonization.

[0040] The diameter of the head artery is smaller than that of the neck and aortic arch. Based on the prior information of the head, the sliding window block size is set to 32 when the sequence voxel spacing is 0.4 mm. 3 (The sliding window block size corresponding to the neck is 64 3 ), which is consistent with the actual vascular morphology, and can not only improve the lesion detection effect, but also reduce the reasoning time of the lesion detection model. Figure 4 As shown in the figure, after skeletonizing the head blood vessel segmentation results, sparse sampling is used to obtain the sliding window block center points required for sub-object analysis model prediction, and the sparse sampling interval is set to 2 for head data. The plaque detection model (sub-object analysis model) adopts a 3D U-Net network structure and completes the training in an iterative manner. Taking the training method of the head blood vessel plaque detection model as an example, it may include:

[0041] (1) Experienced radiologists mark the head vascular plaques in the training sample images and use them as the gold standard for training.

[0042] (2) The training sample images are then input into the plaque model to obtain the head vascular plaque results, and the loss between the head vascular plaque detection results and the gold standard is calculated.

[0043] (3) According to the loss, the network parameters of the head plaque detection model are adjusted in a gradient descent manner. When the loss is less than or equal to a preset threshold or convergence is reached, it indicates that the model training converges and the first plaque detection model is obtained. Optionally, the loss is generally calculated using a Dice loss function, a cross entropy loss function or other types of loss functions, which are not specifically limited here; when adjusting the network parameters, a stochastic gradient descent SGD optimizer or other types of optimizers can be used, which are not specifically limited here.

[0044] (4) Using the first plaque detection model (the head plaque detection model after network parameter adjustment), the head plaque detection results are predicted according to the specified sliding window block, and false positive samples in some detection results are selected and combined with the gold standard to form new training samples.

[0045] (5) Repeat steps (2)-(3) to iteratively obtain the second plaque detection model.

[0046] (6) Repeat steps (2)-(5) several times to iteratively obtain the final plaque detection model.

[0047] In step S5, the sub-image sequence of each channel may be analyzed using the sub-object analysis model corresponding to each part to obtain a sub-object analysis result;

[0048] In this embodiment, step S5 may specifically include: based on the sub-image sequence of each channel, referring to the prior information of each part and the object segmentation result after skeletonization, using the sub-object analysis model corresponding to each part to perform analysis (such as Figure 3 Step S34) to obtain the sub-object analysis result.

[0049] In step S6, the processor can be used to fuse the analysis results of each sub-object to obtain the object analysis result of the 3D medical image. After predicting the plaque detection results of several sub-parts of the CTA medical image, the vascular plaque detection result is obtained by fusing the plaque detection results of the sub-parts. Taking the head and neck CTA as an example, it generally contains three sub-image sequences of the head, neck, and chest. The plaque detection results of the three sub-image sequences can be re-stacked according to the sub-sequence slice classification to obtain the detection result of the CTA medical image (such as Figure 5 Step S51).

[0050] The present disclosure uses a slice classification model to divide a 3D medical image according to the part to obtain a sub-image sequence of each part, sets three window widths and window positions according to the vascular plaque category, and adjusts the sub-image sequence according to the set three window widths and window positions to obtain a multi-channel image, and replaces the single-channel image with the multi-channel image as the input of the sub-object analysis model. Since the CT values ​​(derived from the attenuation coefficient) of different types of lesions are different, the difference in CT values ​​is reflected in the difference in grayscale values. The window width and window position of the grayscale value window are set for each type of object and the window is adjusted. The sub-image sequence of each channel obtained after the window adjustment can highlight the grayscale information of the corresponding type of object, so that the multi-channel sub-image sequence is used as the input of the sub-object analysis model, and good and accurate analysis results for various types of objects can be obtained, which can improve the recognition rate of the sub-object analysis model for different types of lesions and effectively solve the problem of CT value differences corresponding to different types of vascular lesions. Compared with the manual analysis scheme, the present disclosure can automatically, quickly and accurately complete the detection of vascular lesions, while improving the diagnostic efficiency and greatly reducing the workload of doctors and the waiting time of patients.

[0051] As another possible embodiment, the vascular lesion is an aneurysm. The difference between the analysis process of the aneurysm and the object analysis process of this embodiment is that there is only one type of aneurysm. Therefore, only one window width and window position need to be set. Based on this one window width and window position, each sub-image sequence is adjusted to obtain a sub-image sequence of a single-channel image. The sub-image sequence of the single-channel image is input into the sub-object analysis model for analysis.

[0052] As another possible embodiment, the vascular lesion is a stent. The difference between the analysis process of the stent and the object analysis process of this embodiment is that there is only one type of aneurysm. Therefore, only one window width and window position need to be set. Based on this one window width and window position, each sub-image sequence is adjusted to obtain a sub-image sequence of a single-channel image. The sub-image sequence of the single-channel image is input into the sub-object analysis model for analysis.

[0053] Figure 6 FIG. 1 is a block diagram illustrating an exemplary apparatus for performing object analysis on a medical image according to an embodiment of the present disclosure, Figure 6 As shown, an object analysis device 600 may include an interface 607 and a processor 601. The interface 607 may be configured to receive a 3D medical image containing an object. The processor 601 may be configured to: execute the method for performing object analysis on a medical image according to various embodiments of the present disclosure.

[0054] Through the interface 607, the device for object analysis of medical images can be connected to a network (not shown), such as but not limited to a local area network or the Internet in a hospital. However, the communication mode implemented by the interface 607 is not limited to the network, and may include NFC, Bluetooth, WIFI, etc.; it may be a wired connection or a wireless connection. Taking the network as an example, the interface 607 can connect the device for object analysis of medical images to external devices such as an image acquisition device (not shown), a medical image database 608, and an image data storage device 609. The image acquisition device can be any type of imaging modality, such as but not limited to computed tomography (CT), digital subtraction angiography (DSA), magnetic resonance imaging (MRI), functional MRI, dynamic contrast enhancement-MRI, diffusion MRI, spiral CT, cone beam computed tomography (CBCT), positron emission tomography (PET), single photon emission computed tomography (SPECT), X-ray imaging, optical tomography, fluorescence imaging, ultrasound imaging, and radiotherapy field imaging.

[0055] In some embodiments, the object analysis device 600 may be a dedicated intelligent device or a general intelligent device. For example, the object analysis device 600 may be a computer customized for image data acquisition and image data processing tasks, or a server placed in the cloud. For example, the device 600 is integrated into an image acquisition device.

[0056] The object analysis apparatus 600 may include a processor 601 and a memory 604 , and may further include at least one of an input / output 602 and an image display 603 .

[0057] The processor 601 may be a processing device that may include one or more general-purpose processing devices, such as a microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), etc. More specifically, the processor 601 may be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor that runs other instruction sets, or a processor that runs a combination of instruction sets. The processor 601 may also be one or more special-purpose processing devices, such as an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a system on a chip (SoC), etc. As will be appreciated by those skilled in the art, in some embodiments, the processor 601 may be a special-purpose processor rather than a general-purpose processor. The processor 601 may include one or more known processing devices, such as those from Intel. TM Pentium manufactured TM 、Core TM , Xeon TMor Itanium series of microprocessors, Turion manufactured by AMD TM TM , Athlon TM 、Sempron TM 、Opteron TM FX TM 、Phenom TM series, or various processors manufactured by Sun Microsystems. Processor 601 may also include a graphics processing unit, such as from GPUs by Nvidia TM Manufactured series, by Intel TM Made by GMA, Iris TM series, or by AMD TM Made by Radeon TM Processor 601 may also include an accelerated processing unit, such as an AMD TM Desktop A-4(6,6) series manufactured by Intel TM Xeon Phi TM Series. The disclosed embodiments are not limited to any type of processor or processor circuit, which is otherwise configured to acquire a 3D medical image containing an object; segment the 3D medical image to obtain a segmentation result of the object; acquire a set of image slices in the 3D medical image in an extended direction; acquire internal representative points of the segmented object in each image slice in the set of image slices; acquire a set of image blocks in the 3D medical image based on a set of internal representative points of the object in the set of image slices; perform object analysis based on the set of image blocks; or manipulate any other type of data consistent with the disclosed embodiments. In addition, the term "processor" or "image processor" may include more than one processor, for example, a multi-core design or multiple processors, each processor having a multi-core design. The processor 601 can execute a sequence of computer program instructions stored in the memory 604 to perform various operations, processes, and methods disclosed herein.

[0058] The processor 601 may be communicatively coupled to the memory 604 and configured to execute computer executable instructions stored therein. The memory 604 may include a read-only memory (ROM), a flash memory, a random access memory (RAM), a dynamic random access memory (DRAM) such as a synchronous DRAM (SDRAM) or a Rambus DRAM, a static memory (e.g., a flash memory, a static random access memory), etc., on which the computer executable instructions are stored in any format. In some embodiments, the memory 604 may store computer executable instructions of one or more image processing programs 605. The computer program instructions may be accessed by the processor 601, read from the ROM or any other suitable memory location, and loaded into the RAM for execution by the processor 601. For example, the memory 604 may store one or more software applications. The software applications stored in the memory 604 may include, for example, an operating system (not shown) for a general computer system and an operating system for a soft control device.

[0059] In addition, the memory 604 may store the entire software application or only a portion of the software application (e.g., image processing program 605) that can be executed by the processor 601. In addition, the memory 604 may store a plurality of software modules for implementing the various steps of the method for object analysis on a medical image consistent with the present disclosure or the process for training a sub-object analysis model, a slice classification model, and a segmentation model.

[0060] In addition, the memory 604 may store data generated / buffered when executing the computer program, for example, medical image data 606, which may include medical images sent from an image acquisition device, a medical image database 608, an image data storage device 609, etc. In some embodiments, the medical image data 606 may include a 3D medical image containing an object to be analyzed, and the image processing program 605 will segment it, obtain image slices, obtain internal representative points, crop image blocks, and perform object analysis.

[0061] In some embodiments, an image data storage device 609 may be provided to exchange image data with a medical image database 608, and the memory 604 may communicate with the medical image database 608 to obtain a medical image including a plurality of parts for which blood vessel segmentation is to be performed. For example, the image data storage device 609 may reside in other medical image acquisition devices (e.g., a CT scan performed on the patient). The medical image of the patient may be transmitted and stored in the medical image database 608, and the object analysis device 600 may obtain the medical image of a specific patient from the medical image database 608 and perform object analysis on the medical image of the specific patient.

[0062] In some embodiments, the memory 604 may communicate with the medical image database 608 to transmit and store the object segmentation result together with the obtained object analysis result into the medical image database 608 .

[0063] In addition, the parameters of the trained sub-object analysis model and / or slice classification model and / or segmentation model can be stored in the medical image database 608, so as to be accessed, acquired and used by other object analysis devices when needed. In this way, when facing a patient, the processor 601 can obtain the trained sub-object analysis model, slice classification model and / or segmentation model of the corresponding population, so as to perform blood vessel segmentation based on the obtained trained model.

[0064] In some embodiments, the sub-object analysis model, slice classification model and / or segmentation model (especially the learning network) can be stored in the memory 604. Optionally, the learning network can be stored in a remote device, a separate database (such as a medical image database 608), a distributed device, and can be used by the image processing program 605.

[0065] In addition to displaying medical images, the image display 603 can also display other information, such as object segmentation results, center point calculation results, and object analysis results. For example, the image display 603 can be an LCD, CRT, or LED display.

[0066] Input / output 602 may be configured to allow object analysis device 600 to receive and / or send data. Input / output 602 may include one or more digital and / or analog communication devices that allow the device to communicate with a user or other machines and devices. For example, input / output 602 may include a keyboard and mouse that allow a user to provide input.

[0067] In some embodiments, the image display 603 may present a user interface so that the user may conveniently and intuitively modify (such as edit, move, modify, etc.) the generated anatomical labels using the input / output 602 together with the user interface.

[0068] The interface 607 may include a network adapter, a cable connector, a serial connector, a USB connector, a parallel connector, a high-speed data transmission adapter such as an optical fiber, USB 6.0, Lightning, a wireless network adapter such as a Wi-Fi adapter, a telecommunications (6G, 4G / LTE, etc.) adapter. The device may be connected to a network through the interface 607. The network may provide a local area network (LAN), a wireless network, a cloud computing environment (e.g., software as a service, a platform as a service, an infrastructure as a service, etc.), a client-server, a wide area network (WAN), etc.

[0069] The embodiments of the present disclosure also provide a computer storage medium on which computer executable instructions are stored, and when the computer executable instructions can be executed by a processor, the method of performing object analysis on a medical image according to various embodiments of the present disclosure is implemented. The storage medium may include a read-only memory (ROM), a flash memory, a random access memory (RAM), a dynamic random access memory (DRAM) such as a synchronous DRAM (SDRAM) or a Rambus DRAM, a static memory (e.g., a flash memory, a static random access memory), etc., on which computer executable instructions can be stored in any format.

[0070] In addition, although exemplary embodiments have been described herein, the scope may include any and all embodiments based on the present disclosure with equivalent elements, modifications, omissions, combinations (e.g., various embodiments intersecting schemes), adaptations or changes. The elements in the claims are to be interpreted broadly based on the language adopted in the claims, and are not limited to the examples described in this specification or during the practice of the present disclosure, which examples are to be interpreted as non-exclusive. Therefore, this specification and examples are intended to be considered as examples only, and the true scope and spirit are indicated by the following claims and the full scope of their equivalents.

[0071] The above description is intended to be illustrative rather than restrictive. For example, the above examples (or one or more of them) can be used in combination with each other. For example, a person of ordinary skill in the art can use other embodiments when reading the above description. In addition, in the above-mentioned specific embodiments, various features can be grouped together to simplify the present disclosure. This should not be interpreted as an intention that a disclosed feature that is not required to be protected is necessary for any claim. On the contrary, the subject matter of the present disclosure may be less than all the features of a specific disclosed embodiment. Thus, the following claims are incorporated into the specific embodiments as examples or embodiments, wherein each claim is independently used as a separate embodiment, and it is considered that these embodiments can be combined with each other in various combinations or arrangements. The scope of the present invention should be determined with reference to the attached claims and the full scope of equivalent forms granted by these claims.

[0072] The above embodiments are only exemplary embodiments of the present disclosure and are not intended to limit the present invention. The protection scope of the present invention is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present invention within the essence and protection scope of the present disclosure, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present invention.

Claims

1. A method for object analysis of medical images, It is characterized in that include: Acquire a 3D medical image containing an object; Using a processor, dividing the 3D medical image into sub-image sequences of each part according to the part; Setting corresponding window widths and window levels for each type of object, and adjusting windows for each sub-image sequence based on each window width and window level to obtain a sub-image sequence for each channel; Determine the size of the sliding window block based on the prior information of each part, determine the internal representative point of the sliding window block based on the object segmentation result of each sub-image sequence after skeletonization, intercept the sliding window block of the training sample containing the lesion annotation information according to the size of the sliding window block based on the internal representative point, train each sub-object analysis model using the sliding window block as the training sample, and adjust the model parameters for each sub-object analysis model; Based on the sub-image sequence of each channel, referring to the prior information of each part, using the sub-object analysis model corresponding to each part to perform analysis, the sub-object analysis result is obtained; The processor is used to fuse the analysis results of each sub-object to obtain the object analysis result of the 3D medical image.

2. The method according to claim 1, It is characterized in that Dividing the 3D medical image into sub-image sequences of each part according to the parts specifically includes: based on the 3D medical image, using a slice classification model, identifying key slices in the 3D medical image that are the junctions of adjacent parts; using the identified key slices to achieve the division of sub-images according to the parts.

3. The method according to claim 2, It is characterized in that The slice classification model is implemented by using a two-dimensional learning network and is trained by using training samples with classification information of slices at corresponding parts.

4. The method according to claim 1, It is characterized in that The object is at least one of blood vessels, digestive tract, mammary duct, respiratory tract or lesions therein.

5. The method according to claim 1, It is characterized in that The object is a vascular lesion.

6. The method according to claim 5, It is characterized in that The vascular lesion is at least one of a calcified plaque, a non-calcified plaque, a mixed plaque, an aneurysm and a stent image.

7. The method according to any one of claims 1 to 6, It is characterized in that Based on the sub-image sequence of each channel, the sub-object analysis model corresponding to each part is used for analysis to obtain the sub-object analysis result, which specifically includes: based on the sub-image sequence of each channel, referring to the prior information of each part and the object segmentation result after skeletonization, the sub-object analysis model corresponding to each part is used for analysis to obtain the sub-object analysis result.

8. The method according to claim 1, It is characterized in that The method of determining the internal representative points of the sliding window block based on the object segmentation results of each sub-image sequence after skeletonization specifically includes: Based on the sub-image sequence of each part, using the processor, using the corresponding segmentation model of each part to determine the corresponding object segmentation result; Perform skeletonization on the object segmentation results of each sub-image sequence: The skeletonized object segmentation result is sparsely sampled to obtain the internal representative points of the sliding window block.

9. The method according to claim 8, It is characterized in that Each blood vessel segmentation model is trained separately using training samples with blood vessel information of corresponding parts.

10. The method according to claim 1, It is characterized in that Based on the prior information of each part and the object segmentation result after skeletonization, adjusting the model parameters for each sub-object analysis model also specifically includes: Use sliding window blocks as training samples to train each sub-object analysis model; The false positive samples obtained from the training are used together with the training samples containing lesion annotation information as new training samples to train each sub-object analysis model.

11. The method according to any one of claims 8 to 10, It is characterized in that The prior information of the part includes at least one of the size, shape and quantity of objects contained in the part.

12. The method according to any one of claims 8 to 10, It is characterized in that The internal representative point is the center point of the sliding window block.

13. The method according to any one of claims 1 to 6, It is characterized in that The 3D medical image is a CTA image containing blood vessels, a CT image containing ribs, or a CT image containing lungs.

14. A device for performing object analysis on a medical image, It is characterized in that include: an interface configured to acquire a 3D medical image containing an object; and a processor configured to: execute the method for object analysis on a medical image according to any one of claims 1-13.

15. A non-transitory computer-readable medium having instructions stored thereon, which when executed by a processor implement the method for object analysis of a medical image according to any one of claims 1-13.

Images