CN111292289A - CT lung tumor segmentation method, device, equipment and medium based on segmentation network - Google Patents
CT lung tumor segmentation method, device, equipment and medium based on segmentation network Download PDFInfo
- Publication number
- CN111292289A CN111292289A CN201811493978.0A CN201811493978A CN111292289A CN 111292289 A CN111292289 A CN 111292289A CN 201811493978 A CN201811493978 A CN 201811493978A CN 111292289 A CN111292289 A CN 111292289A
- Authority
- CN
- China
- Prior art keywords
- image
- lung
- segmentation
- sampling
- region
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30061—Lung
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30096—Tumor; Lesion
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Quality & Reliability (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Apparatus For Radiation Diagnosis (AREA)
- Image Processing (AREA)
Abstract
The invention is suitable for the technical field of medical image processing, and provides a CT lung tumor segmentation method, a device, equipment and a medium based on a segmentation network, wherein the method comprises the following steps: the method comprises the steps of carrying out image segmentation on a lung preprocessing image obtained through preprocessing by adopting a first segmentation network to obtain a lung tumor initial segmentation image, resampling the lung preprocessing image and the lung tumor initial segmentation image according to pixel intervals to obtain a corresponding lung sampling image and an initial segmentation sampling image, cutting the initial segmentation sampling image into initial segmentation image blocks with the size of a preset number of image blocks by taking the centroid of the initial segmentation sampling image as the center, and obtaining the lung tumor segmentation image corresponding to a lung CT image by adopting a second segmentation network according to the lung sampling image and the initial segmentation image blocks, so that the accuracy and integrity of lung tumor segmentation on the lung CT image are improved, a high-precision lung tumor segmentation image is obtained, and the safety degree of an operation is improved.
Description
Technical Field
The invention belongs to the technical field of medical image processing, and particularly relates to a CT lung tumor segmentation method, a CT lung tumor segmentation device, CT lung tumor segmentation equipment and a CT lung tumor segmentation medium based on a segmentation network.
Background
Lung cancer is the most common malignant tumor in the world, and not only has high incidence rate, but also ranks first in the list of malignant tumors and the mortality rate. In recent years, due to economic development, the living environment and the living style of people are greatly changed, the smoking people are expanded, the long-term psychological stress of people is overlarge, the people lack of exercise, the atmospheric environment pollution is aggravated, and the incidence rate of lung cancer is continuously increased. The rapid development of Computed Tomography (CT) technology continuously affects the diagnosis of human diseases, and CT images have become one of the important conventional methods in clinical diagnosis and treatment.
At present, the treatment means of lung tumor mainly comprises tumor resection, intervention, radiotherapy and the like, and the tumor resection is the most effective treatment mode. CT tumor segmentation is generally used for surgical planning before tumor resection, three-dimensional visualization during surgery, surgical resection scheme design, surgical risk assessment, and the like, and is used for extracting features of organs or tissues in computer-aided diagnosis, and then performing qualitative and quantitative analysis to observe the change condition of tumors in the radiotherapy process, however, lung tumors have variable positions, sizes and shapes, gray levels are similar to those of adjacent pulmonary vessels, and small tumors are difficult to identify. Traditional manual segmentation needs to have anatomical knowledge and experience, subjective diversity and a great deal of time and energy until the breakthrough development of a deep convolutional neural network makes automatic segmentation of lung tumors possible. The existing lung tumor automatic segmentation method depends on the lung tumor segmentation based on a threshold segmentation method, and the method easily loses rich texture information of the lung, so that large lung tumors are easily removed, and the segmented lung tumors are incomplete.
Disclosure of Invention
The invention aims to provide a CT lung tumor segmentation method, a device, equipment and a storage medium based on a segmentation network, and aims to solve the problems that the prior art cannot provide an effective method for segmenting lung tumors in a lung CT image, so that the segmentation precision is low, and the segmented lung tumors are incomplete.
In one aspect, the present invention provides a segmentation network-based CT lung tumor segmentation method, including the following steps:
when a request for lung tumor segmentation of a lung CT image is received, preprocessing the lung CT image to obtain a corresponding lung preprocessed image;
carrying out image segmentation on the lung preprocessing image through a pre-trained first segmentation network to obtain a corresponding lung tumor initial segmentation image;
resampling the lung preprocessing image and the lung tumor initial segmentation image according to a preset pixel interval to obtain a corresponding lung sampling image and an initial segmentation sampling image;
cutting the initial segmentation sampling image according to the size of a preset image block by taking the center of mass of the initial segmentation sampling image as the center to obtain a preset number of initial segmentation image blocks;
and obtaining a lung tumor segmentation image corresponding to the lung CT image by adopting a pre-trained second segmentation network according to the lung sampling image and the initial segmentation image block.
In another aspect, the present invention provides a segmentation network-based CT lung tumor segmentation apparatus, including:
the CT image preprocessing unit is used for preprocessing the lung CT image to obtain a corresponding lung preprocessing image when a lung tumor segmentation request for the lung CT image is received;
the first image segmentation unit is used for carrying out image segmentation on the lung preprocessing image through a pre-trained first segmentation network to obtain a corresponding lung tumor initial segmentation image;
the image resampling unit is used for resampling the lung preprocessing image and the lung tumor initial segmentation image according to a preset pixel interval to obtain a corresponding lung sampling image and a corresponding lung initial segmentation sampling image;
the image block obtaining unit is used for cutting the initial segmentation sampling image according to the size of a preset image block by taking the mass center of the initial segmentation sampling image as the center to obtain a preset number of initial segmentation image blocks; and
and the second image segmentation unit is used for obtaining a lung tumor segmentation image corresponding to the lung CT image by adopting a pre-trained second segmentation network according to the lung sampling image and the initial segmentation image block.
In another aspect, the present invention further provides a computing device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the segmentation network-based CT lung tumor segmentation method when executing the computer program.
In another aspect, the present invention further provides a computer-readable storage medium, which stores a computer program, which when executed by a processor, implements the steps of the above-mentioned segmentation network-based CT lung tumor segmentation method.
The invention adopts a first segmentation network to carry out image segmentation on a lung preprocessed image obtained by preprocessing so as to obtain a lung tumor initial segmentation image, resamples the lung preprocessed image and the lung tumor initial segmentation image according to the pixel pitch so as to obtain a corresponding lung sampling image and an initial segmentation sampling image, calculates the centroid of the initial segmentation sampling image, cuts the initial segmentation sampling image into initial segmentation image blocks with the preset number of image block sizes by taking the centroid as the center, adopts a second segmentation network to obtain the lung tumor segmentation image corresponding to the lung CT image according to the lung sampling image and the initial segmentation image blocks, thereby positioning the lung tumor region in the lung CT image through the first segmentation network, and segmenting the lung tumor through the second segmentation network, thereby improving the accuracy and the integrity of the lung tumor segmentation of the lung CT image, and then obtain the lung tumour segmentation image of high accuracy, improve the safe degree of operation.
Drawings
Fig. 1 is a flowchart illustrating an implementation of a segmentation network-based CT lung tumor segmentation method according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating a method for preprocessing a CT image of a lung according to a second embodiment of the present invention;
fig. 3 is a flowchart of an implementation of training a preset segmentation network according to a third embodiment of the present invention;
fig. 4 is a schematic structural diagram of a segmentation network-based CT lung tumor segmentation apparatus according to a fourth embodiment of the present invention;
fig. 5 is a schematic structural diagram of a CT lung tumor segmentation apparatus based on a segmentation network according to a fourth embodiment of the present invention;
fig. 6 is a schematic structural diagram of a segmentation network-based CT lung tumor segmentation apparatus according to a fifth embodiment of the present invention; and
fig. 7 is a schematic structural diagram of a computing device according to a sixth embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The following detailed description of specific implementations of the present invention is provided in conjunction with specific embodiments:
the first embodiment is as follows:
fig. 1 shows an implementation flow of a segmentation network-based CT lung tumor segmentation method according to an embodiment of the present invention, and for convenience of description, only the relevant portions related to the embodiment of the present invention are shown, which is detailed as follows:
in step S101, when a request for lung tumor segmentation is received for a lung CT image, the lung CT image is preprocessed to obtain a corresponding lung preprocessed image.
The embodiment of the invention is suitable for medical image processing platforms, systems or equipment, such as personal computers, servers and the like. In the embodiment of the invention, because the CT image is formed by arranging a certain number of pixels with different gray scales from black to white according to a matrix, the different gray scales of the CT image reflect the absorption degree of corresponding organs and tissues to X rays, the black area represents a low absorption area, namely a low density area, such as a lung containing a large amount of gas, and the white area represents a high absorption area, namely a high density area, such as a bone, the anatomical structure of certain organs of a human body and the characteristics of lesion tissues with changed densities in the organs can be clearly obtained according to the CT image, and the CT image of the lung is generated by carrying out tomography on the lung of a patient through CT equipment, so that the lesion tissues of the lung of the patient can be found through the CT image of the lung for further treatment. When a user needs to perform lung tumor segmentation on a lung CT image of a patient, a lung tumor segmentation request is sent, wherein the user can obtain the lung CT image of the patient from a public medical image database or an operation image provided by a hospital, and when the lung tumor segmentation request is received, the lung CT image input by the user is obtained, and the obtained lung CT image is preprocessed to obtain a corresponding lung preprocessed image.
In step S102, image segmentation is performed on the preprocessed lung image through a pre-trained first segmentation network, so as to obtain a corresponding initial segmentation image of the lung tumor.
In the embodiment of the invention, the image segmentation is a technology and a process for dividing an image into a plurality of specific areas with unique properties and proposing an interested target. And inputting the lung preprocessing image into a first segmentation network trained in advance for image segmentation, and extracting a lung tumor region in the lung preprocessing image to obtain a corresponding coarse-grained lung tumor initial segmentation image.
In step S103, the lung preprocessing image and the lung tumor initial segmentation image are resampled according to a preset pixel interval, and a corresponding lung sampling image and an initial segmentation sampling image are obtained.
In the embodiment of the invention, the lung preprocessing image is resampled to a preset pixel space (namely, a pixel space) to obtain a lung sampling image, and the lung tumor initial segmentation image is also resampled to the pixel space to obtain a corresponding initial segmentation sampling image.
Before resampling the lung preprocessed image and the lung tumor initial segmentation image according to the preset pixel interval, preferably, the pixel interval is set to be 1mm × 1mm × 2mm, so that image information of the lung sampled image and the lung tumor initial segmentation sampled image is increased, and image quality of the lung sampled image and the lung tumor initial segmentation sampled image is improved.
In step S104, the initial divided sample image is clipped according to the size of the preset image block with the centroid of the initial divided sample image as the center, so as to obtain a preset number of initial divided image blocks.
In the embodiment of the invention, the centroid of the initial segmentation sampling image is calculated, and the initial segmentation image blocks with the preset number and the preset image block size are extracted from the initial segmentation sampling image by taking the calculated centroid as the center.
Before the initial divided sample image is cropped, the image block size is preferably set to 128 × 128 × 64, thereby improving the image quality of the initial divided image block.
When the initial segmentation sampling image is cropped, the gray value of the pixel of the centroid of the initial segmentation sampling image is preferably set to 2, so that the brightness of the pixel at the centroid is improved, and the accuracy of the subsequent lung tumor segmentation is further improved.
In step S105, a lung tumor segmentation image corresponding to the lung CT image is obtained by using a pre-trained second segmentation network according to the lung sampling image and the initial segmentation image block.
In the embodiment of the invention, the lung sampling image and all the initial segmentation image blocks are input into a second segmentation network which is trained in advance, and the lung tumor segmentation image is output through the second segmentation network, namely, the final lung tumor segmentation image with fine granularity is obtained by segmenting the lung tumor region in the lung CT image.
In the embodiment of the invention, a first segmentation network is adopted to carry out image segmentation on a lung preprocessed image obtained by preprocessing so as to obtain a lung tumor initial segmentation image, the lung preprocessed image and the lung tumor initial segmentation image are resampled according to pixel intervals so as to obtain a corresponding lung sampling image and an initial segmentation sampling image, the initial segmentation sampling image is cut into initial segmentation image blocks with the preset number of image block sizes by taking the centroid of the initial segmentation sampling image as the center, a second segmentation network is adopted to obtain a lung tumor segmentation image corresponding to the lung CT image according to the lung sampling image and the initial segmentation image blocks, so that a lung tumor region in the lung CT image is positioned through the first segmentation network, the lung tumor is segmented through the second segmentation network, and the accuracy and the integrity of the lung tumor segmentation of the lung CT image are improved, and then obtain the lung tumour segmentation image of high accuracy, improve the safe degree of operation.
Example two:
fig. 2 shows a flow of implementing the preprocessing of the lung CT image according to the second embodiment of the present invention, and for convenience of description, only the relevant portions of the second embodiment of the present invention are shown, which is detailed as follows:
in step S201, an air connected region in the lung CT image is extracted according to a preset seed point.
In the embodiment of the invention, the air communication region in the lung CT image is extracted by using a region growing method with the preset seed point as a starting point, and the air communication region is also the background region of the lung CT image.
Before extracting the air connected region in the lung CT image, preferably, the first pixel point in the upper left corner of the lung CT image is set as a seed point, so that the accuracy and completeness of extraction of the air connected region are improved.
In step S202, the extracted air connected region is removed from the CT image of the lung, and a corresponding mask region image of the body is obtained.
In the embodiment of the invention, the air communication area is cut out from the lung CT image to obtain the corresponding body mask area image, so that the noise in the lung CT image is removed, and the image quality of the obtained body mask area image is improved.
In step S203, threshold segmentation is performed on the body mask region image to obtain a maximum left-lung connected region and a maximum right-lung connected region.
In the embodiment of the invention, the body mask region is subjected to image segmentation by a threshold segmentation method to obtain the maximum left lung connected region and the maximum right lung connected region in the body mask region image.
When the image of the body mask region is subjected to threshold segmentation, the image segmentation of the body mask region is preferably performed by a threshold segmentation method according to a threshold range of [ -1024, -400], so that the accuracy and completeness of the segmentation of the maximum connected regions of the left lung and the maximum connected regions of the right lung are improved while the image segmentation of the body mask region is simplified.
In step S204, according to the maximum connected region of the left lung and the maximum connected region of the right lung, region-of-interest extraction is performed on the lung CT image, so as to obtain a corresponding lung region-of-interest image.
In step S205, gray-scale transformation is performed on the lung roi image according to a preset gray-scale transformation formula, and normalization processing is performed on the lung roi image after gray-scale transformation, so as to obtain a lung preprocessed image corresponding to the lung CT image.
In the embodiment of the present invention, preferably, the grayscale conversion formula is f (x) ═ 0.2 × i (x), where i (x) is a grayscale value of an xth pixel point in the maximum left lung communication area or the maximum right lung communication area, and f (x) is a grayscale value of the xth pixel point after grayscale conversion, so as to improve the contrast of the lung interesting area image, so that the lung interesting area image becomes clearer, finer and more easily recognized, and further improve the display effect of the lung interesting area image.
When the lung interesting region image after gray level transformation is normalized, preferably, the lung interesting region image after gray level transformation is normalized to [0,1], so that the contrast of the lung interesting region image is further improved, and the details of the lung interesting region image are clearer.
In the embodiment of the invention, preprocessing such as air communication region cutting, region-of-interest extraction, threshold segmentation, gray level conversion, gray level normalization and the like is carried out on the lung CT image, so that the robustness and the execution efficiency of the algorithm are improved.
Example three:
fig. 3 shows an implementation process of training a preset segmentation network according to a third embodiment of the present invention, and for convenience of description, only the parts related to the third embodiment of the present invention are shown, which are detailed as follows:
in step S301, the training sample obtained by preprocessing is resampled according to the preset sample size, so as to obtain a first sampled training sample.
In the embodiment of the invention, the training sample obtained by preprocessing is resampled to the preset sample sampling size to obtain the first sampling training sample, wherein the training sample is a lung tumor CT image data set with different shapes and sizes.
Before resampling the preprocessed training samples, the sample sampling size is preferably set to 128 × 128 × 64, so as to increase the image information of the first sampled training sample, and further improve the image quality of the first sampled training sample.
Before resampling the preprocessed training samples, it is further preferable to preprocess the training samples, specifically, to preprocess the training samples by:
1) extracting an air communication area in the training sample according to a preset seed point;
2) removing the extracted air communication area from the training sample to obtain a corresponding body mask sample image;
3) performing threshold segmentation on the body mask sample image to obtain a left lung maximum connected region and a right lung maximum connected region;
4) extracting the region of interest of the training sample according to the maximum connected region of the left lung and the maximum connected region of the right lung to obtain a corresponding lung region of interest sample image;
5) and carrying out gray level transformation on the lung region-of-interest sample image according to a preset gray level transformation formula, and carrying out normalization processing on the lung region-of-interest sample image after gray level transformation to obtain a preprocessed training sample.
Therefore, the training samples are preprocessed through the steps 1) to 5), the image feature contrast of the training samples is improved, the training samples become clearer, finer and easier to recognize, and the training effect of the subsequent segmentation network is improved.
In step S302, a preset segmentation network is trained through a first sampling training sample according to a preset number of training iterations, so as to obtain a first segmentation network.
In the embodiment of the invention, iterative training of preset training iteration times is performed on a preset segmentation network through a first sampling training sample to obtain a first segmentation network, wherein the preset segmentation network is a Full Convolution Network (FCN), SegNet, U-Net or V-Net and other semantic segmentation Networks, and the obtained first segmentation network is obtained by inputting the first sampling training sample into the semantic segmentation network and training the semantic segmentation network, so that the first segmentation network can roughly position a lung tumor region in a lung CT image, and the integrity of image segmentation of the lung tumor region in the lung CT image through the first segmentation network is improved.
In step S303, the training samples are resampled according to the pixel pitch to obtain a second sampling training sample.
In the embodiment of the present invention, the training sample obtained by the preprocessing is resampled to a preset pixel space (i.e., pixel space) to obtain a second sampling training sample, where the training sample is the same as the training sample in step S301.
In step S304, the second sampling training sample is clipped according to the size of the image block with a preset positioning point as a center to obtain a preset number of training sample image blocks, where the positioning point is a pixel point of the tumor region in the second sampling training sample.
In the embodiment of the invention, a pixel point is randomly selected from a tumor region in the second sampling training sample as an anchor point, and then training sample image blocks with the sizes of a preset number of preset image blocks are extracted from the second sampling training sample by taking the anchor point as a center.
When the second sampling training sample is cut, preferably, the pixel gray value of the positioning point is set to be 2, so that the brightness of the pixel at the positioning point is improved, and the training effect of the subsequent second segmentation network is further improved.
In step S305, training a preset segmentation network through training sample image blocks according to the training iteration number to obtain a second segmentation network.
In the embodiment of the invention, the preset segmentation network is subjected to iterative training for the preset training iteration times through the training sample image blocks to obtain the second segmentation network, and the obtained second segmentation network is obtained by inputting all the training sample image blocks into the semantic segmentation network and training the semantic segmentation network, so that the second segmentation network can accurately position the lung tumor in the lung CT image, and the accuracy of lung tumor image segmentation on the lung CT image through the second segmentation network is improved.
In the embodiment of the invention, the first segmentation network and the second segmentation network are trained in parallel through the lung tumor CT image data sets with different shapes and sizes, so that the training efficiency is improved, and the consumption of a memory in the training process is reduced.
Example four:
fig. 4 shows a structure of a segmentation network-based CT lung tumor segmentation apparatus according to a fourth embodiment of the present invention, and for convenience of description, only the parts related to the embodiment of the present invention are shown, which include:
the CT image preprocessing unit 41 is configured to, when a request for lung tumor segmentation on a lung CT image is received, preprocess the lung CT image to obtain a corresponding lung preprocessed image;
a first image segmentation unit 42, configured to perform image segmentation on the lung preprocessing image through a pre-trained first segmentation network to obtain a corresponding lung tumor initial segmentation image;
the image resampling unit 43 is configured to resample the lung preprocessing image and the lung tumor initial segmentation image according to a preset pixel interval, so as to obtain a corresponding lung sampling image and an initial segmentation sampling image;
an image block obtaining unit 44, configured to cut the initial segmented sample image according to a preset image block size with a centroid of the initial segmented sample image as a center, so as to obtain a preset number of initial segmented image blocks; and
and the second image segmentation unit 45 is configured to obtain a lung tumor segmentation image corresponding to the lung CT image by using a pre-trained second segmentation network according to the lung sampling image and the initial segmentation image block.
As shown in fig. 5, the CT image preprocessing unit 41 preferably includes:
a first region extraction unit 411, configured to extract an air connected region in the lung CT image according to a preset seed point;
a second region extraction unit 412, configured to remove the extracted air connected region from the lung CT image, so as to obtain a corresponding body mask region image;
an image threshold segmentation unit 413, configured to perform threshold segmentation on the body mask region image to obtain a left lung maximum connected region and a right lung maximum connected region;
a third region extraction unit 414, configured to perform region-of-interest extraction on the lung CT image according to the maximum left lung connected region and the maximum right lung connected region, so as to obtain a corresponding lung region-of-interest image; and
the normalization processing unit 415 is configured to perform gray-scale transformation on the lung region-of-interest image according to a preset gray-scale transformation formula, and perform normalization processing on the lung region-of-interest image after gray-scale transformation to obtain a lung preprocessed image corresponding to the lung CT image.
In the embodiment of the present invention, each unit of the CT lung tumor segmentation apparatus based on the segmentation network may be implemented by corresponding hardware or software units, and each unit may be an independent software or hardware unit, or may be integrated into a software or hardware unit, which is not limited herein. Specifically, the implementation of each unit can refer to the description of the foregoing method embodiment, and is not repeated herein.
Example five:
fig. 6 shows a structure of a CT lung tumor segmentation apparatus based on a segmentation network according to a fifth embodiment of the present invention, which only shows parts related to the embodiment of the present invention for convenience of description, and includes:
the first sample sampling unit 60 is configured to resample the training sample obtained by preprocessing according to a preset sample sampling size to obtain a first sampling training sample;
the first network training unit 61 is configured to train a preset segmentation network through a first sampling training sample according to a preset training iteration number to obtain a first segmentation network;
the second sample sampling unit 62 is configured to resample the training samples according to the pixel intervals to obtain second sampling training samples;
the second sample clipping unit 63 is configured to clip the second sampling training sample according to the size of the image block by taking a preset positioning point as a center to obtain a preset number of training sample image blocks, where the positioning point is a pixel point of a tumor region in the second sampling training sample;
the second network training unit 64 is configured to train the preset segmentation network through the training sample image blocks according to the training iteration number to obtain a second segmentation network;
the CT image preprocessing unit 65 is configured to, when a request for segmenting lung tumor from a lung CT image is received, preprocess the lung CT image to obtain a corresponding lung preprocessed image;
a first image segmentation unit 66, configured to perform image segmentation on the lung preprocessing image through a pre-trained first segmentation network to obtain a corresponding lung tumor initial segmentation image;
the image resampling unit 67 is configured to resample the lung preprocessing image and the lung tumor initial segmentation image according to a preset pixel interval, so as to obtain a corresponding lung sampling image and an initial segmentation sampling image;
an image block obtaining unit 68, configured to cut the initial segmented sample image according to a preset image block size with a centroid of the initial segmented sample image as a center, so as to obtain a preset number of initial segmented image blocks; and
and a second image segmentation unit 69, configured to obtain a lung tumor segmentation image corresponding to the lung CT image by using a pre-trained second segmentation network according to the lung sampling image and the initial segmentation image block.
In the embodiment of the present invention, each unit of the CT lung tumor segmentation apparatus based on the segmentation network may be implemented by corresponding hardware or software units, and each unit may be an independent software or hardware unit, or may be integrated into a software or hardware unit, which is not limited herein. Specifically, the implementation of each unit can refer to the description of the foregoing method embodiment, and is not repeated herein.
Example six:
fig. 7 shows a structure of a computing device according to a sixth embodiment of the present invention, and for convenience of explanation, only a part related to the embodiment of the present invention is shown.
The computing device 7 of an embodiment of the invention comprises a processor 70, a memory 71 and a computer program 72 stored in the memory 71 and executable on the processor 70. The processor 70, when executing the computer program 72, implements the steps in the above-mentioned segmentation network-based CT lung tumor segmentation method embodiment, such as the steps S101 to S105 shown in fig. 1. Alternatively, the processor 70, when executing the computer program 72, implements the functions of the units in the above-described device embodiments, such as the functions of the units 41 to 45 shown in fig. 4.
In the embodiment of the invention, a first segmentation network is adopted to carry out image segmentation on a lung preprocessed image obtained by preprocessing so as to obtain a lung tumor initial segmentation image, the lung preprocessed image and the lung tumor initial segmentation image are resampled according to pixel intervals so as to obtain a corresponding lung sampling image and an initial segmentation sampling image, the initial segmentation sampling image is cut into initial segmentation image blocks with the preset number of image block sizes by taking the centroid of the initial segmentation sampling image as the center, a second segmentation network is adopted to obtain a lung tumor segmentation image corresponding to the lung CT image according to the lung sampling image and the initial segmentation image blocks, so that a lung tumor region in the lung CT image is positioned through the first segmentation network, the lung tumor is segmented through the second segmentation network, and the accuracy and the integrity of the lung tumor segmentation of the lung CT image are improved, and then obtain the lung tumour segmentation image of high accuracy, improve the safe degree of operation.
The computing equipment of the embodiment of the invention can be a personal computer and a server. The steps of the method for segmenting the CT lung tumor based on the segmentation network implemented by the processor 70 in the computing device 7 executing the computer program 72 can refer to the description of the foregoing method embodiments, and are not described herein again.
Example seven:
in an embodiment of the present invention, a computer-readable storage medium is provided, which stores a computer program, which when executed by a processor implements the steps in the above-mentioned segmentation network-based CT lung tumor segmentation method embodiment, for example, the steps S101 to S105 shown in fig. 1. Alternatively, the computer program realizes the functions of the units in the above-described device embodiments, such as the functions of the units 41 to 45 shown in fig. 4, when executed by the processor.
In the embodiment of the invention, a first segmentation network is adopted to carry out image segmentation on a lung preprocessed image obtained by preprocessing so as to obtain a lung tumor initial segmentation image, the lung preprocessed image and the lung tumor initial segmentation image are resampled according to pixel intervals so as to obtain a corresponding lung sampling image and an initial segmentation sampling image, the initial segmentation sampling image is cut into initial segmentation image blocks with the preset number of image block sizes by taking the centroid of the initial segmentation sampling image as the center, a second segmentation network is adopted to obtain a lung tumor segmentation image corresponding to the lung CT image according to the lung sampling image and the initial segmentation image blocks, so that a lung tumor region in the lung CT image is positioned through the first segmentation network, the lung tumor is segmented through the second segmentation network, and the accuracy and the integrity of the lung tumor segmentation of the lung CT image are improved, and then obtain the lung tumour segmentation image of high accuracy, improve the safe degree of operation.
The computer readable storage medium of the embodiments of the present invention may include any entity or device capable of carrying computer program code, a recording medium, such as a ROM/RAM, a magnetic disk, an optical disk, a flash memory, or the like.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (10)
1. A CT lung tumor segmentation method based on a segmentation network is characterized by comprising the following steps:
when a request for lung tumor segmentation of a lung CT image is received, preprocessing the lung CT image to obtain a corresponding lung preprocessed image;
carrying out image segmentation on the lung preprocessing image through a pre-trained first segmentation network to obtain a corresponding lung tumor initial segmentation image;
resampling the lung preprocessing image and the lung tumor initial segmentation image according to a preset pixel interval to obtain a corresponding lung sampling image and an initial segmentation sampling image;
cutting the initial segmentation sampling image according to the size of a preset image block by taking the center of mass of the initial segmentation sampling image as the center to obtain a preset number of initial segmentation image blocks;
and obtaining a lung tumor segmentation image corresponding to the lung CT image by adopting a pre-trained second segmentation network according to the lung sampling image and the initial segmentation image block.
2. The method of claim 1, wherein the step of pre-processing the acquired CT images of the lungs comprises:
extracting an air communication region in the lung CT image according to a preset seed point;
removing the extracted air communication area from the lung CT image to obtain a corresponding body mask area image;
performing threshold segmentation on the body mask region image to obtain a left lung maximum connected region and a right lung maximum connected region;
extracting interested areas of the lung CT image according to the maximum connected area of the left lung and the maximum connected area of the right lung to obtain a corresponding lung interested area image;
and carrying out gray level transformation on the lung region-of-interest image according to a preset gray level transformation formula, and carrying out normalization processing on the lung region-of-interest image after gray level transformation to obtain the lung preprocessing image corresponding to the lung CT image.
3. The method of claim 2 wherein the step of thresholding the image of the body mask region includes
And performing threshold segmentation on the body mask region image according to a threshold range of < -1024 < -400 >.
4. The method according to claim 2, wherein the gray-scale transformation formula is f (x) 0.2 x i (x), where i (x) is the gray-scale value of the xth pixel point in the maximum left lung communication region or the maximum right lung communication region, and f (x) is the gray-scale value of the xth pixel point after gray-scale transformation.
5. The method of claim 1, wherein prior to preprocessing the pulmonary CT image, the method further comprises:
resampling the training sample obtained by preprocessing according to the preset sample sampling size to obtain a first sampling training sample;
training a preset segmentation network through the first sampling training sample according to preset training iteration times to obtain a first segmentation network;
resampling the training sample according to the pixel interval to obtain a second sampling training sample;
cutting the second sampling training sample by taking a preset positioning point as a center according to the size of the image block to obtain a preset number of training sample image blocks, wherein the positioning point is a pixel point of a tumor region in the second sampling training sample;
and training the segmentation network through the training sample image blocks according to the training iteration times to obtain the second segmentation network.
6. A segmentation network-based CT lung lesion segmentation apparatus, the apparatus comprising:
the CT image preprocessing unit is used for preprocessing the lung CT image to obtain a corresponding lung preprocessing image when a lung tumor segmentation request for the lung CT image is received;
the first image segmentation unit is used for carrying out image segmentation on the lung preprocessing image through a pre-trained first segmentation network to obtain a corresponding lung tumor initial segmentation image;
the image resampling unit is used for resampling the lung preprocessing image and the lung tumor initial segmentation image according to a preset pixel interval to obtain a corresponding lung sampling image and a corresponding lung initial segmentation sampling image;
the image block obtaining unit is used for cutting the initial segmentation sampling image according to the size of a preset image block by taking the mass center of the initial segmentation sampling image as the center to obtain a preset number of initial segmentation image blocks; and
and the second image segmentation unit is used for obtaining a lung tumor segmentation image corresponding to the lung CT image by adopting a pre-trained second segmentation network according to the lung sampling image and the initial segmentation image block.
7. The apparatus of claim 6, wherein the CT image pre-processing unit comprises:
the first region extraction unit is used for extracting an air communication region in the lung CT image according to a preset seed point;
the second region extraction unit is used for removing the extracted air communication region from the lung CT image to obtain a corresponding body mask region image;
an image threshold segmentation unit, configured to perform threshold segmentation on the body mask region image to obtain a left lung maximum connected region and a right lung maximum connected region;
the third region extraction unit is used for extracting the region of interest of the lung CT image according to the maximum left lung communication region and the maximum right lung communication region to obtain a corresponding lung region of interest image; and
and the normalization processing unit is used for carrying out gray level transformation on the lung region-of-interest image according to a preset gray level transformation formula and carrying out normalization processing on the lung region-of-interest image after gray level transformation to obtain the lung preprocessing image corresponding to the lung CT image.
8. The apparatus of claim 6, wherein the apparatus further comprises:
the first sample sampling unit is used for resampling the training sample obtained by preprocessing according to the sampling size of the preset sample to obtain a first sampling training sample;
the first network training unit is used for training a preset segmentation network through the first sampling training sample according to preset training iteration times to obtain the first segmentation network;
the second sample sampling unit is used for resampling the training samples according to the pixel spacing to obtain second sampling training samples;
the second sample cutting unit is used for cutting the second sampling training sample by taking a preset positioning point as a center according to the size of the image block to obtain a preset number of training sample image blocks, wherein the positioning point is one pixel point of a tumor region in the second sampling training sample; and
and the second network training unit is used for training the segmentation network through the training sample image blocks according to the training iteration times to obtain the second segmentation network.
9. A computing device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any of claims 1 to 5 when executing the computer program.
10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 5.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811493978.0A CN111292289B (en) | 2018-12-07 | 2018-12-07 | CT lung tumor segmentation method, device, equipment and medium based on segmentation network |
| PCT/CN2019/122103 WO2020114332A1 (en) | 2018-12-07 | 2019-11-29 | Segmentation-network-based ct lung tumor segmentation method, apparatus and device, and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811493978.0A CN111292289B (en) | 2018-12-07 | 2018-12-07 | CT lung tumor segmentation method, device, equipment and medium based on segmentation network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111292289A true CN111292289A (en) | 2020-06-16 |
| CN111292289B CN111292289B (en) | 2023-09-26 |
Family
ID=70975202
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811493978.0A Active CN111292289B (en) | 2018-12-07 | 2018-12-07 | CT lung tumor segmentation method, device, equipment and medium based on segmentation network |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN111292289B (en) |
| WO (1) | WO2020114332A1 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112614143A (en) * | 2020-12-30 | 2021-04-06 | 深圳市联影高端医疗装备创新研究院 | Image segmentation method and device, electronic equipment and storage medium |
| CN117934480A (en) * | 2024-03-25 | 2024-04-26 | 天津市第五中心医院 | Lung CT image processing method and system for lung injury prediction |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112767411B (en) * | 2020-12-31 | 2024-08-06 | 罗雄彪 | Synchronous segmentation method for multiple organs of lung |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130216116A1 (en) * | 2012-02-02 | 2013-08-22 | Peter Yim | Method and system for image segmentation |
| CN107392910A (en) * | 2017-07-06 | 2017-11-24 | 沈阳东软医疗系统有限公司 | A kind of lobe of the lung dividing method and device based on CT images |
| CN108492297A (en) * | 2017-12-25 | 2018-09-04 | 重庆理工大学 | The MRI brain tumors positioning for cascading convolutional network based on depth and dividing method in tumor |
| CN108615237A (en) * | 2018-05-08 | 2018-10-02 | 上海商汤智能科技有限公司 | A kind of method for processing lung images and image processing equipment |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080267499A1 (en) * | 2007-04-30 | 2008-10-30 | General Electric Company | Method and system for automatic detection of objects in an image |
| CN103942785A (en) * | 2014-04-09 | 2014-07-23 | 苏州大学 | Lung tumor segmentation method based on PET and CT images of image segmentation |
| GB201705876D0 (en) * | 2017-04-11 | 2017-05-24 | Kheiron Medical Tech Ltd | Recist |
| CN108537784B (en) * | 2018-03-30 | 2021-08-24 | 四川元匠科技有限公司 | CT image pulmonary nodule detection method based on deep learning |
-
2018
- 2018-12-07 CN CN201811493978.0A patent/CN111292289B/en active Active
-
2019
- 2019-11-29 WO PCT/CN2019/122103 patent/WO2020114332A1/en active Application Filing
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130216116A1 (en) * | 2012-02-02 | 2013-08-22 | Peter Yim | Method and system for image segmentation |
| CN107392910A (en) * | 2017-07-06 | 2017-11-24 | 沈阳东软医疗系统有限公司 | A kind of lobe of the lung dividing method and device based on CT images |
| CN108492297A (en) * | 2017-12-25 | 2018-09-04 | 重庆理工大学 | The MRI brain tumors positioning for cascading convolutional network based on depth and dividing method in tumor |
| CN108615237A (en) * | 2018-05-08 | 2018-10-02 | 上海商汤智能科技有限公司 | A kind of method for processing lung images and image processing equipment |
Non-Patent Citations (3)
| Title |
|---|
| HUANG X ET AL: "Lung nodule detection in CT using3D convolutional neural networks", 《PROCEEDINGS OF IEEE14TH INTERNATIONAL SYMPOSIUM ON BIOMEDICAL IMAGING.LOS ALAMITOS:IEEE COMPUTER SOCIETY PRESS》, pages 379 - 383 * |
| 张鹏 等: "基于深度学习的计算机辅助肺癌诊断方法", 《计算机辅助设计与图形学学报》, no. 01, pages 92 - 101 * |
| 胡光亮等: "基于卷积神经网络的鼻咽肿瘤MR图像分割", pages 213 - 217 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112614143A (en) * | 2020-12-30 | 2021-04-06 | 深圳市联影高端医疗装备创新研究院 | Image segmentation method and device, electronic equipment and storage medium |
| CN117934480A (en) * | 2024-03-25 | 2024-04-26 | 天津市第五中心医院 | Lung CT image processing method and system for lung injury prediction |
| CN117934480B (en) * | 2024-03-25 | 2024-05-28 | 天津市第五中心医院 | Lung CT image processing method and system for lung injury prediction |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111292289B (en) | 2023-09-26 |
| WO2020114332A1 (en) | 2020-06-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021115313A1 (en) | System for automatically sketching contour line of normal organ in medical image | |
| WO2019200753A1 (en) | Lesion detection method, device, computer apparatus and storage medium | |
| CN111292289B (en) | CT lung tumor segmentation method, device, equipment and medium based on segmentation network | |
| CN1663530A (en) | Methods and apparatus for processing image data to aid in detecting disease | |
| CN110097557B (en) | Medical image automatic segmentation method and system based on 3D-UNet | |
| CN111260671A (en) | Lung leaf segmentation method and system for CT image | |
| CN110189324B (en) | Medical image processing method and processing device | |
| CN114092450B (en) | Real-time image segmentation method, system and device based on gastroscopy video | |
| CN110766659A (en) | Medical image recognition method, apparatus, device and medium | |
| CN112950552B (en) | Rib segmentation marking method and system based on convolutional neural network | |
| CN110706236B (en) | Three-dimensional reconstruction method and device of blood vessel image | |
| CN112634291B (en) | Burn wound area automatic segmentation method based on neural network | |
| CN112561917A (en) | Image segmentation method and system, electronic device and readable storage medium | |
| CN116824173A (en) | Medical image processing method, medical image processing device and storage medium | |
| CN117409002A (en) | Visual identification detection system for wounds and detection method thereof | |
| CN115115657A (en) | Focus segmentation method and device, electronic device and storage medium | |
| CN113706541B (en) | Image processing method and device | |
| CN116309551B (en) | Method, device, electronic equipment and readable medium for determining focus sampling area | |
| CN107610772A (en) | A kind of thyroid nodule CT image diagnostic system design methods | |
| CN110858412B (en) | Heart coronary artery CTA model building method based on image registration | |
| CN113658198B (en) | Interactive emphysema focus segmentation method, device, storage medium and equipment | |
| CN111127636B (en) | Intelligent complex intra-articular fracture desktop-level three-dimensional diagnosis system | |
| CN117649418B (en) | Chest multi-organ segmentation method and system and computer readable storage medium | |
| CN116344001B (en) | Medical information visual management system and method based on artificial intelligence | |
| CN115409771B (en) | Ovarian cancer survival prediction system constructed based on sample change stability characteristics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |