CN112950554A - Lung lobe segmentation optimization method and system based on lung segmentation - Google Patents
Lung lobe segmentation optimization method and system based on lung segmentation Download PDFInfo
- Publication number
- CN112950554A CN112950554A CN202110164279.7A CN202110164279A CN112950554A CN 112950554 A CN112950554 A CN 112950554A CN 202110164279 A CN202110164279 A CN 202110164279A CN 112950554 A CN112950554 A CN 112950554A
- Authority
- CN
- China
- Prior art keywords
- lung
- result
- segmentation
- lobe
- connected domain
- 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
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- 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
Abstract
The invention discloses a lung segmentation-based lung lobe segmentation optimization method and system, and relates to the field of medical images. The method comprises the following steps: step 1, obtaining a first output result of a first lung automatic segmentation algorithm model and a second output result of a second lung lobe automatic segmentation algorithm model, wherein the precision of the first lung automatic segmentation algorithm model is higher than that of the second lung lobe automatic segmentation algorithm model; step 2, performing point multiplication on the first output result and the second output result to obtain an independent mask of each lung lobe; step 3, respectively obtaining a maximum connected domain matrix of the independent mask of each lung lobe; step 4, performing superposition processing on each maximum connected domain matrix to obtain first optimized data; and 5, calculating and eliminating the first optimized data to obtain a final optimized result. The invention can achieve the effect of improving the prediction precision.
Description
Technical Field
The invention relates to the field of medical images, in particular to a lung lobe segmentation optimization method and system based on lung segmentation.
Background
In the field of medical images, because training data is difficult to acquire, neural networks are often difficult to train to an optimal state, which may lead to inaccuracy of prediction results. The lung lobe segmentation is difficult to delineate, and only a small number of training samples can be obtained, so that the lung lobe segmentation generally needs to be subjected to post-processing, so that errors in a prediction result are reduced as much as possible. Compared with lung lobes, the lung has more clear boundaries, so that the delineation difficulty is much smaller, and the training data can be obtained easily; meanwhile, as the boundary characteristics of the lung are clear, the neural network is easier to train, and the prediction result is more accurate.
In the deep learning lung lobe segmentation prediction task, training is often performed in a mask-supervised manner. Namely, the output of the deep neural network is a matrix with the same size as the original input image, and each pixel of the matrix is filled with 0 or an integer of 1-5, wherein 0 represents that the pixel position is predicted to be a non-lung region, and 1-5 represents that the pixel position corresponds to different lung lobes. This prediction method often results in that each of the predicted lung lobe masks may be discontinuous, and the accuracy of the prediction of the lung lobe partitions is affected.
Disclosure of Invention
The invention aims to solve the technical problem of the prior art and provides a lung lobe segmentation optimization method and system based on lung segmentation.
The technical scheme for solving the technical problems is as follows: a lung lobe segmentation optimization method based on lung segmentation comprises the following steps:
step 1, obtaining a first output result of a first lung automatic segmentation algorithm model and a second output result of a second lung lobe automatic segmentation algorithm model, wherein the precision of the first lung automatic segmentation algorithm model is higher than that of the second lung lobe automatic segmentation algorithm model;
step 2, performing point multiplication on the first output result and the second output result to obtain an independent mask of each lung lobe;
step 3, respectively obtaining a maximum connected domain matrix of the independent mask of each lung lobe;
step 4, performing superposition processing on each maximum connected domain matrix to obtain first optimized data;
and 5, calculating and eliminating the first optimized data to obtain a final optimized result.
The invention has the beneficial effects that: the output results can be optimized through the calculation processing of the two model output results, the lung lobe segmentation prediction results are optimized through the lung segmentation prediction results, meanwhile, the characteristics of the lung lobes are combined, error results in the model processing results can be corrected, and the post-processing results are more accurate.
Further, step 2 specifically comprises:
and performing point multiplication on the first output result and the second output result to obtain first data, and performing formula operation on the first data to obtain the independent masks of the five lung lobes.
Further, step 3 specifically comprises:
and correspondingly obtaining five maximum connected domain matrixes through the identification of the five masks.
Further, step 4 specifically comprises:
step 401, adding the five maximum connected domain matrixes to obtain a superposition matrix;
and 402, performing exclusive-or operation on the first result and the superposition matrix to obtain first optimized data which is removed from the first result and does not belong to the superposition matrix.
The beneficial effect of adopting the above further scheme is that the result can be more refined through rejecting, and the subsequent optimization effect can be better.
Further, step 5 specifically comprises:
step 501, performing connected domain calculation on the first optimized data to obtain label image data;
step 502, performing dilation operation on each maximum connected domain image data in the label image data to obtain a dilation operation result, and eliminating the maximum connected domain image data in the dilation operation result to obtain optimized second optimized data;
step 503, multiplying the second optimized data by the second result to obtain a final optimized result.
Another technical solution of the present invention for solving the above technical problems is as follows: a lung segmentation-based lung lobe segmentation optimization system, comprising:
the system comprises an acquisition module, a calculation module and a display module, wherein the acquisition module is used for acquiring a first output result of a first lung automatic segmentation algorithm model and a second output result of a second lung lobe automatic segmentation algorithm model, and the precision of the first lung automatic segmentation algorithm model is higher than that of the second lung lobe automatic segmentation algorithm model;
the first optimization module is used for performing point multiplication on the first output result and the second output result to obtain an independent mask of each lung lobe;
the calculation module is used for respectively acquiring the maximum connected domain matrix of the independent mask of each lung lobe;
the second optimization module is used for performing superposition processing on each maximum connected domain matrix to obtain first optimization data;
and the result module is used for calculating and eliminating the first optimized data to obtain a final optimized result.
The invention has the beneficial effects that: the output results can be optimized through the calculation processing of the two model output results, the lung lobe segmentation prediction results are optimized through the lung segmentation prediction results, meanwhile, the characteristics of the lung lobes are combined, error results in the model processing results can be corrected, and the post-processing results are more accurate.
Further, the first optimization module is specifically configured to:
and performing point multiplication on the first output result and the second output result to obtain first data, and performing formula operation on the first data to obtain the independent masks of the five lung lobes.
Further, the calculation module is specifically configured to:
and correspondingly obtaining five maximum connected domain matrixes through the identification of the five masks.
Further, the second optimization module is specifically configured to:
and adding the five maximum connected domain matrixes to obtain a superposed matrix, and performing exclusive-OR operation on the first result and the superposed matrix to obtain first optimized data which is removed from the first result and does not belong to the superposed matrix.
The beneficial effect of adopting the above further scheme is that the result can be more refined through rejecting, and the subsequent optimization effect can be better.
Further, the result module is specifically configured to:
performing connected domain calculation on the first optimized data to obtain tag image data, performing expansion operation on each maximum connected domain image data in the tag image data to obtain an expansion operation result, eliminating the maximum connected domain image data in the expansion operation result to obtain optimized second optimized data, and multiplying the second optimized data with the second result to obtain a final optimized result.
Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
Fig. 1 is a schematic flowchart of a lung segmentation-based lung lobe segmentation optimization method according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram provided by an embodiment of a lung segmentation-based lung lobe segmentation optimization system according to the present invention.
In the drawings, the components represented by the respective reference numerals are listed below:
100. the system comprises an acquisition module 200, a first optimization module 300, a calculation module 400, a second optimization module 500 and a result module.
Detailed Description
The principles and features of this invention are described below in conjunction with the following drawings, which are set forth to illustrate, but are not to be construed to limit the scope of the invention.
As shown in fig. 1, a lung lobe segmentation optimization method based on lung segmentation includes:
step 1, obtaining a first output result of a first lung automatic segmentation algorithm model and a second output result of a second lung lobe automatic segmentation algorithm model, wherein the precision of the first lung automatic segmentation algorithm model is higher than that of the second lung lobe automatic segmentation algorithm model;
step 2, point-multiplying the first output result and the second output result to obtain an independent mask of each lung lobe;
step 3, respectively obtaining a maximum connected domain matrix of the independent mask of each lung lobe;
step 4, performing superposition processing on each maximum connected domain matrix to obtain first optimized data;
and 5, calculating and eliminating the first optimized data to obtain a final optimized result.
In some possible embodiments, the calculation processing of the two model output results can optimize the output results, combine the characteristics of lung lobes while optimizing the prediction results of lung lobe segmentation by using the prediction results of lung segmentation, and correct the error results in the model processing results, so that the post-processing results are more accurate.
It should be noted that the first lung automatic segmentation algorithm model may be any one of the lung automatic segmentation algorithms, the second lung lobe automatic segmentation algorithm model may be any one of the lung lobe automatic segmentation algorithms, may be a deep learning algorithm, may also be a net model, but is not limited to the net model, and may also be a conventional segmentation algorithm, the output result of the first lung automatic segmentation algorithm model is a lung mask, the lung region is a foreground, the foreground pixel filling is 1, the non-lung region is a background, and the background filling is 0; the output result of the second lung lobe automatic segmentation algorithm model is a lung lobe mask, the pixel values of five regions, namely the left lung upper lobe, the left lung lower lobe, the right lung upper lobe, the right lung middle lobe and the right lung lower lobe, are respectively filled with 1/2/3/4/5, and the other background regions are filled with 0;
for oneObtaining an original image IMG of Lung Lobe segmentation of the original image IMG to be predicted by using a first Lung automatic segmentation algorithm model to obtain an automatic Lung segmentation result Lung, and obtaining an automatic Lung Lobe segmentation algorithm result Lobe by using a second Lung Lobe automatic segmentation algorithm modelorigin(ii) a Will LobeoriginMultiply by Lung in order to multiply LobeoriginIn the foreground, the pixel point which is not considered by Lung as the foreground is set as 0 to obtain Lobeorigin'; because the foreground in Lung is 1 and the background is 0, after dot multiplication, the pixel points considered as the background in Lung will all become 0, and the points considered as the foreground in Lung still maintain the original value. Therefore, the result of dot multiplication is to remove LobeoriginThe point which is not considered as a foreground by the Lung is set as 0, and the purpose of removing is achieved. According to Lobeorigin' obtaining a separate mask for each lung lobe. I.e. 5 matrices are obtained, which are consistent with the original input image size and which fill values of 0 or 1 per pixel. 0 represents that the pixel position is not the current lung lobe position, and 1 corresponds to the current lung lobe position. Naming these 5 masks as Lobe1～Lobe5；Lobeorigin' and LobeoriginSimilarly, a fill of 0 indicates that the pixel position is a non-lung region, 1 indicates that the pixel position is a left lung upper lobe, 2 indicates that the pixel position is a left lung lower lobe, 3 indicates that the pixel position is a right lung upper lobe, 4 indicates that the pixel position is a right lung middle lobe, and 5 indicates that the pixel position is a right lung lower lobe, the mask contains information of five lung lobes, that is, the mask containing information of five lung lobes is split into five masks, each mask has a filled pixel value consisting of only 0 and 1, 1 indicates that the pixel corresponds to a foreground of the lung lobe, 0 indicates that the position corresponding to the pixel does not belong to a range of the lung lobe, and the mask is a matrix with the same size as that of the original picture. Each pixel in the mask image is filled with 0 or 1, which represents that the position in the original image is a background or a foreground; to each LobeiWherein i ∈ [1, 2, 3, 4, 5 ]]Obtaining the maximum connected domain matrix Lobe thereofi_maxConnect，Lobei_maxConnectThe foreground in (1) is considered to belong to the lung lobe i, and the foreground is the region filled with 1.
A communication area: the image area is composed of foreground pixel points which have the same pixel value and are adjacent in position in the image.
Connected region labeling image: the same connected domain is filled with the same value, different connected domains are filled with different values, and the background is filled with 0.
And (3) analyzing a connected region: finding and marking each connected region in the mask image to obtain a connected region marked image corresponding to the mask image;
connected component analysis algorithm:
the first step is as follows: the image is scanned line by line, the successive white pixels in each line are grouped into a sequence called a blob, and its start, its end, and the line number in which it is located are noted.
The second step is that: for a blob in all rows except the first row, if it has no overlap with all blobs in the previous row, giving it a new label; if it has a coincidence region with only one blob in the previous row, assigning the reference number of the blob in the previous row to it; if it has an overlap area with more than 2 clusters in the previous row, the current cluster is assigned a minimum label of the connected cluster and the labels of the several clusters in the previous row are written into the equivalence pairs, indicating that they belong to one class.
The third step: equivalent pairs are converted to equivalent sequences, each of which is given the same reference numeral because they are equivalent. Starting with 1, each equivalent sequence is given a reference number.
The fourth step: the labels of the start cliques are traversed, equivalent sequences are searched, and new labels are given to the equivalent sequences.
The fifth step: the label of each blob is filled in the label image.
And a sixth step: and (6) ending.
The above algorithm is one of the methods for acquiring the connected region marker image, and other algorithms may be used to acquire the connected region marker image.
Maximum connected region image: marking an image in a connected region, finding a connected region with the maximum number of pixels (counting the number of pixels in each connected region), assigning the pixel points in the region to be 1, and assigning the other pixel points to be 10, obtaining the mask which is the image of the maximum connected region; mixing 5 Lobei_maxConnectAdding to obtain Lobeall_maxConnect(ii) a According to the formula Loberest＝Lung xor Lobeall_maxConnectObtain the matrix LoberestThe purpose of this equation is to find the Lung but not the Lobeall_maxConnectIn the formula, xor represents exclusive or; to LoberestCalculating a connected domain to obtain a label image Loberest_label(ii) a For Loberest_labelEach of the maximum connected region images Lobe in (1)rest_label_jFirstly, performing dilation operation, and then subtracting Lobe from the result of dilation operationrest_label_jTo obtain Loberest_label_j_round。
Expansion: image a is convolved with a kernel B of arbitrary shape, typically a small square or circle. Core B has a definable anchor point, typically defined as the core center point. When the expansion operation is carried out, the kernel B is drawn through the image, the maximum pixel value of the coverage area of the kernel B is extracted, and the pixel at the anchor point position is replaced.
Will Loberest_label_j_roundAnd LobeoriginMultiplying, counting to obtain the lung Lobe to which the pixel adjacent to the connected region j belongs, and then using the connected region Loberest_label_jAnd (4) dividing the lung into the lung lobes. If a plurality of lung lobes are adjacent to the connected region j, dividing the lung lobes into the lung lobes with the largest pixel ratio; loberest_label_j_roundIs actually Loberest_label_jA circle of neighboring pixels. Will Loberest_label_j_roundAnd LobeoriginMultiplication, Loberest_label_j_roundThe pixels with 0 in are all set to 0, Loberest_label_j_roundPixel fill and Lobe of 1originIs consistent with (1). Due to LobeoriginA middle fill of 0 indicates that the pixel location is a non-lung region, 1 indicates that the pixel location is a left superior lung Lobe, 2 indicates that the pixel location is a left inferior lung Lobe, 3 indicates that the pixel location is a right superior lung Lobe, 4 indicates that the pixel location is a right middle lung Lobe, and 5 indicates that the pixel location is a right inferior lung Lobe, thus Loberest_label_j_roundAnd LobeoriginIn the multiplication result, frontThe scene value is which number, representing which lobe this pixel belongs to. Statistics Loberest_label_j_roundAnd LobeoriginThe result of the multiplication, the most significant number, represents Loberest_label_j_roundThat the neighboring pixel belongs to which Lobe is the most, i.e. describes Loberest_label_jThe area corresponding to the foreground of the mask is more likely to belong to which lobe.
For example, the following steps are carried out: loberest_label_j_roundAnd LobeoriginAs a result of the multiplication, 421 pixels having a value of 2, 120 pixels having a value of 1, and 0 pixels having a value of 3/4/5 are included in addition to the pixels having a value of 0. Then, consider Loberest_label_jThe surrounding pixel ratio is 2 at most, so Loberest_label_jMore likely to belong to the same region as 2, so Loberest_label_jAnd dividing the lung lobes into lung lobes corresponding to 2, namely the left lower lung lobe, so far, finishing the post-processing step of the invention, and enabling each lung lobe of the processed lung lobe segmentation mask matrix to be communicated and completely conform to the shape of the lung.
Preferably, in any of the above embodiments, step 2 is specifically:
and performing point multiplication on the first output result and the second output result to obtain first data, and performing formula operation on the first data to obtain the independent masks of the five lung lobes.
Preferably, in any of the above embodiments, step 3 is specifically:
and correspondingly obtaining five maximum connected domain matrixes through the identification of the five masks.
Preferably, in any of the above embodiments, step 4 is specifically:
step 401, adding the five maximum connected domain matrixes to obtain a superposition matrix;
and 402, performing exclusive-or operation on the first result and the superposition matrix to obtain first optimized data which is removed from the first result and does not belong to the superposition matrix.
In some possible embodiments, the result can be refined through elimination, and the subsequent optimization effect can be better.
Preferably, in any of the above embodiments, step 5 is specifically:
step 501, performing connected domain calculation on the first optimized data to obtain label image data;
step 502, performing dilation operation on each maximum connected domain image data in the label image data to obtain a dilation operation result, and eliminating the maximum connected domain image data in the dilation operation result to obtain optimized second optimized data;
and 503, multiplying the second optimization data by the second result to obtain a final optimization result.
As shown in fig. 2, a lung segmentation-based lung lobe segmentation optimization system includes:
an obtaining module 100, configured to obtain a first output result of a first lung automatic segmentation algorithm model and a second output result of a second lung lobe automatic segmentation algorithm model, where accuracy of the first lung automatic segmentation algorithm model is greater than that of the second lung lobe automatic segmentation algorithm model;
a first optimization module 200, configured to perform point multiplication on the first output result and the second output result to obtain an independent mask for each lung lobe;
a calculating module 300, configured to obtain a maximum connected domain matrix of an independent mask of each lung lobe;
a second optimization module 400, configured to perform superposition processing on each maximum connected domain matrix to obtain first optimization data;
and the result module 500 is configured to perform calculation and elimination processing on the first optimized data to obtain a final optimized result.
In some possible embodiments, the calculation processing of the two model output results can optimize the output results, combine the characteristics of lung lobes while optimizing the prediction results of lung lobe segmentation by using the prediction results of lung segmentation, and correct the error results in the model processing results, so that the post-processing results are more accurate.
Preferably, in any of the above embodiments, the first optimization module 200 is specifically configured to:
and performing point multiplication on the first output result and the second output result to obtain first data, and performing formula operation on the first data to obtain the independent masks of the five lung lobes.
Preferably, in any of the above embodiments, the calculation module 300 is specifically configured to:
and correspondingly obtaining five maximum connected domain matrixes through the identification of the five masks.
Preferably, in any of the above embodiments, the second optimization module 400 is specifically configured to:
and adding the five maximum connected domain matrixes to obtain a superposition matrix, and performing exclusive-or operation on the first result and the superposition matrix to obtain first optimized data which is removed from the first result and does not belong to the superposition matrix.
In some possible embodiments, the result can be refined through elimination, and the subsequent optimization effect can be better.
Preferably, in any of the above embodiments, the result module 500 is specifically configured to:
performing connected domain calculation on the first optimized data to obtain tag image data, performing expansion operation on each maximum connected domain image data in the tag image data to obtain an expansion operation result, eliminating the maximum connected domain image data in the expansion operation result to obtain optimized second optimized data, and multiplying the second optimized data with the second result to obtain a final optimized result.
It is understood that some or all of the alternative embodiments described above may be included in some embodiments.
It should be noted that the above embodiments are product embodiments corresponding to the previous method embodiments, and for the description of each optional implementation in the product embodiments, reference may be made to corresponding descriptions in the above method embodiments, and details are not described here again.
The reader should understand that in the description of this specification, reference to the description of the terms "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., means 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, the schematic representations of the terms used above are not necessarily intended to 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. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described method embodiments are merely illustrative, and for example, the division of steps into only one logical functional division may be implemented in practice in another way, for example, multiple steps may be combined or integrated into another step, or some features may be omitted, or not implemented.
The above method, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention essentially or partially contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
While the invention has been described with reference to specific embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. A lung lobe segmentation optimization method based on lung segmentation is characterized by comprising the following steps:
step 1, obtaining a first output result of a first lung automatic segmentation algorithm model and a second output result of a second lung lobe automatic segmentation algorithm model, wherein the precision of the first lung automatic segmentation algorithm model is higher than that of the second lung lobe automatic segmentation algorithm model;
step 2, performing point multiplication on the first output result and the second output result to obtain an independent mask of each lung lobe;
step 3, respectively obtaining a maximum connected domain matrix of the independent mask of each lung lobe;
step 4, performing superposition processing on each maximum connected domain matrix to obtain first optimized data;
and 5, calculating and eliminating the first optimized data to obtain a final optimized result.
2. The lung segmentation-based lung lobe segmentation optimization method according to claim 1, wherein the step 2 specifically comprises:
and performing point multiplication on the first output result and the second output result to obtain first data, and performing formula operation on the first data to obtain the independent masks of the five lung lobes.
3. The lung segmentation-based lung lobe segmentation optimization method according to claim 2, wherein the step 3 specifically comprises:
and correspondingly obtaining five maximum connected domain matrixes through the identification of the five masks.
4. The lung segmentation-based lung lobe segmentation optimization method according to claim 3, wherein the step 4 specifically comprises:
step 401, adding the five maximum connected domain matrixes to obtain a superposition matrix;
and 402, performing exclusive-or operation on the first result and the superposition matrix to obtain first optimized data which is removed from the first result and does not belong to the superposition matrix.
5. The lung segmentation-based lung lobe segmentation optimization method according to claim 4, wherein the step 5 specifically comprises:
step 501, performing connected domain calculation on the first optimized data to obtain label image data;
step 502, performing dilation operation on each maximum connected domain image data in the label image data to obtain a dilation operation result, and eliminating the maximum connected domain image data in the dilation operation result to obtain optimized second optimized data;
step 503, multiplying the second optimized data by the second result to obtain a final optimized result.
6. A lung segmentation-based lung lobe segmentation optimization system, comprising:
the system comprises an acquisition module, a calculation module and a display module, wherein the acquisition module is used for acquiring a first output result of a first lung automatic segmentation algorithm model and a second output result of a second lung lobe automatic segmentation algorithm model, and the precision of the first lung automatic segmentation algorithm model is higher than that of the second lung lobe automatic segmentation algorithm model;
the first optimization module is used for performing point multiplication on the first output result and the second output result to obtain an independent mask of each lung lobe;
the calculation module is used for respectively acquiring the maximum connected domain matrix of the independent mask of each lung lobe;
the second optimization module is used for performing superposition processing on each maximum connected domain matrix to obtain first optimization data;
and the result module is used for calculating and eliminating the first optimized data to obtain a final optimized result.
7. The lung segmentation-based lung lobe segmentation optimization system according to claim 6, wherein the first optimization module is specifically configured to:
and performing point multiplication on the first output result and the second output result to obtain first data, and performing formula operation on the first data to obtain the independent masks of the five lung lobes.
8. The lung segmentation-based lung lobe segmentation optimization system according to claim 7, wherein the calculation module is specifically configured to:
and correspondingly obtaining five maximum connected domain matrixes through the identification of the five masks.
9. The lung segmentation-based lung lobe segmentation optimization system according to claim 8, wherein the second optimization module is specifically configured to:
and adding the five maximum connected domain matrixes to obtain a superposed matrix, and performing exclusive-OR operation on the first result and the superposed matrix to obtain first optimized data which is removed from the first result and does not belong to the superposed matrix.
10. The system of claim 9, wherein the result module is specifically configured to:
performing connected domain calculation on the first optimized data to obtain tag image data, performing expansion operation on each maximum connected domain image data in the tag image data to obtain an expansion operation result, eliminating the maximum connected domain image data in the expansion operation result to obtain optimized second optimized data, and multiplying the second optimized data with the second result to obtain a final optimized result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110164279.7A CN112950554B (en) | 2021-02-05 | 2021-02-05 | Lung lobe segmentation optimization method and system based on lung segmentation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110164279.7A CN112950554B (en) | 2021-02-05 | 2021-02-05 | Lung lobe segmentation optimization method and system based on lung segmentation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112950554A true CN112950554A (en) | 2021-06-11 |
| CN112950554B CN112950554B (en) | 2021-12-21 |
Family
ID=76242870
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110164279.7A Active CN112950554B (en) | 2021-02-05 | 2021-02-05 | Lung lobe segmentation optimization method and system based on lung segmentation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112950554B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114862861A (en) * | 2022-07-11 | 2022-08-05 | 珠海横琴圣澳云智科技有限公司 | Lung lobe segmentation method and device based on few-sample learning |
| CN115049735A (en) * | 2022-08-12 | 2022-09-13 | 季华实验室 | Mask optimization processing method and device, electronic equipment and storage medium |
| CN117523207A (en) * | 2024-01-04 | 2024-02-06 | 广东欧谱曼迪科技股份有限公司 | Method, device, electronic equipment and storage medium for lung lobe segmentation correction processing |
| CN117576126A (en) * | 2024-01-16 | 2024-02-20 | 广东欧谱曼迪科技股份有限公司 | Optimization method and device for lung lobe segmentation, electronic equipment and storage medium |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170224301A1 (en) * | 2016-02-05 | 2017-08-10 | Pulmonx Corporation | Methods, systems, and devices for analyzing lung imaging data |
| CN107230204A (en) * | 2017-05-24 | 2017-10-03 | 东北大学 | A kind of method and device that the lobe of the lung is extracted from chest CT image |
| CN107480675A (en) * | 2017-07-28 | 2017-12-15 | 祁小龙 | A kind of construction method of the HVPG computation model based on radiation group |
| CN109615636A (en) * | 2017-11-03 | 2019-04-12 | 杭州依图医疗技术有限公司 | Vascular tree building method, device in the lobe of the lung section segmentation of CT images |
| CN110706236A (en) * | 2019-09-03 | 2020-01-17 | 西人马帝言(北京)科技有限公司 | Three-dimensional reconstruction method and device of blood vessel image |
| CN110766713A (en) * | 2019-10-30 | 2020-02-07 | 上海微创医疗器械（集团）有限公司 | Lung image segmentation method and device and lung lesion region identification equipment |
| CN110910348A (en) * | 2019-10-22 | 2020-03-24 | 上海联影智能医疗科技有限公司 | Method, device, equipment and storage medium for classifying positions of pulmonary nodules |
| CN111242931A (en) * | 2020-01-15 | 2020-06-05 | 东北大学 | Method and device for judging small airway lesion of single lung lobe |
| CN111275673A (en) * | 2020-01-15 | 2020-06-12 | 深圳前海微众银行股份有限公司 | Lung lobe extraction method, device and storage medium |
| CN111292343A (en) * | 2020-01-15 | 2020-06-16 | 东北大学 | Lung lobe segmentation method and device based on multiple visual angles |
| CN111340773A (en) * | 2020-02-24 | 2020-06-26 | 齐鲁工业大学 | Retinal image blood vessel segmentation method |
| CN111986206A (en) * | 2019-05-24 | 2020-11-24 | 深圳市前海安测信息技术有限公司 | Lung lobe segmentation method and device based on UNet network and computer-readable storage medium |
-
2021
- 2021-02-05 CN CN202110164279.7A patent/CN112950554B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170224301A1 (en) * | 2016-02-05 | 2017-08-10 | Pulmonx Corporation | Methods, systems, and devices for analyzing lung imaging data |
| CN107230204A (en) * | 2017-05-24 | 2017-10-03 | 东北大学 | A kind of method and device that the lobe of the lung is extracted from chest CT image |
| CN107480675A (en) * | 2017-07-28 | 2017-12-15 | 祁小龙 | A kind of construction method of the HVPG computation model based on radiation group |
| CN109615636A (en) * | 2017-11-03 | 2019-04-12 | 杭州依图医疗技术有限公司 | Vascular tree building method, device in the lobe of the lung section segmentation of CT images |
| CN111986206A (en) * | 2019-05-24 | 2020-11-24 | 深圳市前海安测信息技术有限公司 | Lung lobe segmentation method and device based on UNet network and computer-readable storage medium |
| CN110706236A (en) * | 2019-09-03 | 2020-01-17 | 西人马帝言(北京)科技有限公司 | Three-dimensional reconstruction method and device of blood vessel image |
| CN110910348A (en) * | 2019-10-22 | 2020-03-24 | 上海联影智能医疗科技有限公司 | Method, device, equipment and storage medium for classifying positions of pulmonary nodules |
| CN110766713A (en) * | 2019-10-30 | 2020-02-07 | 上海微创医疗器械（集团）有限公司 | Lung image segmentation method and device and lung lesion region identification equipment |
| CN111242931A (en) * | 2020-01-15 | 2020-06-05 | 东北大学 | Method and device for judging small airway lesion of single lung lobe |
| CN111275673A (en) * | 2020-01-15 | 2020-06-12 | 深圳前海微众银行股份有限公司 | Lung lobe extraction method, device and storage medium |
| CN111292343A (en) * | 2020-01-15 | 2020-06-16 | 东北大学 | Lung lobe segmentation method and device based on multiple visual angles |
| CN111340773A (en) * | 2020-02-24 | 2020-06-26 | 齐鲁工业大学 | Retinal image blood vessel segmentation method |
Non-Patent Citations (5)
| Title |
|---|
| BIANCA LASSEN 等: "Lung and Lung Lobe Segmentation Methods at Fraunhofer MEVIS", 《LOCA11 CHALLENGE》 * |
| WENJIA WANG 等: "AUTOMATED SEGMENTATION OF PULMONARY LOBES USING COORDINATION-GUIDED DEEP NEURAL NETWORKS", 《ARXIV》 * |
| 刘莹芳 等: "一种基于CT图像的肺实质分割方法", 《云南大学学报（自然科学版）》 * |
| 袁甜 等: "基于U-Net网络的肺部CT图像分割算法", 《自动化与仪器仪表》 * |
| 边子健 等: "肺部CT图像中的解剖结构分割方法综述", 《中国图象图形学报》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114862861A (en) * | 2022-07-11 | 2022-08-05 | 珠海横琴圣澳云智科技有限公司 | Lung lobe segmentation method and device based on few-sample learning |
| CN115049735A (en) * | 2022-08-12 | 2022-09-13 | 季华实验室 | Mask optimization processing method and device, electronic equipment and storage medium |
| CN117523207A (en) * | 2024-01-04 | 2024-02-06 | 广东欧谱曼迪科技股份有限公司 | Method, device, electronic equipment and storage medium for lung lobe segmentation correction processing |
| CN117523207B (en) * | 2024-01-04 | 2024-04-09 | 广东欧谱曼迪科技股份有限公司 | Method, device, electronic equipment and storage medium for lung lobe segmentation correction processing |
| CN117576126A (en) * | 2024-01-16 | 2024-02-20 | 广东欧谱曼迪科技股份有限公司 | Optimization method and device for lung lobe segmentation, electronic equipment and storage medium |
| CN117576126B (en) * | 2024-01-16 | 2024-04-09 | 广东欧谱曼迪科技股份有限公司 | Optimization method and device for lung lobe segmentation, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112950554B (en) | 2021-12-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112950554B (en) | Lung lobe segmentation optimization method and system based on lung segmentation | |
| CN111833237B (en) | Image registration method based on convolutional neural network and local homography transformation | |
| CN110189341B (en) | Image segmentation model training method, image segmentation method and device | |
| CN111723815B (en) | Model training method, image processing device, computer system and medium | |
| CN108334879B (en) | Region extraction method, system and terminal equipment | |
| CN111382758A (en) | Training image classification model, image classification method, device, equipment and medium | |
| CN111951154B (en) | Picture generation method and device containing background and medium | |
| CN111784564B (en) | Automatic image matting method and system | |
| CN110827312A (en) | Learning method based on cooperative visual attention neural network | |
| CN111598076B (en) | Method and device for detecting and processing date in label image | |
| CN109145906B (en) | Target object image determination method, device, equipment and storage medium | |
| CN111080654A (en) | Image lesion region segmentation method and device and server | |
| CN115272691A (en) | Training method, recognition method and equipment for steel bar binding state detection model | |
| CN116137061B (en) | Training method and device for quantity statistical model, electronic equipment and storage medium | |
| CN111738310B (en) | Material classification method, device, electronic equipment and storage medium | |
| CN111798481B (en) | Image sequence segmentation method and device | |
| CN115908363A (en) | Tumor cell counting method, device, equipment and storage medium | |
| CN108876701A (en) | A kind of single sweep operation based on the distance of swimming leads to field mark method and its hardware configuration | |
| CN110827243B (en) | Method and device for detecting abnormity of coverage area of grid beam | |
| KR102108308B1 (en) | Individual cell identifying device and cotrolling method thereof | |
| CN114078245A (en) | Image processing method and image processing device | |
| CN110378930A (en) | Extracting of Moving Object, device and electronic equipment and readable storage medium storing program for executing | |
| CN111815689A (en) | Semi-automatic labeling method, equipment, medium and device | |
| CN105335935A (en) | Image processing apparatus and method | |
| CN110706190B (en) | Color enhancement method and device for pathological digital image |
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 | ||
| CP01 | Change in the name or title of a patent holder |
Address after: 100192 A206, 2 / F, building B-2, Dongsheng Science Park, Zhongguancun, 66 xixiaokou Road, Haidian District, Beijing Patentee after: Huiying medical technology (Beijing) Co.,Ltd. Address before: 100192 A206, 2 / F, building B-2, Dongsheng Science Park, Zhongguancun, 66 xixiaokou Road, Haidian District, Beijing Patentee before: HUIYING MEDICAL TECHNOLOGY (BEIJING) Co.,Ltd. |
|
| CP01 | Change in the name or title of a patent holder |