Bone scan image annotation method, device, medium and product

Abstract

The present invention discloses a bone scan image annotation method, device, medium and product, which relates to the field of image processing. The method comprises: obtaining a bone scan image to be annotated of a target object; preprocessing the bone scan image to obtain a processed bone scan image; determining a preliminary annotation result of the bone scan image using a bone scan image annotation model according to the processed bone scan image; wherein the bone scan image annotation model is obtained by training an image segmentation network using a training data set; the training data set includes a bone scan image of a training object and a corresponding label; the image segmentation network includes but is not limited to a fully convolutional neural network and a U‑Net network; post-processing the preliminary annotation result of the bone scan image to obtain a bone scan image annotation result. The present invention improves the annotation efficiency of bone scan images.

Description

Technical Field

[0001] The present invention relates to the field of image annotation, and in particular to a bone scan image annotation method, device, medium and product. Background Art

[0002] Medical imaging technology is a technology that uses different methods and equipment to obtain information about the internal structure and function of the human body. It plays an important role in clinical diagnosis, treatment and research, helping doctors understand pathological changes, formulate treatment plans and evaluate treatment effects.

[0003] Nuclear medicine imaging is a special type of medical imaging technology that uses radioactive isotopes or tracers injected into the body, then captures the photons or positrons released by radioactive decay through detectors and reconstructs them into images through computers. Nuclear medicine imaging can provide information about organ function, metabolism and molecular levels, and is often used in cancer diagnosis, cardiovascular disease assessment and neurological disease research.

[0004] Single photon emission computed tomography (SPECT) and positron emission computed tomography (PET) are two common nuclear medicine imaging technologies. However, SPECT is more affordable than PET, with lower costs and more widespread use worldwide. Many medical centers are equipped with SPECT equipment. At the same time, SPECT can use a variety of radioactive isotopes for imaging. Each isotope has different characteristics and uses. The most suitable isotope can be selected as needed to meet different clinical needs.

[0005] Bone scan images provide information about organ function and metabolism in clinical diagnosis. Doctors usually determine the location and extent of lesions by examining the intensity distribution of radioisotopes in different areas of the image, and determine possible diseases by the morphological structure of the lesion location. At the same time, it is necessary to judge the relationship between the lesion location and the surrounding tissues and refer to the results of other imaging technologies (such as CT or MRI) to determine whether there are abnormal lesions. Traditional diagnostic methods rely on the professional knowledge of doctors and have certain subjective factors. In recent years, computer-aided diagnosis technology based on bone scan images has received attention from the industry and has been studied to a certain extent. Since automated lesion segmentation can provide doctors with support in quantitative analysis, treatment planning, and disease monitoring, automated lesion segmentation based on SPECT images has received widespread attention. However, SPECT image annotation requires operators to not only master professional nuclear medicine knowledge, but also learn professional computer annotation programs, which requires operators to pay a lot of learning costs and cannot be carried out in large quantities, resulting in low annotation efficiency.

[0006] The development of interactive annotation based on SPECT images is difficult due to the lack of public research data and the large differences in data between different data centers. Summary of the invention

[0007] The purpose of the present invention is to provide a bone scan image annotation method, device, medium and product to improve the annotation efficiency of bone scan images.

[0008] To achieve the above object, the present invention provides the following solutions:

[0009] A bone scan image annotation method, comprising:

[0010] Acquire a bone scan image to be annotated of a target object;

[0011] Preprocessing the bone scan image to obtain a processed bone scan image; the preprocessing includes noise reduction, contrast enhancement and normalization processing;

[0012] According to the processed bone scan image, a preliminary annotation result of the bone scan image is determined using a bone scan image annotation model; wherein the bone scan image annotation model is obtained by training an image segmentation network using a training data set; the training data set includes a bone scan image of a training object and a corresponding label;

[0013] The preliminary annotation result of the bone scan image is post-processed to obtain the bone scan image annotation result; the post-processing includes morphological operation, connected component labeling algorithm and pixel-by-pixel labeling technology.

[0014] Optionally, obtaining a bone scan image to be annotated specifically includes:

[0015] Using a nuclide detector to capture nuclide count values ​​outside the body of the target object;

[0016] The bone scan image is determined according to the radionuclide count value.

[0017] Optionally, preprocessing the bone scan image to obtain a processed bone scan image specifically includes:

[0018] Performing noise reduction processing on the bone scan image to obtain a noise-reduced bone scan image;

[0019] Performing contrast enhancement processing on the bone scan image to obtain an enhanced bone scan image;

[0020] The enhanced bone scan image is normalized to obtain a processed bone scan image.

[0021] Optionally, according to the processed bone scan image, using a bone scan image annotation model, determining a preliminary annotation result of the bone scan image specifically includes:

[0022] marking a region of interest of the processed bone scan image to obtain a bone scan image with the region of interest marked;

[0023] A preliminary annotation result of the bone scan image is determined based on the bone scan image with the annotated region of interest using a bone scan image annotation model.

[0024] Optionally, post-processing the preliminary annotation result of the bone scan image to obtain the bone scan image annotation result specifically includes:

[0025] Filling holes in the preliminary annotation result of the bone scan image using the morphological operation to obtain a filled annotation result;

[0026] Determine the maximum connected domain of the filled annotation result by using the connected component labeling algorithm;

[0027] The maximum connected domain of the filled annotation result is corrected by using a pixel-by-pixel labeling technique to correct edge pixels or wrongly labeled pixels to obtain a bone scan image annotation result.

[0028] A computer device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-mentioned bone scan image annotation method.

[0029] A computer-readable storage medium stores a computer program, which implements the steps of the above-mentioned bone scan image annotation method when executed by a processor.

[0030] A computer program product includes a computer program, which implements the steps of the above-mentioned bone scan image annotation method when executed by a processor.

[0031] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0032] The present invention provides a bone scan image annotation method, device, medium and product, firstly obtaining a target object's bone scan image to be annotated; preprocessing the bone scan image to obtain a processed bone scan image; according to the processed bone scan image, using a bone scan image annotation model, determining a preliminary annotation result of the bone scan image; wherein the bone scan image annotation model is obtained by training an image segmentation network using a training data set; the training data set includes a bone scan image of a training object and corresponding labels; the image segmentation network is a fully convolutional neural network or a U-Net network; post-processing the preliminary annotation result of the bone scan image to obtain a bone scan image annotation result. The present invention improves the annotation efficiency of bone scan images. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 A schematic diagram of a bone scan image annotation method flow chart provided in Example 1 of the present invention;

[0035] Figure 2 It is a schematic diagram of the flow chart of the bone scan image annotation method of the present invention in practical application;

[0036] Figure 3 This is a schematic diagram of the labeling results of Example 1 of the present invention;

[0037] Figure 4 This is a diagram of the internal structure of a computer device. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] The purpose of the present invention is to provide a bone scan image annotation method, device, medium and product to improve the annotation efficiency of bone scan images.

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] Example 1

[0042] like Figure 1 and Figure 2 As shown, the bone scan image annotation method in this embodiment includes:

[0043] Step 101: Obtain a bone scan image of a target object to be labeled.

[0044] As an optional implementation, step 101 specifically includes:

[0045] The nuclide detector is used to capture the nuclide count value outside the target object.

[0046] The bone scan image is determined according to the radionuclide count value.

[0047] In practical applications, bone scan image data is imported to obtain a bone scan image data matrix, specifically including:

[0048] Step 1.1: Use a radionuclide detector to capture radionuclide count values ​​outside the patient's body, display them as bone scan images with the help of computer-assisted equipment, and store the bone scan images as DICOM files.

[0049] Step 1.2: According to the DICOM 3.0 protocol standard, divide and extract the data matrix representing the bone scan image in the DICOM file.

[0050] Step 1.3: Provide multiple reading methods for different types of bone scan image formats.

[0051] Step 102: preprocessing the bone scan image to obtain a processed bone scan image; the preprocessing includes noise reduction, contrast enhancement and normalization processing.

[0052] As an optional implementation, step 102 specifically includes:

[0053] The bone scan image is subjected to noise reduction processing to obtain a noise-reduced bone scan image. In practical applications, computer technologies such as threshold noise reduction, smoothing filtering, and wavelet denoising are used to perform noise reduction processing on imaging noise.

[0054] The bone scan image is subjected to contrast enhancement processing to obtain an enhanced bone scan image. In practical applications, contrast enhancement methods including but not limited to histogram equalization, adaptive histogram equalization, contrast stretching and other computer technologies are used to enhance the contrast of the bone scan image and increase the dynamic range of the bone scan image.

[0055] The enhanced bone scan image is normalized to obtain a processed bone scan image. In practical applications, the element spacing of the bone scan data matrix is ​​detected, and the data resampling technology is used to perform element spacing normalization on the bone scan image data matrix.

[0056] Performing noise reduction, contrast enhancement, and resampling operations on the bone scan image data matrix can eliminate image noise and enhance image details.

[0057] Step 103: According to the processed bone scan image, a preliminary annotation result of the bone scan image is determined using a bone scan image annotation model; wherein the bone scan image annotation model is obtained by training an image segmentation network using a training data set; the training data set includes bone scan images of training objects and corresponding labels; the image segmentation network includes but is not limited to a fully convolutional neural network and a U-Net network.

[0058] As an optional implementation, step 103 specifically includes:

[0059] The region of interest of the processed bone scan image is marked to obtain a bone scan image with the region of interest marked.

[0060] A preliminary annotation result of the bone scan image is determined based on the bone scan image with the annotated region of interest using a bone scan image annotation model.

[0061] In practical applications, model training is as follows:

[0062] Step 3.1: Obtain a bone scan image without position information and annotation information.

[0063] Step 3.2: Use the above collected data to train image segmentation networks, including but not limited to FCN networks, U-Net and other training lesion segmentation and annotation models.

[0064] Step 3.3: Get the model training weights and use them as the bone scan image annotation model.

[0065] In actual application, the bone scan image is enlarged, moved and pre-processed in step 102 to find the marked area through the interactive operation platform.

[0066] Manual guidance information is provided to the marked area to obtain the region of interest of the bone scan image.

[0067] The obtained bone scan image region of interest is filled or cropped to meet the model input requirements.

[0068] The processed bone scan image lesion region of interest (bone scan image with annotated region of interest) is sent to the network model (bone scan image annotation model) to obtain the preliminary annotation result of the lesion.

[0069] Step 104: post-processing the preliminary annotation result of the bone scan image to obtain a bone scan image annotation result; the post-processing includes morphological operation, connected component labeling algorithm and pixel-by-pixel labeling technology.

[0070] As an optional implementation, step 104 specifically includes:

[0071] Step 4.1: Use the morphological operation to fill the holes in the preliminary annotation results of the bone scan image to obtain the filled annotation results. In practical applications, morphological operations (erosion, expansion, opening operation, closing operation, etc.) are used to fill the holes in the model annotation results.

[0072] Step 4.2: Determine the maximum connected domain of the filled annotation result using the connected component labeling algorithm. In practical applications, the connected component labeling algorithm is used to obtain the maximum connected domain of the model annotation result.

[0073] Step 4.3: Correct edge pixels or wrongly labeled pixels using pixel-by-pixel labeling technology for the largest connected domain of the filled labeling result to obtain the bone scan image labeling result. In practical applications, the edge pixels or wrongly labeled pixels are manually corrected using pixel-by-pixel labeling technology for the result obtained in step 4.2.

[0074] Repeat the above steps to obtain accurate bone scan image lesion annotation results, such as Figure 3 shown.

[0075] Save the bone scan image lesion annotation results. The saving formats include but are not limited to ".nii.gz", ".npy", ".png", etc.

[0076] Manually labeling bone scan image lesions requires operators to not only have extensive knowledge in the field of nuclear medicine, but also to master the relevant techniques of using computers to label lesions. The interactive labeling method for low-resolution large-size bone scan images (bone scan image labeling method) provided by the present invention can achieve lesion labeling by simply providing simple guidance information by the operator, which greatly reduces the learning and communication costs of the operator; in view of the imaging differences caused by imaging equipment in different data centers and the existence of bone scan images with poor imaging, a complete data preprocessing operation is provided to eliminate imaging differences and enhance image quality to a certain extent, providing identifiable high-quality bone scan images for subsequent manual guidance information; in view of the possible false positives or omissions of lesions in deep learning models, a manual guidance method is used to determine the region of interest, accurately locate the lesions and complete the labeling, and obtain preliminary labels. annotation results; in view of the problems such as holes and incomplete edges that may exist in the above preliminary annotation results, morphological methods are used to fill and smooth the holes and incomplete edges to obtain complete annotation results, and the connected component labeling algorithm is used to obtain the maximum connected domain to obtain the true annotation information of the lesion; due to differences in training data and environment, the above annotation results may have lesion edges that are inconsistent with the diagnosis of professional physicians, that is, the area of ​​the annotation results is inconsistent with the actual lesion area. The manual pixel-by-pixel modification technology is used to modify the annotation information to make it consistent with the actual lesion size; in view of the different formats of lesion annotation required by different researchers and users, various formats of storage formats are provided to realize the preservation of annotation information.

[0077] In summary, an interactive annotation method for low-resolution large-size bone scan images can realize bone scan image lesion annotation through simple manual guidance information, which not only greatly reduces the human resource requirements required for manual annotation and improves the efficiency of image annotation, but also provides real and reliable labels for the use of deep learning technology to realize lesion segmentation of bone scan images.

[0078] Example 2

[0079] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the bone scan image annotation method in embodiment 1.

[0080] Example 3

[0081] A computer-readable storage medium stores a computer program, which implements the steps of the bone scan image annotation method in embodiment 1 when executed by a processor.

[0082] Example 4

[0083] A computer program product includes a computer program, which implements the steps of the bone scan image annotation method in embodiment 1 when executed by a processor.

[0084] Example 5

[0085] A computer device, which may be a database, may have an internal structure as shown in Figure 4 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store pending transactions. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the bone scan image annotation method in Example 1 is implemented.

[0086] It should be noted that the object information (including but not limited to object device information, object personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data authorized by the object or fully authorized by all parties, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of relevant countries and regions.

[0087] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided by the present invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided by the present invention may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited thereto. The processor involved in each embodiment provided by the present invention may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited thereto.

[0088] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0089] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A bone scan image annotation method, characterized in that: The method comprises: Acquire a bone scan image to be annotated of a target object; Preprocessing the bone scan image to obtain a processed bone scan image; the preprocessing includes noise reduction, contrast enhancement and normalization processing; According to the processed bone scan image, a preliminary annotation result of the bone scan image is determined using a bone scan image annotation model; wherein the bone scan image annotation model is obtained by training an image segmentation network using a training data set; the training data set includes a bone scan image of a training object and a corresponding label; According to the processed bone scan image, using the bone scan image annotation model, determining the preliminary annotation result of the bone scan image, specifically including: Marking the region of interest of the processed bone scan image to obtain the bone scan image with the marked region of interest specifically includes: The processed bone scan image is enlarged and moved to find the marked area through the interactive operation platform; Providing manual guidance information to the marked area to obtain the region of interest of the bone scan image, thereby obtaining a bone scan image with the marked region of interest; Determine a preliminary annotation result of the bone scan image by using a bone scan image annotation model according to the bone scan image with the annotated region of interest; Post-processing the preliminary annotation results of the bone scan image to obtain the bone scan image annotation results; the post-processing includes morphological operations, connected component labeling algorithms and pixel-by-pixel labeling technology; Post-processing the preliminary annotation results of the bone scan image to obtain the bone scan image annotation results, specifically including: Filling holes in the preliminary annotation result of the bone scan image using the morphological operation to obtain a filled annotation result; Determine the maximum connected domain of the filled annotation result by using the connected component labeling algorithm; The maximum connected domain of the filled annotation result is corrected by using a pixel-by-pixel labeling technique to correct edge pixels or wrongly labeled pixels to obtain a bone scan image annotation result.

2. The bone scan image annotation method according to claim 1, characterized in that: Obtain the bone scan image to be annotated, including: Using a nuclide detector to capture nuclide count values ​​outside the body of the target object; The bone scan image is determined according to the radionuclide count value.

3. The bone scan image annotation method according to claim 1, characterized in that: Preprocessing the bone scan image to obtain a processed bone scan image specifically includes: Performing noise reduction processing on the bone scan image to obtain a noise-reduced bone scan image; Performing contrast enhancement processing on the bone scan image to obtain an enhanced bone scan image; The enhanced bone scan image is normalized to obtain a processed bone scan image.

4. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the bone scan image annotation method according to any one of claims 1 to 3.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the bone scan image annotation method according to any one of claims 1 to 3 are implemented.

6. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the bone scan image annotation method according to any one of claims 1 to 3 are implemented.

Images