CT image lung nodule segmentation detection method, system, computer and storage medium

Abstract

The application discloses a CT image lung nodule segmentation detection method, system, computer and storage medium. The method comprises the following steps: acquiring a three-dimensional CT image of a lung to be detected, and performing noise reduction processing; performing prediction segmentation on all two-dimensional images in the three-dimensional CT image after the noise reduction processing, forming a mask picture and combining to form a predicted three-dimensional image; calculating the center points of all connected domains in the predicted three-dimensional image and calibrating as candidate nodule center points, and acquiring the three-dimensional space coordinates of the candidate nodule center points; combining the three-dimensional CT image after the noise reduction processing and the three-dimensional space coordinates of all candidate nodule center points, and generating a candidate lung nodule image; performing 3D CNN false positive filtering on the candidate lung nodule image, and outputting all true positive lung nodule three-dimensional space coordinates. The application has the beneficial effects of improving the accuracy of the lung nodule segmentation task, and in the lung nodule image detection process, the lung nodule judgment accuracy is high, the burden on medical personnel can be reduced, and the diagnosis speed can be accelerated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image data processing, in particular to a CT image lung nodule segmentation detection method and system, a computer and a storage medium. BACKGROUND

[0002] Lung cancer is one of the highest incidence and mortality rates in the world, and the 5-year survival rate of lung cancer is low, but early detection and treatment of lung cancer can greatly reduce mortality.

[0003] The detection of lung nodules is of great significance for the whole section of lung cancer, because lung nodules generally appear in the early stage of lung cancer. Irregular and uncontrollable growth of lung parenchymal cells is an important reason for the formation of lung nodules, but because lung nodules are generally small, and their irregular visual features make the segmentation of lung nodules a great challenge. Computed tomography (CT) is widely used in lung nodule detection and analysis, and CT screening can significantly reduce severe and fatal cases of lung cancer.

[0004] However, the current lung nodule segmentation detection method has many defects, the main defects are: (1) the screening method based on artificial analysis is time-consuming and laborious, and is severely dependent on the level of doctors. (2) The method of segmenting and detecting lung nodules based on deep learning, the existing U-Net and its improved methods, etc. perform poorly on this task. SUMMARY

[0005] Therefore, one purpose of the present application is to provide a CT image lung nodule segmentation detection method, system, computer and storage medium, which can automatically segment and detect the position of lung nodules, reduce the burden of doctors, and speed up the diagnosis of the disease.

[0006] The CT image lung nodule segmentation detection method provided by the embodiment of the present application comprises the following steps:

[0007] Obtain a three-dimensional CT image of a lung to be detected, and perform noise reduction processing on the three-dimensional CT image;

[0008] Perform predictive segmentation on all two-dimensional images in the three-dimensional CT image after noise reduction processing to form a mask picture;

[0009] Combine the mask picture to form a predicted three-dimensional image;

[0010] Calculate the center points of all connected domains in the predicted three-dimensional image using an image operation function;

[0011] Mark the center points as candidate node centers, and obtain the three-dimensional space coordinates of the candidate node centers;

[0012] combine the three-dimensional CT image after noise reduction processing and the three-dimensional spatial coordinates of all candidate nodule centers to generate a candidate lung nodule image;

[0013] perform 3D CNN false positive filtering on the candidate lung nodule image, and output all true positive lung nodule three-dimensional spatial coordinates.

[0014] The CT image lung nodule segmentation detection method provided by the application mainly includes the following steps: three-dimensional image input, image preprocessing, two-dimensional segmentation of the image to obtain the three-dimensional spatial coordinates of all candidate lung nodule nodes, generation of a candidate lung nodule image, and then 3D CNN false positive filtering to output all true positive lung nodule coordinates. The beneficial effects are: compared with the traditional U-Net and its improvement, the two-dimensional segmentation method greatly improves the accuracy in the lung nodule segmentation task. In the lung nodule image detection process, the overall operation is simple, the lung nodule determination accuracy is high, the burden on medical personnel can be reduced, and the diagnosis speed can be accelerated.

[0015] In addition, the CT image lung nodule segmentation detection method provided by the application can also have the following additional technical features:

[0016] Further, the step of performing noise reduction processing on the three-dimensional CT image comprises:

[0017] reading the CT scan value of each part in the three-dimensional CT image;

[0018] intercepting the CT scan value of each part in the three-dimensional CT image according to a preset interval value, and performing normalization processing on the intercepted region.

[0019] Further, before the step of performing prediction segmentation on all two-dimensional images in the three-dimensional CT image after noise reduction processing to form a mask picture combination, the method further comprises a two-dimensional segmentation model training step, which comprises:

[0020] selecting the LUNA16 dataset, which includes 888 CTs, and selecting 1186 nodules labeled in the annotations.csv file;

[0021] According to each label information, select a slice image with a lung nodule as a center point in a three-dimensional space;

[0022] generate a lung nodule mask using the lung nodule center coordinates and radius information to obtain a dataset;

[0023] Divide the dataset into a training set, a validation set and a test set according to a ratio of 8:1:1 for model training.

[0024] Further, all two-dimensional images in the three-dimensional CT image after noise reduction processing are predicted to be segmented to form a mask picture by using a U-net network improved by a Floor structure and an attention mechanism.

[0025] Further, the step of combining the three-dimensional CT image after noise reduction processing and three-dimensional spatial coordinates of all candidate nodule centers to generate a candidate lung nodule image specifically comprises:

[0026] A 48*48*48 cube is intercepted in the three-dimensional CT image after noise reduction processing with the three-dimensional spatial coordinates of the candidate nodule center as a center point, and the cube is the candidate lung nodule image.

[0027] Further, the step of filtering the candidate lung nodule image by 3D CNN false positive and outputting all true positive lung nodule three-dimensional spatial coordinates specifically comprises:

[0028] The candidate lung nodule image is input into a 3D CNN false positive filtering network for filtering;

[0029] When the output is a false positive result, the coordinates of the candidate lung nodule at the corresponding position in the candidate lung nodule image are discarded;

[0030] When the output is a true positive result, the coordinates of the candidate lung nodule at the corresponding position in the candidate lung nodule image are retained.

[0031] The application further provides a CT image lung nodule segmentation detection system, which comprises:

[0032] An image input module is used to acquire a three-dimensional CT image of a lung to be detected;

[0033] A preprocessing module is used to perform noise reduction processing on the three-dimensional CT image;

[0034] A prediction segmentation module is used to perform prediction segmentation on all two-dimensional images in the three-dimensional CT image after noise reduction processing to form a mask picture;

[0035] An image combination module is used to combine the mask picture to form a predicted three-dimensional image;

[0036] A calculation module is used to calculate the center points of all connected domains in the predicted three-dimensional image by using an image operation function;

[0037] A coordinate acquisition module is used to mark the center points as candidate nodule centers and acquire three-dimensional spatial coordinates of the candidate nodule centers;

[0038] An image generation module is configured to combine the three-dimensional CT image after the noise reduction processing and the three-dimensional spatial coordinates of all candidate nodule centers to generate a candidate lung nodule image.

[0039] A false positive filtering module is configured to perform 3D CNN false positive filtering on the candidate lung nodule image and output all true positive lung nodule three-dimensional spatial coordinates.

[0040] Further, the preprocessing module specifically comprises:

[0041] A reading unit is configured to read CT scan values of each part in the three-dimensional CT image.

[0042] A clipping unit is configured to clip the CT scan values of each part in the three-dimensional CT image according to a preset interval value.

[0043] A normalization unit is configured to perform normalization processing on the clipping region.

[0044] In addition, an embodiment of the present application provides a computer device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the CT image lung nodule segmentation detection method according to the first aspect.

[0045] In addition, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the program is executable on a processor to implement the CT image lung nodule segmentation detection method according to the first aspect.

[0046] Additional aspects and advantages of the application will be made apparent by the following description and the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0047] The above and / or additional aspects and advantages of the application will become apparent and be readily understood by considering the following detailed description, including the accompanying drawings, in which:

[0048] Figure 1 is a CT image lung nodule segmentation detection method flowchart according to an embodiment of the present application;

[0049] Figure 2 is a method flowchart for performing noise reduction processing on the three-dimensional CT image according to an embodiment of the present application;

[0050] Figure 3 is a structure diagram of the U-net improved based on the Floor structure and the attention mechanism according to an embodiment of the present application;

[0051] Figure 4is a method flowchart for filtering false positives of candidate lung nodule images and outputting true positive lung nodules in the first embodiment of the present application.

[0052] Figure 5 is a structure schematic diagram of a CT image lung nodule segmentation detection system in the second embodiment of the present application. DETAILED DESCRIPTION

[0053] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application is described and explained below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0054] Obviously, the drawings described below are only some examples or embodiments of the present application, and for those of ordinary skill in the art, the present application can be applied to other similar scenarios without creative labor on the basis of these drawings. In addition, it can be understood that although the efforts made in this development process can be complex and lengthy, for those of ordinary skill in the art related to the content disclosed in the present application, some design, manufacture or production changes on the basis of the technical content disclosed in the present application are only routine technical means and should not be understood as insufficient disclosure of the present application.

[0055] In the present application, "embodiment" means that the specific features, structures or characteristics described in combination with the embodiment can be contained in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily mean the same embodiment, nor is it an independent or alternative embodiment that is not mutually exclusive with other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in the present application can be combined with other embodiments without conflict.

[0056] Unless otherwise defined, technical terms and scientific terms used in the present application shall have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terms "a", "an", "one", "this", and similar referents in the context of describing the application are to be construed to be inclusive, not exclusive. For example, the use of the term "comprises" or "comprising" or "includes" or "including" or "has" or "having" or "contains" or "containing" or "consists" or "consisting" or "consists of" or "consisting of" to describe certain steps or modules (units) of the processes, methods, systems, products, or devices described herein, is intended to mean that the processes, methods, systems, products, or devices can consist of, but are not limited to, the listed steps or modules (units), and can also include other steps or modules (units) not listed, or can also include other steps or modules (units) inherent to the processes, methods, systems, products, or devices. The terms "connected", "coupled", or "linked" and the like, unless otherwise defined, are not limited to physical or mechanical connections or linkages, but can include electrical connection, whether direct or indirect. The term "plurality" means two or more. The term "and / or" describes association between name objects. For example, "A and / or B" can mean A alone, A and B together, or B alone. The character " / " generally means "or". The terms "first", "second", "third", and the like, merely distinguish similar objects, and do not represent a specific order.

[0057] The detection of lung nodules is of great importance to the whole section of lung cancer, because lung nodules generally accompany the initial stage of lung cancer. Irregular and uncontrollable growth of lung parenchymal cells is an important reason for the formation of lung nodules, but due to the small size of lung nodules and their irregular visual features, the segmentation of lung nodules poses a great challenge. Computed tomography (CT) is widely used in lung nodule detection and analysis, and CT screening can significantly reduce the number of severe and fatal cases of lung cancer.

[0058] However, the current lung nodule segmentation detection method has many defects, the main defects are: (1) the screening method based on artificial analysis is time-consuming and laborious, and is severely dependent on the level of doctors. (2) The method of segmenting and detecting lung nodules based on deep learning, such as U-Net and its improved method, performs poorly on this task.

[0059] Therefore, the present application provides a CT image lung nodule segmentation detection method to overcome the problems in the prior art.

[0060] Please refer to Figure 1 A CT image lung nodule segmentation detection method is provided for the first embodiment of the present application. Specifically, the CT image lung nodule segmentation detection method comprises the following steps:

[0061] Step S11, a three-dimensional CT image of a lung to be detected is acquired, and noise reduction processing is performed on the three-dimensional CT image.

[0062] It can be understood that generally, the CT image will be disturbed by noise in the acquisition process, such as quantum noise of X-rays itself, electronic noise of the detector and the like, and the existence of these noises will reduce the quality of the reconstructed image, and then affect the diagnosis of the disease by the doctor. In order to ensure the quality of the reconstructed image, the image is generally subjected to noise reduction processing.

[0063] Step S12, performing prediction segmentation on all two-dimensional images in the three-dimensional CT image after noise reduction processing to form a mask picture.

[0064] In the embodiment of the application, the two-dimensional prediction segmentation model is used for prediction, and the lung nodule image in each two-dimensional image is formed into a mask picture.

[0065] Compared with the traditional U-Net and its improvement, the two-dimensional prediction segmentation model in the embodiment of the application greatly improves the accuracy in the lung nodule segmentation task.

[0066] Step S13, combining the mask picture to form a predicted three-dimensional image.

[0067] Step S14, using an image operation function to calculate the center point of all connected domains in the predicted three-dimensional image.

[0068] In the embodiment of the application, the image operation function is specifically realized by using the function in the scipy.ndimage (multi-dimensional image processing package) library.

[0069] Step S15, marking the center point as a candidate nodule center, and acquiring the three-dimensional space coordinates of the candidate nodule center.

[0070] Step S16, combining the three-dimensional CT image after noise reduction processing and the three-dimensional space coordinates of all candidate nodule centers to generate a candidate lung nodule image.

[0071] Step S17, performing 3D CNN false positive filtering on the candidate lung nodule image, and outputting all true positive lung nodule three-dimensional space coordinates.

[0072] In summary, the CT image lung nodule segmentation detection method provided by the application obtains the three-dimensional space coordinates of all candidate lung nodule nodes through three-dimensional image input, image preprocessing and image two-dimensional segmentation, generates a candidate lung nodule image, and then performs 3D CNN false positive filtering to output all true positive lung nodule coordinates. The beneficial effects are: compared with the traditional U-Net and its improvement, the two-dimensional segmentation method improves the accuracy of the lung nodule segmentation task. In the lung nodule image detection process, the overall operation is simple, the lung nodule determination accuracy is high, the burden on medical personnel can be reduced, and the diagnosis speed can be accelerated.

[0073] Further, please refer to Figure 2 For the CT image lung nodule segmentation detection method provided in Embodiment One of the application, the method flow chart for the noise reduction processing of the three-dimensional CT image specifically includes the following steps:

[0074] Step S21, reading the CT scan value of each part in the three-dimensional CT image.

[0075] Step S22, intercepting the CT scan value of each part in the three-dimensional CT image according to a preset interval value, and performing normalization processing on the intercepted region.

[0076] By way of example but not limitation, in order to achieve good lung nodule structure determination, in the embodiment of the application, the setting interval range of the CT scan value (HU value in medicine) is specifically [-1000, 400]. This can achieve accurate scanning of the lung nodule in the image.

[0077] In summary, by filtering and scanning the lung nodule according to the HU value of the image, the characteristics of the region where the lung nodule exists are normalized, and the region where the lung nodule does not exist is zeroed, so as to achieve the purpose of inputting the image into the deep learning model for data processing.

[0078] Further, the model training step of the two-dimensional prediction segmentation model in the embodiment of the application specifically includes:

[0079] The LUNA16 data set is selected, which includes 888 CTs in the data set, and 1186 nodes labeled in the annotations.csv file are selected.

[0080] According to each label information, a slice image with the lung nodule as the center point in the three-dimensional space is selected.

[0081] The lung nodule mask is generated by using the lung nodule center coordinates and radius information, and the data set is obtained.

[0082] The data set is divided into a training set, a validation set and a test set according to a ratio of 8:1:1 for model training.

[0083] Further, in the embodiment of the present application, all two-dimensional images in the three-dimensional CT image after noise reduction processing are predicted and segmented to form a mask picture, which is realized by using a U-net network improved by a Floor structure and an attention mechanism.

[0084] It can be understood that in the convolutional neural network, the low-resolution information obtained after multiple downsampling can provide the global context semantic information of the segmentation target in the whole image, learn a large number of high-level semantic features of the image, and this feature is helpful for the category judgment of the object. Therefore, in the traditional classification problem, the traditional convolutional neural network only uses the high-level semantic feature to guide the model to complete the classification task, and does not use the shallow semantic feature.

[0085] However, the traditional U-Net does not make good use of the high-level semantic feature, but only uses it as the starting point of the expansion path. Please refer to 3, which is a structure diagram of the U-net improved by the Floor structure and the attention mechanism in the embodiment of the present application. In view of this situation, the embodiment of the present application proposes a structure for fusing high-level semantic features into each decoder of the U-Net to guide segmentation. Since the feature layer is located at the bottom of the U-Net, it is called Floor structure.

[0086] Just like when humans observe objects, they will pay attention to those more important features or visual information, so when designing the model, the trained network model should also pay attention to those more important features. The core idea of the attention mechanism is to let the designed network model pay attention to the places in the picture that need more attention, and the attention mechanism is a way to realize adaptive attention. The attention mechanism used in this paper is based on this method of fusing channel attention and spatial attention, and the formula can be expressed as:

[0087] res=in*CA(in);

[0088] out=res*SA(res);

[0089] Where out represents the output result, in represents the input, res is the intermediate value, CA(Channel Attention) represents the channel attention module, and SA(Spatial Attention) represents the spatial attention module.

[0090] Further, in combination with the three-dimensional CT image after noise reduction processing and the three-dimensional space coordinates of all candidate nodule centers, the application generates a candidate lung nodule image by taking the three-dimensional space coordinates of the candidate nodule center as a center point, and intercepting a 48*48*48 cube in the three-dimensional CT image after noise reduction processing. The cube is the candidate lung nodule image. Through this step, the candidate lung nodule region in the three-dimensional CT image is modularly extracted, effectively reducing the volume of the input parameters for subsequent false positive filtering, and the filtering efficiency and accuracy of the false positive network model are obviously optimized.

[0091] Further, please refer to Figure 4 The method flow chart for filtering false positives of the candidate lung nodule image and outputting true positive lung nodules in the CT image lung nodule segmentation and detection method provided by the first embodiment of the application, which specifically includes:

[0092] Step S41, input the candidate lung nodule image into the 3D CNN false positive filtering network for filtering.

[0093] Step S42, when the output is a false positive result, discard the coordinates of the candidate lung nodule at the corresponding position in the candidate lung nodule image.

[0094] Step S43, when the output is a true positive result, keep the coordinates of the candidate lung nodule at the corresponding position in the candidate lung nodule image.

[0095] By using the above method, the 3D CNN false positive filtering of the reorganized image of the application effectively eliminates false positive results, retains true positive results, and optimizes the accuracy of the candidate lung nodule image results.

[0096] It should be noted that before 3D CNN false positive filtering, the 3D CNN false positive model also needs to be trained. The training process is as follows:

[0097] 1) Dataset generation: First, the data is processed using the preprocessing method in the two-dimensional CT image segmentation module. Then the coordinate data is read from candidates.csv, and a (48, 48, 48) size region image centered on the coordinates is selected as the candidate lung nodule image. The candidates.csv file contains 551065 candidate nodule information, of which there are 1351 lung nodule images and 549714 non-lung nodule images, and the corresponding label values are 0 and 1 respectively. Because the positive and negative samples are very different, 26000 samples are selected from the non-lung nodule images as negative samples, and 1300 samples are selected from the lung nodule images, and then 20 times data augmentation is performed to obtain 26000 samples as positive samples. The positive and negative samples are divided into training set, validation set and test set according to the ratio of 8:1:1, and then the corresponding positive and negative sample sets are combined for training.

[0098] 2) Model training: The model is trained using 3D Resnet50, a total of 15 epochs, a learning rate of 0.0001, a binary cross-entropy loss function for updating, and accuracy as the evaluation index.

[0099] Referring to Figure 5 , the CT image lung nodule segmentation detection system proposed for the second embodiment of the application, specifically, the CT image lung nodule segmentation detection system comprises:

[0100] An image input module 51 is configured to acquire a three-dimensional CT image of a lung to be detected.

[0101] A preprocessing module 52 is configured to perform noise reduction processing on the three-dimensional CT image.

[0102] A prediction segmentation module 53 is configured to perform prediction segmentation on all two-dimensional images in the three-dimensional CT image after noise reduction processing to form a mask picture.

[0103] An image combination module 54 is configured to combine the mask picture to form a predicted three-dimensional image.

[0104] A calculation module 55 is configured to calculate the center points of all connected domains in the predicted three-dimensional image using an image operation function.

[0105] A coordinate acquisition module 56 is configured to mark the center points as candidate nodule centers and acquire three-dimensional space coordinates of the candidate nodule centers.

[0106] An image generation module 57 is configured to combine the three-dimensional CT image after noise reduction processing and the three-dimensional space coordinates of all candidate nodule centers to generate a candidate lung nodule image.

[0107] False positive filtering module 58: used for 3D CNN false positive filtering of the candidate lung nodule image, and outputting all true positive lung nodule three-dimensional spatial coordinates.

[0108] Further, the preprocessing module specifically comprises:

[0109] The reading unit is used for reading the CT scan value of each part in the three-dimensional CT image.

[0110] The intercepting unit intercepts the CT scan value of each part in the three-dimensional CT image according to a preset interval value.

[0111] The normalization unit is used for normalizing the intercepted region.

[0112] In summary, the CT image lung nodule segmentation detection system provided by the present application, in combination with the above-mentioned CT image lung nodule segmentation detection method, obtains the three-dimensional spatial coordinates of all candidate lung nodule nodes by inputting three-dimensional images, image preprocessing, and two-dimensional image segmentation, generates candidate lung nodule images, and then performs 3D CNN false positive filtering to output all true positive lung nodule coordinates. The beneficial effects are: compared with the traditional U-Net and its improvement, the two-dimensional segmentation method improves the accuracy of the lung nodule segmentation task. In the lung nodule image detection process, the overall operation is simple, the lung nodule determination accuracy is high, the burden on medical personnel can be reduced, and the diagnosis speed can be accelerated.

[0113] It should be noted that each of the above modules can be a functional module or a program module, which can be implemented by software or hardware. For the modules implemented by hardware, each of the above modules can be located in the same processor; or each of the above modules can also be located in different processors in any combination.

[0114] In addition, in combination with Figure 1 The CT image lung nodule segmentation detection method described in the present application can be implemented by a computer device. The computer device can include a processor and a memory having computer program instructions.

[0115] Specifically, the processor can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0116] The memory can include mass storage for data or instructions. By way of example, and not limitation, the memory can include a Hard Disk Drive (HDD), a floppy disk drive, a Solid State Drive (SSD), a flash drive, a Compact Disc (CD) ROM, a Digital Versatile Disc (DVD) ROM, a Blu-ray disk, a magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. The memory can be removable and / or non-removable (or fixed) as appropriate. The memory can be internal or external as appropriate. In certain embodiments, the memory is a non-volatile memory. In certain embodiments, the memory includes a Read-Only Memory (ROM) and a Random Access Memory (RAM). The ROM can be a mask-programmed ROM, a Programmable Read-Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), an Electrically Erasable Programmable Read-Only Memory (EEPROM), an Electrically Alterable Read-Only Memory (EAROM), or a FLASH memory, or a combination of two or more of these, as appropriate. The RAM can be a Static Random-Access Memory (SRAM) or a Dynamic Random Access Memory (DRAM), which can be a Fast Page Mode Dynamic Random Access Memory (FPMDRAM), an Extended Data Output Dynamic Random Access Memory (EDODRAM), a Synchronous Dynamic Random-Access Memory (SDRAM), or the like, as appropriate.

[0117] The memory can be used to store or buffer various data files required for processing and / or communication, and possible computer program instructions executed by the processor.

[0118] The processor realizes the CT image lung nodule segmentation detection method in any one of the above embodiments by reading and executing the computer program instructions stored in the memory.

[0119] The computer device can further include a communication interface and a bus. The processor, the memory, the communication interface are connected through the bus and complete the communication among each other.

[0120] The communication interface is used to realize the communication among the modules, devices, units and / or equipment in the embodiments of the application. The communication interface can also realize the data communication between other components, such as: external devices, image / data acquisition devices, databases, external storage and image / data processing workstations, etc.

[0121] The bus includes hardware, software, or both, that couples components of the computer device to each other. A bus includes, but is not limited to, at least one of a data bus, an address bus, a control bus, an expansion bus, a local bus, and the like. By way of example, and not limitation, a bus can include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand (IB) interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, and the like, or a combination of two or more of these. Where appropriate, a bus can include one or more buses. Although the present application is described and illustrated with a particular bus, the present application contemplates any suitable bus or interconnect.

[0122] The computer device can execute the CT image lung nodule segmentation and detection method in the embodiments of the present application based on the obtained data information, so as to realize the combination of Figure 1 The CT image lung nodule segmentation and detection method is described.

[0123] In addition, in combination with the CT image lung nodule segmentation and detection method in the above embodiments, the embodiments of the present application can provide a computer readable storage medium to realize. The computer readable storage medium has computer program instructions stored thereon; the computer program instructions are executed by the processor to realize any one of the CT image lung nodule segmentation and detection methods in the above embodiments.

[0124] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Also, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0125] The above-described embodiments only express several implementation manners of the present application, which are described in a more specific and detailed manner, but cannot be understood as a limitation on the patent scope of the present application. It should be noted that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, which are all within the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A CT image lung nodule segmentation detection method, characterized in that, The method comprises: acquiring a three-dimensional CT image of a lung to be detected, and performing noise reduction processing on the three-dimensional CT image; performing predictive segmentation on all two-dimensional images in the three-dimensional CT image after noise reduction processing to form a mask picture; the performing predictive segmentation on all two-dimensional images in the three-dimensional CT image after noise reduction processing to form a mask picture is implemented by using a U-net network improved by using a Floor structure and an attention mechanism; combining the mask picture to form a predicted three-dimensional image; calculating center points of all connected domains in the predicted three-dimensional image by using an image operation function; labeling the center points as candidate node centers, and acquiring three-dimensional space coordinates of the candidate node centers; combining the three-dimensional CT image after noise reduction processing and the three-dimensional space coordinates of all candidate node centers to generate a candidate lung nodule image, specifically comprising: taking the three-dimensional space coordinates of the candidate node centers as center points to intercept a 48*48*48 cube in the three-dimensional CT image after noise reduction processing, and the cube is the candidate lung nodule image; performing 3D CNN false positive filtering on the candidate lung nodule image, and outputting all true positive lung nodule three-dimensional space coordinates, specifically comprising: inputting the candidate lung nodule image into a 3D CNN false positive filtering network for filtering; when the output is a false positive result, discarding the coordinates of the candidate lung nodule at the corresponding position in the candidate lung nodule image; when the output is a true positive result, retaining the coordinates of the candidate lung nodule at the corresponding position in the candidate lung nodule image. 2.The CT image lung nodule segmentation detection method of claim 1, wherein, The step of performing noise reduction processing on the three-dimensional CT image comprises: reading CT scan values of each part in the three-dimensional CT image; intercepting the CT scan values of each part in the three-dimensional CT image according to a preset interval value, and performing normalization processing on the intercepted region. 3.The CT image lung nodule segmentation detection method of claim 1, wherein, Before the step of combining the mask picture formed by performing predictive segmentation on all two-dimensional images in the three-dimensional CT image after noise reduction processing, a two-dimensional segmentation model training step is further included, and the step comprises: selecting a LUNA16 dataset, the dataset comprising 888 CTs, and selecting 1186 nodes labeled in an annotations.csv file; selecting a slice image with a lung nodule as a center point in a three-dimensional space according to each label information; generating a lung nodule mask by using lung nodule center coordinates and radius information to obtain a dataset; dividing the dataset into a training set, a verification set and a test set according to a ratio of 8:1:1 for model training. 4.A CT image lung nodule segmentation detection system, characterized in that, The system is used to implement the CT image lung nodule segmentation detection method according to any one of claims 1-3, and the system comprises: an image input module for acquiring a three-dimensional CT image of a lung to be detected; a preprocessing module for performing noise reduction processing on the three-dimensional CT image; a predictive segmentation module for performing predictive segmentation on all two-dimensional images in the three-dimensional CT image after noise reduction processing to form a mask picture; an image combination module for combining the mask picture to form a predicted three-dimensional image; The computing module is configured to calculate center points of all connected domains in the predicted three-dimensional image using an image operation function; The coordinate obtaining module is configured to calibrate the center points as candidate node center and obtain three-dimensional space coordinates of the candidate node center; The image generating module is configured to combine the three-dimensional CT image after the noise reduction processing and the three-dimensional space coordinates of all candidate node centers and generate a candidate lung nodule image; The false positive filtering module is configured to perform 3D CNN false positive filtering on the candidate lung nodule image and output three-dimensional space coordinates of all true positive lung nodules.

5. The CT image lung nodule segmentation detection system of claim 4, wherein, The preprocessing module specifically comprises: A reading unit configured to read CT scanning values of each part in the three-dimensional CT image; An intercepting unit configured to intercept the CT scanning values of each part in the three-dimensional CT image according to preset interval values; A normalization unit configured to perform normalization processing on the intercepted region.

6. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the CT image lung nodule segmentation detection method according to any one of claims 1 to 3.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the CT image lung nodule segmentation detection method according to any one of claims 1 to 3.

Images