CN116310329A - Skin lesion image segmentation method based on lightweight multi-scale UNet - Google Patents
Skin lesion image segmentation method based on lightweight multi-scale UNet Download PDFInfo
- Publication number
- CN116310329A CN116310329A CN202310194557.2A CN202310194557A CN116310329A CN 116310329 A CN116310329 A CN 116310329A CN 202310194557 A CN202310194557 A CN 202310194557A CN 116310329 A CN116310329 A CN 116310329A
- Authority
- CN
- China
- Prior art keywords
- module
- depth
- convolution
- scale
- unet
- 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.)
- Pending
Links
- 206010040882 skin lesion Diseases 0.000 title claims abstract description 38
- 231100000444 skin lesion Toxicity 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000003709 image segmentation Methods 0.000 title claims abstract description 12
- 238000011176 pooling Methods 0.000 claims abstract description 16
- 230000011218 segmentation Effects 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims abstract description 12
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 230000009191 jumping Effects 0.000 claims abstract description 4
- 230000004913 activation Effects 0.000 claims description 18
- 238000010606 normalization Methods 0.000 claims description 18
- 230000004927 fusion Effects 0.000 claims description 16
- 238000010586 diagram Methods 0.000 claims description 5
- 230000002708 enhancing effect Effects 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 3
- 238000011426 transformation method Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 2
- 230000003902 lesion Effects 0.000 claims 1
- 239000011800 void material Substances 0.000 claims 1
- 230000006870 function Effects 0.000 description 15
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000003745 diagnosis Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- 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
-
- 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
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- 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
- G06T2207/30088—Skin; Dermal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/03—Recognition of patterns in medical or anatomical images
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Computational Linguistics (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a skin lesion image segmentation method based on a lightweight multi-scale UNet, which comprises the following steps: obtaining a skin lesion image and preprocessing; building a lightweight multi-scale UNet network structure, namely LMunet: based on an original UNet model, a multi-scale inversion residual error module is used for replacing an original convolution module in a UNet coding path, an asymmetric cavity space pyramid pooling module is added between the coding path and a decoding path, the number of channels of each layer is reduced, and original jumping connection of UNet is modified into channel addition; training the LMUNet network by utilizing the preprocessed skin lesion image; and inputting the skin lesion image to be segmented into a trained LMUNet network to obtain a segmentation result. Experimental results show that the method solves the problems of high complexity, large calculated amount and multiple parameters of the skin lesion image model based on UNet, improves the accuracy and the speed of dividing the skin lesion image, and realizes the rapid division of the skin lesion image.
Description
Technical Field
The invention belongs to the technical field of medical image processing, and particularly relates to a skin lesion image segmentation method based on a lightweight multi-scale UNet.
Background
Skin lesion image segmentation is an important part of skin diagnosis research. The traditional approach is to make high resolution images of damaged skin under a dermatoscope and then make a diagnosis by a specialist. However, since the size and shape of the skin lesion area are different, the manual judgment is time-consuming and labor-consuming, and contains subjective components, which increases the difficulty of diagnosis. Early computer-aided medical image segmentation methods typically relied on edge detection, template matching techniques, and traditional machine learning techniques. These methods work well to some extent, but they often require complex pre-processing of the original image, which requires an experienced engineer to design the feature extractor and select the appropriate classifier for classification. Therefore, this method is weak in generalization ability, and is difficult to implement complex multi-classification tasks.
With the advent of the big data age and the tremendous progress of computer hardware, deep learning techniques, and in particular convolutional neural networks, have achieved better results than traditional methods in many tasks such as image classification and detection. In recent years, many researchers have focused on developing a high-precision dividing method. The UNet is a semantic segmentation network based on a full convolution network, is suitable for medical image segmentation, modifies and expands the structure of the full convolution neural network, and obtains accurate segmentation results when training is performed by using a small amount of data, so that the UNet is paid attention to. After that, many neural networks are developed on the basis of UNet, such as unet++, linkNet, 3DUNet, resUNet, etc., and although the improved model improves the segmentation effect, the model is often too many in parameters and large in operation, which is unfavorable for realizing the rapid segmentation of images.
Disclosure of Invention
In order to solve the technical defects in the prior art, the invention provides a skin lesion image segmentation method based on a lightweight multi-scale UNet.
The technical scheme for realizing the purpose of the invention is as follows: a skin lesion image segmentation method based on lightweight multi-scale UNet comprises the following steps:
step 1, obtaining a skin lesion image and preprocessing;
step 2, building a lightweight multi-scale UNet network structure, namely an LMunet: based on an original UNet model, a multi-scale inversion residual error module is used for replacing an original convolution module in a UNet coding path, an asymmetric cavity space pyramid pooling module is added between the coding path and a decoding path, the number of channels of each layer is reduced, and original jumping connection of UNet is modified into channel addition;
step 3, training the LMUNet network by utilizing the preprocessed skin lesion image;
and 4, inputting the skin lesion image to be segmented into a trained LMUNet network to obtain a segmentation result.
Preferably, the preprocessing of the skin lesion image in step 1 specifically includes:
unifying the sizes of the obtained skin lesion images in a scaling or cutting mode, enhancing the data and dividing the data set; the data enhancement adopts one or more of rotation and overturn image geometric transformation methods and histogram equalization; the data set division refers to dividing the image after the data enhancement operation into a training set, a verification set and a test set.
Preferably, the method for establishing the lightweight multi-scale UNet network structure, i.e. LMUNet, in step 2 is as follows:
the LMUNet network structure comprises an encoding path, a decoding path and a multi-scale information fusion module;
the coding path consists of a multi-scale inversion residual error module and a 2 multiplied by 2 maximization pooling module, and is responsible for carrying out feature extraction on an input image, reducing the image size and reducing the redundant parameter quantity; the decoding path consists of a deconvolution layer and a standard convolution layer and is responsible for recovering the characteristic diagram information; the multi-scale information fusion module is positioned between the coding path and the decoding path and used for fusing the characteristics under different scales and enriching the context information.
Preferably, the encoding path comprises a multi-scale inversion residual module and a 2×2 maximization pooling module; the multi-scale inversion residual module is responsible for extracting characteristic information of the image under multiple scales, and the 2×2 maximization pooling module can compress the image, extract main characteristics of the image and reduce the parameter quantity of network redundancy.
The multi-scale inversion residual error module comprises a first 3X 3 depth convolution module, a second 3X 3 depth convolution module, a 3X 3 depth cavity convolution module and a 1X 1 point convolution module, wherein the first 3X 3 depth convolution module comprises a 3X 3 depth convolution module, a batch normalization layer and a ReLU6 activation function, input information passes through the 3X 3 depth convolution module, the batch normalization layer and the ReLU6 activation function once, an obtained result and the input information are subjected to channel splicing, the spliced information is respectively input into the second 3X 3 depth convolution module and the 3X 3 depth cavity convolution module, and the second 3X 3 depth convolution module comprises the 3X 3 depth convolution module, the batch normalization layer and the ReLU6 activation function; the 3×3 depth hole convolution module comprises a 3×3 depth hole convolution with a hole rate of 2, a batch normalization layer and a ReLU6 activation function;
and adding results obtained by the second 3X 3 depth convolution module and the 3X 3 depth cavity convolution module, and then inputting the results into a 1X 1 point convolution module, wherein the 1X 1 point convolution module comprises a 3X 3 depth convolution module, a batch normalization layer and a ReLU6 activation function, and adding the obtained results and input information at the beginning through the 3X 3 depth convolution module, the batch normalization layer and the ReLU6 activation function to obtain characteristic information under different scales.
Preferably, the multi-scale information fusion module is composed of an asymmetric cavity space pyramid pooling module, and comprises 1×1 point convolution and 3 parallel branches, wherein the 3 parallel branches are respectively:
branch one: consists of a 3 x 1 depth asymmetric convolution and a 1 x 3 depth asymmetric convolution;
branch two: the method consists of 3 multiplied by 1 depth asymmetrical cavity convolution with the cavity rate of 2 and 1 multiplied by 3 depth asymmetrical cavity convolution;
branch three: the method consists of 3 multiplied by 1 depth asymmetrical cavity convolution with the cavity rate of 3 and 1 multiplied by 3 depth asymmetrical cavity convolution;
the input information is subjected to 1X 1 point convolution, batch normalization layer and ReLU6 activation function, the number of channels is reduced to half of the original number, the obtained result and the information of the input branches are subjected to channel splicing through 3 branches connected in parallel, and finally, the characteristic information fusion is carried out through 1X 1 point convolution, so that the characteristic information under the multi-scale after fusion is obtained;
preferably, step 3 trains the LMUNet network using the preprocessed skin lesion image, using a cross entropy function with the following function form:
wherein H (P, Q represents cross entropy, P (x) i ) Representing the true probability distribution, Q (x i ) Representing a predictive probability distribution.
Compared with the prior art, the invention has the remarkable advantages that: the invention is based on the UNet model and improves the UNet model by introducing a multi-scale inversion residual module, an asymmetric cavity space pyramid pooling module and reducing the number of channels. Experimental results show that compared with other segmentation models, the LMUNet model provided by the invention has excellent performance and is very light, and a better segmentation effect can be realized only by a small amount of calculation.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.
Drawings
The drawings are only for purposes of illustrating particular embodiments and are not to be construed as limiting the invention, like reference numerals being used to refer to like parts throughout the several views.
Fig. 1 is an overall structure diagram of an LMUNet network in the present invention.
Fig. 2 is a block diagram of a multi-scale inversion residual module in accordance with the present invention.
Fig. 3 is a block diagram of a multi-scale information fusion module according to the present invention.
FIG. 4 is a schematic flow chart of the method of the present invention.
Detailed Description
It is easy to understand that various embodiments of the present invention can be envisioned by those of ordinary skill in the art without altering the true spirit of the present invention in light of the present teachings. Accordingly, the following detailed description and drawings are merely illustrative of the invention and are not intended to be exhaustive or to limit or restrict the invention. Rather, these embodiments are provided so that this disclosure will be thorough and complete by those skilled in the art. Preferred embodiments of the present invention are described in detail below with reference to the attached drawing figures, which form a part of the present application and are used in conjunction with embodiments of the present invention to illustrate the innovative concepts of the present invention.
The invention is conceived of as shown in fig. 1 to 4, a skin lesion image segmentation method based on a lightweight multi-scale UNet, comprising the following specific steps:
step 1, obtaining a skin lesion image and preprocessing;
step 2, building a lightweight multi-scale UNet network structure, namely an LMunet: based on an original UNet model, a multi-scale inversion residual error module is used for replacing an original convolution module in a UNet coding path, an asymmetric cavity space pyramid pooling module is added between the coding path and a decoding path, the number of channels of each layer is reduced, and original jumping connection of UNet is modified into channel addition;
step 3, training the LMUNet network by utilizing the preprocessed skin lesion image;
and 4, inputting the skin lesion image to be segmented into a trained LMUNet network to obtain a segmentation result.
In a further embodiment, the specific method for acquiring and preprocessing the skin lesion image is as follows: unifying the sizes of the obtained skin lesion images in a scaling or cutting mode, enhancing the data and dividing the data set; the data enhancement adopts one or more of geometric transformation methods such as rotation and overturn images and image enhancement methods such as histogram equalization; the data set division divides the image after the data enhancement operation into a training set, a verification set and a test set.
In a further embodiment, as shown in fig. 1, the lightweight multi-scale UNet network structure, i.e., LMUNet, includes an encoding path, a decoding path, and a multi-scale information fusion module. The coding path consists of a multi-scale inversion residual error module and a 2 multiplied by 2 maximization pooling module, wherein the maximum pooling module performs downsampling for 4 times in total; the decoding path consists of a 3 x 3 standard convolution block and a 2 x 2 deconvolution, wherein the deconvolution layer performs up-sampling 4 times in total; the multi-scale information fusion module is positioned between the coding path and the decoding path and used for fusing the characteristics under different scales and enriching the context information.
Specifically, as shown in fig. 2, the multi-scale inverted residual module structure is specifically formed by 3×3 depth convolution, 3×3 depth hole convolution, 1×1 point convolution, batch normalization layer and ReLU6 activation function, input information of the module is firstly subjected to 3×3 depth convolution, batch normalization layer and ReLU6 activation function, an obtained result and the input information are subjected to channel splicing, and then 2 parallel branches are passed:
branch one: consists of a 3 x 3 depth convolution;
branch two: consists of 3 x 3 depth cavity convolution with a cavity rate of 2;
the results from 2 branches are added and then subjected to 3 x 3 depth convolution, batch normalization layer and ReLU6 activation function, and finally the obtained results are channel added with the input information at the beginning. The module can be used for obtaining the characteristic information under different scales.
Further, the structure of the multi-scale information fusion module is shown in fig. 3, specifically, the multi-scale information fusion module is composed of an asymmetric cavity space pyramid pooling module, and includes 3 parallel branches:
branch one: consists of a 3 x 1 depth asymmetric convolution and a 1 x 3 depth asymmetric convolution;
branch two: the method consists of 3 multiplied by 1 depth asymmetrical cavity convolution with the cavity rate of 2 and 1 multiplied by 3 depth asymmetrical cavity convolution;
branch three: the method consists of 3 multiplied by 1 depth asymmetrical cavity convolution with the cavity rate of 3 and 1 multiplied by 3 depth asymmetrical cavity convolution;
the input information firstly passes through a 1X 1 point convolution, a batch normalization layer and a ReLU6 activation function, the number of channels is reduced to half of the original number, then 3 parallel branches are passed, the channel splicing is carried out on the results obtained by the 3 branches and the information of the input branches, and finally the characteristic information fusion is carried out through the 1X 1 point convolution. The module can be used for fusing the characteristic information under multiple scales and enhancing the model capacity.
In the training process, the network model is realized on a Pytorch platform, and the version is Pytorch1.11.0. The super parameters in the training process are as follows: the learning rate was set to 0.01, the number of training iterations was set to 100, and the number of samples per training was 4. The optimizer uses a random gradient descent method to update network parameters, so that the convergence speed is increased. The loss function used is a cross entropy function, the functional form being:
wherein H (P, Q represents cross entropy, P (x) i ) Representing the true probability distribution, Q (x i ) Representing a predictive probability distribution.
TABLE 1 comparison of the segmentation results of the inventive method and the partial convolutional neural network method
As can be seen from the segmentation results in Table 1, compared with other convolutional neural networks in Table 1, the method has the advantages of excellent performance, light weight, reduced parameter amount and calculation amount, and improved accuracy and speed of segmentation of skin lesion images.
The above description is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto,
any changes or substitutions that would be easily recognized by those skilled in the art within the technical scope of the present disclosure are intended to be covered by the present invention.
It should be appreciated that in the above description of exemplary embodiments of the invention, various features of the invention are sometimes described in the context of a single embodiment or with reference to a single figure in order to streamline the invention and aid those skilled in the art in understanding the various aspects of the invention. The present invention should not, however, be construed as including features that are essential to the patent claims in the exemplary embodiments.
It should be understood that modules, units, components, etc. included in the apparatus of one embodiment of the present invention may be adaptively changed to arrange them in an apparatus different from the embodiment. The different modules, units or components comprised by the apparatus of the embodiments may be combined into one module, unit or component or they may be divided into a plurality of sub-modules, sub-units or sub-components.
Claims (6)
1. The skin lesion image segmentation method based on the lightweight multi-scale UNet is characterized by comprising the following steps of:
step 1, obtaining a skin lesion image and preprocessing;
step 2, building a lightweight multi-scale UNet network structure, namely an LMunet: based on an original UNet model, a multi-scale inversion residual error module is used for replacing an original convolution module in a UNet coding path, an asymmetric cavity space pyramid pooling module is added between the coding path and a decoding path, the number of channels of each layer is reduced, and original jumping connection of UNet is modified into channel addition;
step 3, training the LMUNet network by utilizing the preprocessed skin lesion image;
and 4, inputting the skin lesion image to be segmented into a trained LMUNet network to obtain a segmentation result.
2. The method for segmenting a skin lesion image based on lightweight multi-scale UNet according to claim 1, wherein the preprocessing of the skin lesion image in step 1 specifically comprises:
unifying the sizes of the obtained skin lesion images in a scaling or cutting mode, enhancing the data and dividing the data set; the data enhancement adopts one or more of rotation and overturn image geometric transformation methods and histogram equalization; the data set division refers to dividing the image after the data enhancement operation into a training set, a verification set and a test set.
3. The method for segmenting skin lesions based on lightweight multi-scale UNet according to claim 1, wherein the step 2 is to build a lightweight multi-scale UNet network structure, i.e. LMUNet, specifically comprising:
the LMUNet network structure comprises an encoding path, a decoding path and a multi-scale information fusion module;
the coding path consists of a multi-scale inversion residual error module and a 2 multiplied by 2 maximization pooling module, and is responsible for carrying out feature extraction on an input image, reducing the image size and reducing the redundant parameter quantity; the decoding path consists of a deconvolution layer and a standard convolution layer and is responsible for recovering the characteristic diagram information; the multi-scale information fusion module is positioned between the coding path and the decoding path and used for fusing the characteristics under different scales and enriching the context information.
4. A lightweight multi-scale UNet-based dermatological lesion image segmentation method according to claim 3, characterized in that the encoding path comprises a multi-scale inversion residual module and a 2 x 2 maximization pooling module; the multi-scale inversion residual module is responsible for extracting characteristic information of the image under multiple scales, and the 2×2 maximization pooling module can compress the image, extract main characteristics of the image and reduce the parameter quantity of network redundancy.
The multi-scale inversion residual error module comprises a first 3X 3 depth convolution module, a second 3X 3 depth convolution module, a 3X 3 depth cavity convolution module and a 1X 1 point convolution module, wherein the first 3X 3 depth convolution module comprises a 3X 3 depth convolution module, a batch normalization layer and a ReLU6 activation function, input information passes through the 3X 3 depth convolution module, the batch normalization layer and the ReLU6 activation function once, an obtained result and the input information are subjected to channel splicing, the spliced information is respectively input into the second 3X 3 depth convolution module and the 3X 3 depth cavity convolution module, and the second 3X 3 depth convolution module comprises the 3X 3 depth convolution module, the batch normalization layer and the ReLU6 activation function; the 3×3 depth hole convolution module comprises a 3×3 depth hole convolution with a hole rate of 2, a batch normalization layer and a ReLU6 activation function;
and adding results obtained by the second 3X 3 depth convolution module and the 3X 3 depth cavity convolution module, and then inputting the results into a 1X 1 point convolution module, wherein the 1X 1 point convolution module comprises a 3X 3 depth convolution module, a batch normalization layer and a ReLU6 activation function, and adding the obtained results and input information at the beginning through the 3X 3 depth convolution module, the batch normalization layer and the ReLU6 activation function to obtain characteristic information under different scales.
5. The method for segmenting the skin lesion image based on the lightweight multi-scale UNet according to claim 1, wherein the multi-scale information fusion module consists of an asymmetric cavity space pyramid pooling module, and comprises 1 x 1 point convolution and 3 parallel branches, wherein the 3 parallel branches are respectively:
branch one: consists of a 3 x 1 depth asymmetric convolution and a 1 x 3 depth asymmetric convolution;
branch two: the method consists of 3 multiplied by 1 depth asymmetrical cavity convolution with the cavity rate of 2 and 1 multiplied by 3 depth asymmetrical cavity convolution;
branch three: consists of a 3 x 1 depth asymmetric hole convolution with a 3 void fraction and a 1 x 3 depth asymmetric hole convolution.
The input information is subjected to 1X 1 point convolution, batch normalization layer and ReLU6 activation function, the number of channels is reduced to half of the original number, the obtained result and the information of the input branches are subjected to channel splicing through 3 branches connected in parallel, and finally, the characteristic information fusion is carried out through 1X 1 point convolution, so that the characteristic information under the multi-scale after fusion is obtained;
6. the method for segmenting skin lesion images based on lightweight multi-scale UNet according to claim 1, wherein step 3 trains the LMUNet network using preprocessed skin lesion images, and the loss function is a cross entropy function, and the function is as follows:
wherein H (P, Q) represents cross entropy, P (x) i ) Representing the true probability distribution, Q (x i ) Representing a predictive probability distribution, x i And n is the number of variables.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310194557.2A CN116310329A (en) | 2023-03-03 | 2023-03-03 | Skin lesion image segmentation method based on lightweight multi-scale UNet |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310194557.2A CN116310329A (en) | 2023-03-03 | 2023-03-03 | Skin lesion image segmentation method based on lightweight multi-scale UNet |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116310329A true CN116310329A (en) | 2023-06-23 |
Family
ID=86789919
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310194557.2A Pending CN116310329A (en) | 2023-03-03 | 2023-03-03 | Skin lesion image segmentation method based on lightweight multi-scale UNet |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116310329A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116612122A (en) * | 2023-07-20 | 2023-08-18 | 湖南快乐阳光互动娱乐传媒有限公司 | Image significance region detection method and device, storage medium and electronic equipment |
| CN117351202A (en) * | 2023-09-28 | 2024-01-05 | 河北翔拓航空科技有限公司 | Image segmentation method for lightweight multi-scale UNet network |
-
2023
- 2023-03-03 CN CN202310194557.2A patent/CN116310329A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116612122A (en) * | 2023-07-20 | 2023-08-18 | 湖南快乐阳光互动娱乐传媒有限公司 | Image significance region detection method and device, storage medium and electronic equipment |
| CN116612122B (en) * | 2023-07-20 | 2023-10-10 | 湖南快乐阳光互动娱乐传媒有限公司 | Image significance region detection method and device, storage medium and electronic equipment |
| CN117351202A (en) * | 2023-09-28 | 2024-01-05 | 河北翔拓航空科技有限公司 | Image segmentation method for lightweight multi-scale UNet network |
| CN117351202B (en) * | 2023-09-28 | 2024-04-30 | 河北翔拓航空科技有限公司 | Image segmentation method for lightweight multi-scale UNet network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6980958B1 (en) | Rural area classification garbage identification method based on deep learning | |
| CN111860171B (en) | Method and system for detecting irregular-shaped target in large-scale remote sensing image | |
| CN116310329A (en) | Skin lesion image segmentation method based on lightweight multi-scale UNet | |
| CN109410219A (en) | A kind of image partition method, device and computer readable storage medium based on pyramid fusion study | |
| CN113807355A (en) | Image semantic segmentation method based on coding and decoding structure | |
| CN110992351B (en) | sMRI image classification method and device based on multi-input convolution neural network | |
| CN112070158B (en) | Facial flaw detection method based on convolutional neural network and bilateral filtering | |
| CN114038037B (en) | Expression label correction and identification method based on separable residual error attention network | |
| WO2024104035A1 (en) | Long short-term memory self-attention model-based three-dimensional medical image segmentation method and system | |
| CN106408001A (en) | Rapid area-of-interest detection method based on depth kernelized hashing | |
| CN111369565A (en) | Digital pathological image segmentation and classification method based on graph convolution network | |
| CN112348830B (en) | Multi-organ segmentation method based on improved 3D U-Net | |
| CN113344933B (en) | Glandular cell segmentation method based on multi-level feature fusion network | |
| CN113192076B (en) | MRI brain tumor image segmentation method combining classification prediction and multi-scale feature extraction | |
| CN114972254A (en) | Cervical cell image segmentation method based on convolutional neural network | |
| CN115409846A (en) | Colorectal cancer focus region lightweight segmentation method based on deep learning | |
| CN110349170B (en) | Full-connection CRF cascade FCN and K mean brain tumor segmentation algorithm | |
| CN110738660A (en) | Spine CT image segmentation method and device based on improved U-net | |
| CN115526829A (en) | Honeycomb lung focus segmentation method and network based on ViT and context feature fusion | |
| Li et al. | A deep learning method for material performance recognition in laser additive manufacturing | |
| CN112927237A (en) | Honeycomb lung focus segmentation method based on improved SCB-Unet network | |
| CN110633706B (en) | Semantic segmentation method based on pyramid network | |
| CN116363149A (en) | Medical image segmentation method based on U-Net improvement | |
| CN111667488B (en) | Medical image segmentation method based on multi-angle U-Net | |
| CN116823868A (en) | Melanin tumor image segmentation method |
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 |