CN111091564A - Pulmonary nodule size detection system based on 3DUnet - Google Patents
Pulmonary nodule size detection system based on 3DUnet Download PDFInfo
- Publication number
- CN111091564A CN111091564A CN201911352399.9A CN201911352399A CN111091564A CN 111091564 A CN111091564 A CN 111091564A CN 201911352399 A CN201911352399 A CN 201911352399A CN 111091564 A CN111091564 A CN 111091564A
- Authority
- CN
- China
- Prior art keywords
- image
- intermediate image
- lung
- unit
- result
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 44
- 230000002685 pulmonary effect Effects 0.000 title claims abstract description 19
- 230000011218 segmentation Effects 0.000 claims abstract description 58
- 210000004072 lung Anatomy 0.000 claims abstract description 46
- 206010056342 Pulmonary mass Diseases 0.000 claims abstract description 32
- 230000003902 lesion Effects 0.000 claims abstract description 7
- 238000000034 method Methods 0.000 claims description 78
- 238000011176 pooling Methods 0.000 claims description 40
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 claims description 36
- 238000010606 normalization Methods 0.000 claims description 33
- 238000005070 sampling Methods 0.000 claims description 2
- 230000006870 function Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 3
- 206010058467 Lung neoplasm malignant Diseases 0.000 description 2
- 201000005202 lung cancer Diseases 0.000 description 2
- 208000020816 lung neoplasm Diseases 0.000 description 2
- 206010028980 Neoplasm Diseases 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 201000011510 cancer Diseases 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06T2207/20076—Probabilistic image processing
-
- 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
- G06T2207/30064—Lung nodule
Abstract
The invention provides a pulmonary nodule size detection system based on 3DUnet, which comprises a lung lobe segmentation module and a focus detection module; the lung lobe segmentation module is configured to read a lung CT image, remove irrelevant areas in the lung CT image and output a lung segmentation image; the lesion detection module is configured to read the lung segmentation image, locate a lesion, output a feature map, and output a location result. The lung nodule size detection system based on the 3DUnet can accurately give the lung nodule probability, the lung nodule central point and the lung nodule size information through the feature map and the positioning result.
Description
Technical Field
The invention relates to the technical field of pulmonary nodule detection, in particular to a pulmonary nodule size detection system based on 3 DUnet.
Background
The lung cancer always poses great threat to human life as a cancer with high incidence, lung nodules are one of early manifestation forms of the lung cancer, and whether the lung nodule detection is accurate and timely is very important for patients. The traditional pulmonary nodule detection system is to scan the lung of a patient by CT and the doctor manually labels the pulmonary nodule, which has great instability and inaccuracy. Some lung nodule detection systems based on deep learning also exist nowadays, but the detection systems have the problem of large detection error, and meanwhile, the training difficulty is increased by the detection model with a complex structure. Therefore, it is very necessary to invent a pulmonary nodule size detection system based on 3 DUnet.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: in order to solve the problems of large detection error and large difficulty in training a detection model of the conventional pulmonary nodule detection system, the invention provides a pulmonary nodule size detection system based on 3DUnet to solve the problems.
The technical scheme adopted by the invention for solving the technical problems is as follows: a pulmonary nodule size detection system based on 3DUnet comprises a lung lobe segmentation module and a focus detection module;
the lung lobe segmentation module is configured to read a lung CT image, remove irrelevant areas in the lung CT image and output a lung segmentation image; the lung lobe segmentation module comprises ten segmentation volume blocks, four maximum pooling layers and four upper sampling layers;
the focus detection module is configured to read the lung segmentation image, locate a focus, output a feature map, and output a location result; the focus detection module comprises an input residual block, a first residual block, a second residual block, a first deconvolution unit, a third residual block, a second deconvolution unit and an output residual block which are sequentially arranged;
the input residual block comprises two first convolution units, a first pooling unit and a first normalization unit which are arranged in sequence;
the first residual block comprises three second convolution units, a second pooling unit and a second normalization unit;
the second residual block comprises three third convolution units, a third pooling unit and a third normalization unit;
the third residual block comprises a fourth convolution unit, a fifth convolution unit and a fourth normalization unit;
the output residual block includes two sixth convolution units and a fifth normalization unit.
Preferably, the lung lobe segmentation module is specifically configured to:
a first segmentation volume block configured to process the lung CT image into a first result image;
the first largest pooling layer and the second split volume block are configured to process the first resultant image into a second resultant image;
a second largest pooling layer and a third split volume block configured to process the second resultant image into a third resultant image;
a third maximum pooling layer and a fourth split volume block configured to process the third resultant image into a fourth resultant image;
a fourth maximum pooling layer and a fifth split volume block configured to process the fourth resultant image into a fifth resultant image;
the first upsampling layer is configured to process the fifth resulting image into a first intermediate resulting image;
a sixth segmentation volume block configured to process a stitching result of the first intermediate result image and the fourth result image into a sixth result image;
a second upsampling layer configured to process the sixth resulting image into a second intermediate resulting image;
a seventh segmentation volume block configured to process a stitching result of the second intermediate result image and the third result image into a seventh result image;
a third upsampling layer configured to process the seventh resulting image into a third intermediate resulting image;
an eighth segmentation volume block configured to process a stitching result of the third intermediate result image and the second result image into an eighth result image;
a fourth upsampling layer configured to process the eighth resulting image into a fourth intermediate resulting image;
a ninth segmentation volume block and a tenth segmentation volume block are configured to process a stitching result of the fourth intermediate result image and the first result image into a lung segmentation image.
Preferably, the lesion detection module is specifically configured to:
the input residual block is configured to process the lung segmentation image into an input image;
a first one of the second convolution units is configured to process the input image into a first intermediate image, a second one of the second convolution units is configured to process the first intermediate image into a second intermediate image, a third one of the second convolution units is configured to process the second intermediate image into a third intermediate image, and the second pooling unit and the second normalization unit are configured to process the third intermediate image into the second residual block;
a first one of the third convolution units is configured to process the third intermediate image processed by the second pooling unit and the second normalization unit into a fourth intermediate image, a second one of the third convolution units is configured to process the fourth intermediate image into a fifth intermediate image, a third one of the third convolution units is configured to process the fifth intermediate image into a sixth intermediate image, and the third pooling unit and the third normalization unit are configured to process the sixth intermediate image and input the sixth intermediate image into the first deconvolution unit;
the first deconvolution unit is configured to process the sixth intermediate image processed by the third pooling unit and the third normalization unit into a seventh intermediate image;
the seventh intermediate image is processed by a Sigmoid function and then multiplied by a superposed image of the fourth intermediate image, the fifth intermediate image and the sixth intermediate image to obtain an eighth intermediate image;
adding the eighth intermediate image and the superposed image of the fourth intermediate image, the fifth intermediate image and the sixth intermediate image to obtain a ninth intermediate image;
the fourth convolution unit, fifth convolution unit, and fourth normalization unit are configured to process the ninth intermediate image into a tenth intermediate image;
the tenth intermediate image is processed by a Sigmoid function and then multiplied by a superposed image of the first intermediate image, the second intermediate image and the third intermediate image to obtain an eleventh intermediate image;
adding the eleventh intermediate image and a superposed image of the first intermediate image, the second intermediate image and the third intermediate image to obtain a twelfth intermediate image;
the two sixth convolution units and the fifth normalization unit are configured to process the twelfth intermediate image into a feature map.
Preferably, the positioning result is determined by the following formula:
(P,Ox,Oy,Oz,Sx,Sy,Sz)；
in the formula:
p is the lung nodule probability;
Oxis the X-axis coordinate of the lung nodule;
Oyis the Y-axis coordinate of the lung nodule;
Ozis the Z-axis coordinate of the pulmonary nodule;
Sxthe length of the region of the pulmonary nodule in the X-axis direction;
Sythe length of the region where the lung nodule is located in the Y-axis direction;
Szthe length of the region of the lung nodule in the Z-axis direction.
Preferably, the dimension of the lung segmentation image is 200 × 224 × 320, the dimension of the feature map is 100 × 112 × 160, and the step size of the feature map is 2.
The lung nodule size detection system based on the 3DUnet has the advantages that the lung nodule probability, the lung nodule central point and the lung nodule size information can be accurately given through the characteristic diagram and the positioning result; the focus detection module cancels the setting and calculation of anchors and NMS, and reduces the training difficulty of the focus detection module.
Drawings
The invention is further illustrated with reference to the following figures and examples.
Fig. 1 is a schematic structural diagram of a lung lobe segmentation module of a lung nodule size detection system based on 3DUnet according to the present invention.
Fig. 2 is a schematic structural diagram of a lesion detection module of a lung nodule size detection system based on 3DUnet according to the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.
In the description of the present invention, it is to be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "axial", "radial", "circumferential", and the like, indicate orientations and positional relationships based on the orientations and positional relationships shown in the drawings, and are used merely for convenience of description and for simplicity of description, and do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore, should not be considered as limiting the present invention.
Furthermore, the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. In the description of the present invention, it is to be noted that, unless otherwise explicitly specified or limited, the terms "connected" and "connected" are to be interpreted broadly, e.g., as being fixed or detachable or integrally connected; can be mechanically or electrically connected; may be directly connected or indirectly connected through an intermediate. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art. In addition, in the description of the present invention, "a plurality" means two or more unless otherwise specified.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and alternate implementations are included within the scope of the preferred embodiment of the present invention in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present invention.
As shown in fig. 1-2, the present invention provides a lung nodule size detection system based on 3DUnet, which includes a lung lobe segmentation module and a lesion detection module.
The lung lobe segmentation module includes ten segmentation volume blocks, four maximum pooling layers, and four upsampling layers. The lung lobe segmentation module is configured to read a lung CT image, remove irrelevant areas in the lung CT image and output a lung segmentation image, and specifically comprises:
the first segmented volume block is configured to process the lung CT image into a first resultant image having dimensions 96 x 96, the first segmented volume block having a number of output channels of 32.
The first max pooling layer and the second split volume block are configured to process the first result image into a second result image having dimensions of 48 x 48, the first max pooling layer has a number of output channels of 32, and the second split volume block has a number of output channels of 64.
The second max pooling layer and the third split volume block are configured to process the second resultant image into a third resultant image having dimensions of 24 x 24, the second max pooling layer has a number of output channels of 64, and the third split volume block has a number of output channels of 128.
The third maximum pooling layer and the fourth split volume block are configured to process the third result image into a fourth result image having dimensions of 12 × 12 × 12, the third maximum pooling layer has a number of output channels of 128, and the fourth split volume block has a number of output channels of 256.
The fourth maximum pooling layer and the fifth split volume block are configured to process the fourth resultant image into a fifth resultant image having dimensions of 6 × 6 × 6, the number of output channels of the fourth maximum pooling layer is 256, and the number of output channels of the fifth split volume block is 512.
The first upsampling layer is configured to process the fifth resulting image into a first intermediate resulting image.
The sixth segmentation volume block is configured to process a result of stitching the first intermediate result image and the fourth result image into a sixth result image, the sixth result image having dimensions of 12 × 12 × 12, the number of output channels of the first upsampling layer being 512, and the number of output channels of the sixth segmentation volume block being 256.
The second upsampling layer is configured to process the sixth resulting image into a second intermediate resulting image.
The seventh segmentation volume block is configured to process a stitching result of the second intermediate result image and the third result image into a seventh result image, the seventh result image having dimensions of 24 × 24 × 24, the number of output channels of the second upsampling layer being 256, and the number of output channels of the seventh segmentation volume block being 128.
The third upsampling layer is configured to process the seventh resulting image into a third intermediate resulting image.
The eighth segmentation volume block is configured to process a result of stitching the third intermediate result image and the second result image into an eighth result image having dimensions of 48 × 48 × 48, the number of output channels of the third upsampling layer being 128, and the number of output channels of the eighth segmentation volume block being 64.
The fourth upsampling layer is configured to process the eighth resulting image into a fourth intermediate resulting image.
The ninth and tenth segmentation volume blocks are configured to process a stitching result of the fourth intermediate result image and the first result image as a lung segmentation image. The dimensions of the lung segmentation image are 96 × 96 × 96, the number of output channels of the fourth upsampled layer is 64, the number of output channels of the ninth segmentation volume block is 32, and the number of output channels of the tenth segmentation volume block is 6.
In the embodiment of the present disclosure, the obtained lung segmentation image with the dimension of 96 × 96 × 96 needs to be processed into a lung segmentation image with the dimension of 200 × 224 × 320, and then the lung segmentation image with the dimension of 200 × 224 × 320 needs to be input to the lesion detection module.
The focus detection module comprises an input residual block, a first residual block, a second residual block, a first deconvolution unit, a third residual block, a second deconvolution unit and an output residual block which are sequentially arranged.
The input residual block comprises two first convolution units, a first pooling unit and a first normalization unit which are arranged in sequence.
The first residual block includes three second convolution units, a second pooling unit, and a second normalization unit.
The second residual block includes three third convolution units, a third pooling unit, and a third normalization unit.
The third residual block includes a fourth convolution unit, a fifth convolution unit, and a fourth normalization unit.
The output residual block includes two sixth convolution units and a fifth normalization unit.
The focus detection module is configured to read the lung segmentation image, locate a focus, output a feature map and output a location result, and specifically includes:
the input residual block is configured to process the lung segmentation image into an input image.
The first second convolution unit is configured to process the input image into a first intermediate image, the second convolution unit is configured to process the first intermediate image into a second intermediate image, the third second convolution unit is configured to process the second intermediate image into a third intermediate image, and the second pooling unit and the second normalization unit are configured to process the third intermediate image into a second residual block.
The first third convolution unit is configured to process a third intermediate image processed by the second pooling unit and the second normalization unit into a fourth intermediate image, the second third convolution unit is configured to process the fourth intermediate image into a fifth intermediate image, the third convolution unit is configured to process the fifth intermediate image into a sixth intermediate image, and the third pooling unit and the third normalization unit are configured to process the sixth intermediate image and input the sixth intermediate image into the first deconvolution unit.
The first deconvolution unit is configured to process the sixth intermediate image processed by the third pooling unit and the third normalization unit into a seventh intermediate image.
And the seventh intermediate image is multiplied by the superposed image of the fourth intermediate image, the fifth intermediate image and the sixth intermediate image after being processed by the Sigmoid function to obtain an eighth intermediate image.
And adding the eighth intermediate image and the superposed image of the fourth intermediate image, the fifth intermediate image and the sixth intermediate image to obtain a ninth intermediate image.
The fourth convolution unit, the fifth convolution unit, and the fourth normalization unit are configured to process the ninth intermediate image into a tenth intermediate image.
And the tenth intermediate image is multiplied by the superposed image of the first intermediate image, the second intermediate image and the third intermediate image after being processed by a Sigmoid function to obtain an eleventh intermediate image.
The eleventh intermediate image is added to the superimposed image of the first intermediate image, the second intermediate image, and the third intermediate image to obtain a twelfth intermediate image.
The two sixth convolution units and the fifth normalization unit are configured to process the twelfth intermediate image into a feature map.
In this embodiment, the dimension of the feature map is 100 × 112 × 160, and the step size of the feature map is 2. The positioning result is determined by the following formula:
(P,Ox,Oy,Oz,Sx,Sy,Sz)；
in the formula:
p is the pulmonary nodule probability;
Oxis the X-axis coordinate of the lung nodule;
Oyis the Y-axis coordinate of the lung nodule;
Ozis the Z-axis coordinate of the pulmonary nodule;
Sxthe length of the region of the pulmonary nodule in the X-axis direction;
Sythe length of the region where the lung nodule is located in the Y-axis direction;
Szthe length of the region of the lung nodule in the Z-axis direction.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, a schematic representation of the term does not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
In light of the foregoing description of the preferred embodiment of the present invention, many modifications and variations will be apparent to those skilled in the art without departing from the spirit and scope of the invention. The technical scope of the present invention is not limited to the content of the specification, and must be determined according to the scope of the claims.
Claims (5)
1. A pulmonary nodule size detection system based on 3DUnet is characterized by comprising a lung lobe segmentation module and a focus detection module;
the lung lobe segmentation module is configured to read a lung CT image, remove irrelevant areas in the lung CT image and output a lung segmentation image; the lung lobe segmentation module comprises ten segmentation volume blocks, four maximum pooling layers and four upper sampling layers;
the focus detection module is configured to read the lung segmentation image, locate a focus, output a feature map, and output a location result; the focus detection module comprises an input residual block, a first residual block, a second residual block, a first deconvolution unit, a third residual block, a second deconvolution unit and an output residual block which are sequentially arranged;
the input residual block comprises two first convolution units, a first pooling unit and a first normalization unit which are arranged in sequence;
the first residual block comprises three second convolution units, a second pooling unit and a second normalization unit;
the second residual block comprises three third convolution units, a third pooling unit and a third normalization unit;
the third residual block comprises a fourth convolution unit, a fifth convolution unit and a fourth normalization unit;
the output residual block includes two sixth convolution units and a fifth normalization unit.
2. The 3d net-based pulmonary nodule size detection system of claim 1, wherein:
the lung lobe segmentation module is specifically configured to:
a first segmentation volume block configured to process the lung CT image into a first result image;
the first largest pooling layer and the second split volume block are configured to process the first resultant image into a second resultant image;
a second largest pooling layer and a third split volume block configured to process the second resultant image into a third resultant image;
a third maximum pooling layer and a fourth split volume block configured to process the third resultant image into a fourth resultant image;
a fourth maximum pooling layer and a fifth split volume block configured to process the fourth resultant image into a fifth resultant image;
the first upsampling layer is configured to process the fifth resulting image into a first intermediate resulting image;
a sixth segmentation volume block configured to process a stitching result of the first intermediate result image and the fourth result image into a sixth result image;
a second upsampling layer configured to process the sixth resulting image into a second intermediate resulting image;
a seventh segmentation volume block configured to process a stitching result of the second intermediate result image and the third result image into a seventh result image;
a third upsampling layer configured to process the seventh resulting image into a third intermediate resulting image;
an eighth segmentation volume block configured to process a stitching result of the third intermediate result image and the second result image into an eighth result image;
a fourth upsampling layer configured to process the eighth resulting image into a fourth intermediate resulting image;
a ninth segmentation volume block and a tenth segmentation volume block are configured to process a stitching result of the fourth intermediate result image and the first result image into a lung segmentation image.
3. A 3DUnet based lung nodule size detection system as claimed in claim 2 wherein:
the lesion detection module is specifically configured to:
the input residual block is configured to process the lung segmentation image into an input image;
a first one of the second convolution units is configured to process the input image into a first intermediate image, a second one of the second convolution units is configured to process the first intermediate image into a second intermediate image, a third one of the second convolution units is configured to process the second intermediate image into a third intermediate image, and the second pooling unit and the second normalization unit are configured to process the third intermediate image into the second residual block;
a first one of the third convolution units is configured to process the third intermediate image processed by the second pooling unit and the second normalization unit into a fourth intermediate image, a second one of the third convolution units is configured to process the fourth intermediate image into a fifth intermediate image, a third one of the third convolution units is configured to process the fifth intermediate image into a sixth intermediate image, and the third pooling unit and the third normalization unit are configured to process the sixth intermediate image and input the sixth intermediate image into the first deconvolution unit;
the first deconvolution unit is configured to process the sixth intermediate image processed by the third pooling unit and the third normalization unit into a seventh intermediate image;
the seventh intermediate image is processed by a Sigmoid function and then multiplied by a superposed image of the fourth intermediate image, the fifth intermediate image and the sixth intermediate image to obtain an eighth intermediate image;
adding the eighth intermediate image and the superposed image of the fourth intermediate image, the fifth intermediate image and the sixth intermediate image to obtain a ninth intermediate image;
the fourth convolution unit, fifth convolution unit, and fourth normalization unit are configured to process the ninth intermediate image into a tenth intermediate image;
the tenth intermediate image is processed by a Sigmoid function and then multiplied by a superposed image of the first intermediate image, the second intermediate image and the third intermediate image to obtain an eleventh intermediate image;
adding the eleventh intermediate image and a superposed image of the first intermediate image, the second intermediate image and the third intermediate image to obtain a twelfth intermediate image;
the two sixth convolution units and the fifth normalization unit are configured to process the twelfth intermediate image into a feature map.
4. A 3DUnet based lung nodule size detection system as claimed in claim 3 wherein:
the positioning result is determined by the following formula:
(P，Ox，Oy，Oz，Sx，Sy，Sz)；
in the formula:
p is the lung nodule probability;
Oxx-axis coordinates of the lung nodules in the feature map;
Oyis the Y-axis coordinate of the lung nodule in the feature map;
Ozis the Z-axis coordinate of the lung nodule in the feature map;
Sxthe length of the region where the lung nodule is located in the X-axis direction in the feature map;
Sythe length of the region where the lung nodule is located in the Y-axis direction in the feature map;
Szthe length of the region where the lung nodule is located in the Z-axis direction in the feature map.
5. The 3d net-based pulmonary nodule size detection system of claim 4, wherein:
the dimension of the lung segmentation image is 200 × 224 × 320, the dimension of the feature map is 100 × 112 × 160, and the step size of the feature map is 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911352399.9A CN111091564B (en) | 2019-12-25 | Lung nodule size detecting system based on 3DUnet |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911352399.9A CN111091564B (en) | 2019-12-25 | Lung nodule size detecting system based on 3DUnet |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111091564A true CN111091564A (en) | 2020-05-01 |
| CN111091564B CN111091564B (en) | 2024-04-26 |
Family
ID=
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114820584A (en) * | 2022-05-27 | 2022-07-29 | 北京安德医智科技有限公司 | Lung focus positioner |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108765369A (en) * | 2018-04-20 | 2018-11-06 | 平安科技（深圳）有限公司 | Detection method, device, computer equipment and the storage medium of Lung neoplasm |
| WO2018222755A1 (en) * | 2017-05-30 | 2018-12-06 | Arterys Inc. | Automated lesion detection, segmentation, and longitudinal identification |
| CN109685776A (en) * | 2018-12-12 | 2019-04-26 | 华中科技大学 | A kind of pulmonary nodule detection method based on ct images and system |
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018222755A1 (en) * | 2017-05-30 | 2018-12-06 | Arterys Inc. | Automated lesion detection, segmentation, and longitudinal identification |
| CN108765369A (en) * | 2018-04-20 | 2018-11-06 | 平安科技（深圳）有限公司 | Detection method, device, computer equipment and the storage medium of Lung neoplasm |
| WO2019200740A1 (en) * | 2018-04-20 | 2019-10-24 | 平安科技（深圳）有限公司 | Pulmonary nodule detection method and apparatus, computer device, and storage medium |
| CN109685776A (en) * | 2018-12-12 | 2019-04-26 | 华中科技大学 | A kind of pulmonary nodule detection method based on ct images and system |
Non-Patent Citations (3)
| Title |
|---|
| MIDL 2018: "Recurrent Residual Convolutional Neural Network based on U-Net (R2U-Net) for Medical Image Segmentation", <MIDL 2018>, pages 2 - 4 * |
| OZG UN C I CEK: "3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation", <MICCAI 2016> * |
| 褚晶辉;李晓川;张佳祺;吕卫;: "一种基于级联卷积网络的三维脑肿瘤精细分割", 激光与光电子学进展, no. 10 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114820584A (en) * | 2022-05-27 | 2022-07-29 | 北京安德医智科技有限公司 | Lung focus positioner |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP5683065B2 (en) | Improved system and method for positive displacement registration | |
| JP4800127B2 (en) | Medical image segmentation device and medical image segmentation program | |
| CN1890689A (en) | Elastic image registration | |
| ES2875919T3 (en) | Similar case image search program, similar case image search device and similar case image search method | |
| US20080069477A1 (en) | Result filter and method for selecting the result data of an application for automatic pattern recognition | |
| CN104586418B (en) | medical image data processing apparatus and medical image data processing method | |
| CN113674253B (en) | Automatic segmentation method for rectal cancer CT image based on U-transducer | |
| CN110363802B (en) | Prostate image registration system and method based on automatic segmentation and pelvis alignment | |
| US10603016B2 (en) | Image processing apparatus, method of controlling the same, and non-transitory computer-readable storage medium | |
| CN105913442A (en) | Method for automatically matching pulmonary nodules | |
| US11200443B2 (en) | Image processing apparatus, image processing method, and image processing system | |
| JP7101809B2 (en) | Image processing equipment, image processing methods, and programs | |
| WO2019037654A1 (en) | 3d image detection method and apparatus, electronic device, and computer readable medium | |
| CN109191465A (en) | A kind of system for being determined based on deep learning network, identifying human body or so the first rib cage | |
| CN113436173A (en) | Abdomen multi-organ segmentation modeling and segmentation method and system based on edge perception | |
| CN110910348B (en) | Method, device, equipment and storage medium for classifying positions of pulmonary nodules | |
| US20100215228A1 (en) | Medical image displaying apparatus, medical image displaying method, and medical image displaying program | |
| Costa et al. | Data augmentation for detection of architectural distortion in digital mammography using deep learning approach | |
| CN111091564B (en) | Lung nodule size detecting system based on 3DUnet | |
| CN111091564A (en) | Pulmonary nodule size detection system based on 3DUnet | |
| CN110992310A (en) | Method and device for determining partition where mediastinal lymph node is located | |
| US20100260393A1 (en) | Navigation guide | |
| Foo et al. | Interactive segmentation for covid-19 infection quantification on longitudinal ct scans | |
| CN111751371B (en) | Immunohistochemical digital slide reading system and method | |
| JP4393135B2 (en) | Radiation image processing apparatus, radiation image processing method, computer program, and computer-readable recording medium |
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 |