CN108806774B - Medical image retrieval method based on geometric constraint and spatial pixel intensity - Google Patents
Medical image retrieval method based on geometric constraint and spatial pixel intensity Download PDFInfo
- Publication number
- CN108806774B CN108806774B CN201810496727.1A CN201810496727A CN108806774B CN 108806774 B CN108806774 B CN 108806774B CN 201810496727 A CN201810496727 A CN 201810496727A CN 108806774 B CN108806774 B CN 108806774B
- Authority
- CN
- China
- Prior art keywords
- image
- pixel intensity
- vlad
- codebook
- database
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
Abstract
The invention discloses a medical image retrieval method based on geometric constraint and spatial pixel intensity, which relates to the technical field of medical image processing. And then extracting spatial pixel intensity characteristics of the ROI on the premise of geometric constraint, namely extracting the pixel intensity characteristics of each pixel point and extracting the pixel intensity characteristics of surrounding pixels which have a certain geometric relationship with the pixel point, and describing pixel point information together. And performing K-means clustering on the extracted spatial pixel intensity characteristics to generate a VLAD model. And finally, the similarity between the images is judged according to the distance between the VLAD vectors of the images for retrieval, so that the retrieval performance of the brain tumor image retrieval can be effectively improved.
Description
Technical Field
The invention relates to the technical field of medical image processing, in particular to a medical image retrieval method based on geometric constraint and spatial pixel intensity.
Background
Brain tumors, also called intracranial tumors, can be classified into glioma, meningioma, pituitary tumor and the like according to different diseased tissue sites, and are also diseases seriously threatening human health. The death rate of brain tumor is very high, and the incidence of brain tumor is continuously improved along with the acceleration of life rhythm and the continuous increase of working pressure.
The medical image retrieval based on the content also plays a good auxiliary role in diagnosis and treatment of brain tumors, and a plurality of researchers are dedicated to the research of brain tumor image retrieval. Researchers such as Quellec and the like use a wavelet transform algorithm in medical image retrieval, and researchers such as Yangxian and the like perform experimental comparison on retrieval performance of five image feature extraction methods such as a color correlation diagram, a color moment, a gray level co-occurrence matrix, pyramid wavelet transform and tree wavelet transform in brain tumor image retrieval, and the result shows that the effect of wavelet transform is better. Researchers such as Huang H K and the like propose a method for determining a brain tumor image ROI, extracting texture, shape and gray scale features of the image in the ROI and carrying out similarity measurement, wherein the method uses global features of the brain tumor image. On the basis, researchers such as Huang M and the like propose a brain tumor image retrieval method based on a BoW model aiming at the specificity of a brain tumor region, and similarity measurement is carried out on a query image and a database image in a mode of establishing a visual word bag by dividing an obtained ROI into regions and extracting the pixel intensity characteristics of the image on each sub-region.
However, valuable key points which can be extracted from a human brain tumor image area are few, and gradient information used by the SIFT features in the brain tumor image is not rich in pixel intensity information in the brain tumor image, so that the discrimination capability is poor when the SIFT features are used for expressing brain tumor image information.
Disclosure of Invention
The embodiment of the invention provides a medical image retrieval method based on geometric constraint and spatial pixel intensity, which can solve the problems in the prior art.
The invention provides a medical image retrieval method based on geometric constraint and spatial pixel intensity, which comprises the following steps:
step 1, reading a database image;
step 2, according to the set image expansion structure parameter p, creating a structural element to perform expansion processing on the read database image;
step 3, extracting the spatial pixel intensity characteristics of the database image subjected to expansion processing according to the set geometric constraint parameter n, and expressing the spatial pixel intensity characteristics by using a d-dimensional vector;
step 4, performing K-means clustering on the extracted spatial pixel intensity characteristics according to the set number K of clustering centers to generate K clustering centers;
step 5, generating a VLAD codebook for each database image: dividing all spatial pixel intensity characteristics in the database image into k classes according to the k clustering centers generated in the step 4, accumulating difference values after the characteristics belonging to the same clustering center are differed with the clustering center to obtain d-dimensional residual vectors, and integrating the k d-dimensional residual vectors into a k x d-dimensional matrix, namely a VLAD codebook;
step 6, inputting a query image;
step 7, according to the parameter p, creating a structural element to perform expansion processing on the query image;
step 8, extracting the spatial pixel intensity characteristic of the query image subjected to expansion processing according to the geometric constraint parameter n, and representing the spatial pixel intensity characteristic by using a d-dimensional vector;
step 9, generating a VLAD codebook for the query image: dividing all spatial pixel intensity characteristics in the query image into k classes according to the k clustering centers generated in the step 4, accumulating difference values after the characteristics belonging to the same clustering center are differed with the clustering center to obtain d-dimensional residual vectors, and integrating the k d-dimensional residual vectors into a k x d-dimensional matrix, namely a VLAD codebook;
and step 10, performing similarity measurement on the query image and the database image according to the VLAD codebook generated in the steps 5 and 9, and outputting a retrieval result.
According to the medical image retrieval method based on the geometric constraint and the spatial pixel intensity, a local feature aggregation descriptor VLAD model is introduced to quantify a brain tumor image. Firstly, expansion enhancement is carried out on the ROI of the brain tumor image, and the image information of the edge neighborhood of the expanded ROI is fused, so that the ROI of the brain tumor image can be better described. And then extracting spatial pixel intensity characteristics of the ROI on the premise of geometric constraint, namely extracting the pixel intensity characteristics of each pixel point and extracting the pixel intensity characteristics of surrounding pixels which have a certain geometric relationship with the pixel point, and describing pixel point information together. And performing K-means clustering on the extracted spatial pixel intensity characteristics to generate a VLAD model. And finally, the similarity between the images is judged according to the distance between the VLAD vectors of the images for retrieval, so that the retrieval performance of the brain tumor image retrieval can be effectively improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart of a medical image retrieval method based on geometric constraints and spatial pixel intensities in an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, an embodiment of the present invention provides a medical image retrieval method based on geometric constraint and spatial pixel intensity, the method includes the following steps:
step 1, reading a database image.
And 2, creating a structural element according to the set image expansion structure parameter p to perform expansion processing on the read image, wherein the structural element used in the embodiment is a disc.
And 3, extracting the spatial pixel intensity characteristic of the image subjected to expansion processing according to the set geometric constraint parameter n, and representing the spatial pixel intensity characteristic by using a d-dimensional vector.
And 4, performing K-means clustering on the extracted spatial pixel intensity characteristics according to the set number K of clustering centers to generate K clustering centers.
Step 5, generating a VLAD codebook for each image in the database: dividing all spatial pixel intensity characteristics in the image into k classes according to the k clustering centers generated in the step 4, accumulating difference values after the characteristics belonging to the same clustering center are differed with the clustering center to obtain d-dimensional residual vectors, and integrating the k d-dimensional residual vectors into a k x d-dimensional matrix, namely the VLAD codebook.
And 6, inputting a query image.
And 7, according to the parameter p, creating a structural element to perform expansion processing on the input query image.
And 8, extracting the spatial pixel intensity characteristic of the query image subjected to expansion processing according to the geometric constraint parameter n, and representing the spatial pixel intensity characteristic by using a d-dimensional vector.
Step 9, generating a VLAD codebook for the query image: dividing all spatial pixel intensity characteristics in the query image into k classes according to the k clustering centers generated in the step 4, accumulating difference values after the characteristics belonging to the same clustering center are differed with the clustering center to obtain d-dimensional residual vectors, and integrating the k d-dimensional residual vectors into a k x d-dimensional matrix, namely the VLAD codebook.
And step 10, performing similarity measurement on the query image and the database image according to the VLAD codebook generated in the steps 5 and 9, and outputting a retrieval result.
The method for the expansion treatment in the steps 2 and 7 comprises the following steps: and taking the central point of the structural element as an anchor point, positioning the anchor point of the structural element at the first pixel position of the image, and starting to scan the image. And replacing the intensity value of the anchor point position pixel in the image with the maximum intensity value of the pixel in the image area covered by the structural element, and translating the structural element on the image to process the next pixel point of the image until the structural element scans the whole image.
The method for extracting the spatial pixel intensity characteristics in the steps 3 and 8 comprises the following steps: and forming a space pixel intensity characteristic vector by using the pixel intensity values in the neighborhood and the central pixel intensity value to jointly describe the central pixel.
The method for generating the VLAD codebook in steps 5 and 9 is:
substeps 101, countingThe spatial pixel intensity characteristics of the images in the database are represented by a vector x, the vector dimension is d, and the ith clustering center of k clustering centers is represented by ciRepresenting that the cluster center set is represented by Cl;
substep 102, the extracted features of the query image and each database image are classified into k classes, each database image belonging to a clustering center ciIs represented as:
substep 103, making difference between all feature points in each feature subset and the cluster center of the feature subset, and then accumulating all difference values to generate d-dimensional residual vector viExpressed as:
substep 104, integrating the residual vectors calculated for all the cluster centers into a k × d vector V, where the expression is:
a substep 105 of performing PCA dimension reduction processing on the vector V;
a substep 106 of performing L on the vector V subjected to the dimension reduction processing2Normalization treatment:
through the steps, the VLAD codebook is obtained.
The step 10 specifically comprises the following steps: VLAD codebook V of query imageqVLAD codebook V associated with database imagesdCalculating the Euclidean distance:
dist(Vq,Vd)=||Vq-Vd||2 (5)
and performing ascending sequencing according to the Euclidean distances between the obtained query image and all database images, and selecting the first five images with the shortest distance to the query image as retrieval result images to be returned.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (2)
1. A method for medical image retrieval based on geometric constraints and spatial pixel intensities, the method comprising the steps of:
step 1, reading a database image;
step 2, according to the set image expansion structure parameter p, creating a structural element to perform expansion processing on the read database image;
step 3, extracting the spatial pixel intensity characteristics of the database image subjected to expansion processing according to the set geometric constraint parameter n, and expressing the spatial pixel intensity characteristics by using a d-dimensional vector;
step 4, performing K-means clustering on the extracted spatial pixel intensity characteristics according to the set number K of clustering centers to generate K clustering centers;
step 5, generating a VLAD codebook for each database image: dividing all spatial pixel intensity characteristics in the database image into k classes according to the k clustering centers generated in the step 4, accumulating difference values after making differences between all characteristics belonging to the same clustering center and the clustering center to obtain d-dimensional residual vectors, and integrating the k d-dimensional residual vectors into a k x d-dimensional matrix, namely a VLAD codebook;
step 6, inputting a query image;
step 7, according to the parameter p, creating a structural element to perform expansion processing on the query image;
step 8, extracting the spatial pixel intensity characteristic of the query image subjected to expansion processing according to the geometric constraint parameter n, and representing the spatial pixel intensity characteristic by using a d-dimensional vector;
step 9, generating a VLAD codebook for the query image: dividing all spatial pixel intensity characteristics in the query image into k classes according to the k clustering centers generated in the step 4, accumulating difference values after making differences between all characteristics belonging to the same clustering center and the clustering center to obtain d-dimensional residual vectors, and integrating the k d-dimensional residual vectors into a k x d-dimensional matrix, namely a VLAD codebook;
step 10, performing similarity measurement on the query image and the database image according to the VLAD codebook generated in the steps 5 and 9, and outputting a retrieval result;
the method of the expansion treatment in the steps 2 and 7 comprises the following steps: taking the central point of the structural element as an anchor point, positioning the anchor point of the structural element at the first pixel position of the image to start scanning the image, replacing the intensity value of the anchor point position pixel in the image with the maximum intensity value of the pixel in the image area covered by the structural element, and horizontally moving the structural element on the image to process the next pixel point of the image until the structural element scans the whole image;
the method for extracting the spatial pixel intensity characteristics in the steps 3 and 8 comprises the following steps: forming a space pixel intensity characteristic vector by using pixel intensity values in the neighborhood and a center pixel intensity value to jointly describe a center pixel;
the method for generating the VLAD codebook in steps 5 and 9 is:
in the substep 101, the spatial pixel intensity feature of the image in the database is represented by a vector x, the vector dimension is d, and the ith clustering center of the k clustering centers is represented by ciRepresenting that the cluster center set is represented by Cl;
substep 102, the extracted features of the query image and each database image are classified into k classes, each database image belonging to a clustering center ciIs represented as:
substep 103, making difference between all feature points in each feature subset and the cluster center of the feature subset, and then accumulating all difference values to generate d-dimensional residual vector viExpressed as:
substep 104, integrating the residual vectors calculated for all the cluster centers into a k × d vector V, where the expression is:
a substep 105 of performing PCA dimension reduction processing on the vector V;
a substep 106 of performing L on the vector V subjected to the dimension reduction processing2Normalization treatment:
through the steps, the VLAD codebook is obtained.
2. The medical image retrieval method based on geometric constraints and spatial pixel intensities according to claim 1, wherein step 10 specifically comprises the following method: VLAD codebook V of query imageqVLAD codebook V associated with database imagesdCalculating the Euclidean distance:
dist(Vq,Vd)=||Vq-Vd||2 (5)
and performing ascending sequencing according to the Euclidean distances between the obtained query image and all database images, and selecting the first five images with the shortest distance to the query image as retrieval result images to be returned.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810496727.1A CN108806774B (en) | 2018-05-22 | 2018-05-22 | Medical image retrieval method based on geometric constraint and spatial pixel intensity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810496727.1A CN108806774B (en) | 2018-05-22 | 2018-05-22 | Medical image retrieval method based on geometric constraint and spatial pixel intensity |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108806774A CN108806774A (en) | 2018-11-13 |
CN108806774B true CN108806774B (en) | 2022-02-01 |
Family
ID=64092820
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810496727.1A Active CN108806774B (en) | 2018-05-22 | 2018-05-22 | Medical image retrieval method based on geometric constraint and spatial pixel intensity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108806774B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104216949A (en) * | 2014-08-13 | 2014-12-17 | 中国科学院计算技术研究所 | Method and system for expressing clustering of image features by fusion of space information |
CN104951562A (en) * | 2015-07-08 | 2015-09-30 | 天云融创数据科技(北京)有限公司 | Image retrieval method based on VLAD (vector of locally aggregated descriptors) dual self-adaptation |
CN106204545A (en) * | 2016-06-28 | 2016-12-07 | 南方医科大学 | A kind of based on region division and the medical science lesion image feature representation method of Fisher vector |
CN107527058A (en) * | 2017-07-25 | 2017-12-29 | 北京理工大学 | A kind of image search method based on weighting local feature Aggregation Descriptor |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2016037844A1 (en) * | 2014-09-09 | 2016-03-17 | Thomson Licensing | Method and apparatus for image retrieval with feature learning |
-
2018
- 2018-05-22 CN CN201810496727.1A patent/CN108806774B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104216949A (en) * | 2014-08-13 | 2014-12-17 | 中国科学院计算技术研究所 | Method and system for expressing clustering of image features by fusion of space information |
CN104951562A (en) * | 2015-07-08 | 2015-09-30 | 天云融创数据科技(北京)有限公司 | Image retrieval method based on VLAD (vector of locally aggregated descriptors) dual self-adaptation |
CN106204545A (en) * | 2016-06-28 | 2016-12-07 | 南方医科大学 | A kind of based on region division and the medical science lesion image feature representation method of Fisher vector |
CN107527058A (en) * | 2017-07-25 | 2017-12-29 | 北京理工大学 | A kind of image search method based on weighting local feature Aggregation Descriptor |
Non-Patent Citations (1)
Title |
---|
图像检索中判别性增强研究;李清亮;《中国博士学位论文全文数据库 信息科技辑》;20160815;I138-58 * |
Also Published As
Publication number | Publication date |
---|---|
CN108806774A (en) | 2018-11-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10019656B2 (en) | Diagnostic system and method for biological tissue analysis | |
JP7417555B2 (en) | Human body pose analysis system and method | |
Bai et al. | Liver tumor segmentation based on multi-scale candidate generation and fractal residual network | |
Xu et al. | Hierarchical segmentation using tree-based shape spaces | |
CN108921019B (en) | Gait recognition method based on GEI and TripletLoss-DenseNet | |
US9569698B2 (en) | Method of classifying a multimodal object | |
McLean | Vector quantization for texture classification | |
CN107273824B (en) | Face recognition method based on multi-scale and multi-direction local binary pattern | |
Hu et al. | Automatic pancreas segmentation in CT images with distance-based saliency-aware DenseASPP network | |
O’Connor et al. | Facial recognition using modified local binary pattern and random forest | |
He et al. | Automatic magnetic resonance image prostate segmentation based on adaptive feature learning probability boosting tree initialization and CNN-ASM refinement | |
CN113989890A (en) | Face expression recognition method based on multi-channel fusion and lightweight neural network | |
Zhang et al. | Discovering the best feature extraction and selection algorithms for spontaneous facial expression recognition | |
CN112861605A (en) | Multi-person gait recognition method based on space-time mixed characteristics | |
Sivanesan et al. | Unsupervised medical image segmentation with adversarial networks: From edge diagrams to segmentation maps | |
Shahroudnejad et al. | TUN-Det: a novel network for thyroid ultrasound nodule detection | |
Aman et al. | Content-based image retrieval on CT colonography using rotation and scale invariant features and bag-of-words model | |
Suresh et al. | Optimization and Deep Learning–Based Content Retrieval, Indexing, and Metric Learning Approach for Medical Images | |
Juang et al. | Stereo-camera-based object detection using fuzzy color histograms and a fuzzy classifier with depth and shape estimations | |
CN110992310A (en) | Method and device for determining partition where mediastinal lymph node is located | |
Padma et al. | Image segmentation using Mask R-CNN for tumor detection from medical images | |
Pavlov et al. | Weakly supervised fine tuning approach for brain tumor segmentation problem | |
Tang et al. | Tumor segmentation from single contrast MR images of human brain | |
Kamencay et al. | 3D image reconstruction from 2D CT slices | |
CN108806774B (en) | Medical image retrieval method based on geometric constraint and spatial pixel intensity |
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 |