CN108564567A - A kind of ultrahigh resolution pathological image cancerous region method for visualizing - Google Patents

A kind of ultrahigh resolution pathological image cancerous region method for visualizing Download PDF

Info

Publication number
CN108564567A
CN108564567A CN201810215800.3A CN201810215800A CN108564567A CN 108564567 A CN108564567 A CN 108564567A CN 201810215800 A CN201810215800 A CN 201810215800A CN 108564567 A CN108564567 A CN 108564567A
Authority
CN
China
Prior art keywords
probability
image
slice
ultrahigh resolution
pathological image
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
Application number
CN201810215800.3A
Other languages
Chinese (zh)
Inventor
王国利
刘力
李亮
徐琰
郭斌
郭雪梅
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sun Yat Sen University
Original Assignee
Sun Yat Sen University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Sun Yat Sen University filed Critical Sun Yat Sen University
Priority to CN201810215800.3A priority Critical patent/CN108564567A/en
Publication of CN108564567A publication Critical patent/CN108564567A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • G06T7/0012Biomedical image inspection
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/11Region-based segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/136Segmentation; Edge detection involving thresholding
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/194Segmentation; Edge detection involving foreground-background segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10024Color image
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/10Image acquisition modality
    • G06T2207/10056Microscopic image
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20016Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20081Training; Learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20084Artificial neural networks [ANN]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30004Biomedical image processing
    • G06T2207/30024Cell structures in vitro; Tissue sections in vitro
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30004Biomedical image processing
    • G06T2207/30096Tumor; Lesion

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)

Abstract

The present invention provides a kind of ultrahigh resolution pathological image cancerous region method for visualizing, first, the tissue regions of ultrahigh resolution pathological image to be measured are identified using threshold value optimal algorithm, and have overlappingly extraction slice from tissue regions, achieve the purpose that rejecting extraneous background reduces calculation amount and avoid introducing noise.Secondly, the deep neural network model for the slice of extraction being inputted persistence is predicted, will predict that the probability of every obtained slice constitutes probability matrix Mp.Then, it is based on probability matrix generating probability probe image H, trains a disaggregated model to carry out full figure prediction from H extraction features.Finally, it is based on full figure prediction result, probability probe image H is visualized as the expression of cancerous region probability and cancerous region profile is expressed.Cancerous region method for visualizing provided by the invention will greatly mitigate the work load of pathologist, improve accuracy rate of diagnosis for assisting pathologist to carry out analysis and diagnosis to pathological section.

Description

A kind of ultrahigh resolution pathological image cancerous region method for visualizing
Technical field
The present invention relates to ultrahigh resolution pathological image cancerous region method for visualizing fields, more particularly, to one kind Ultrahigh resolution pathological image cancerous region method for visualizing.
Background technology
Pathological basic medicine task is to obtain medical diagnosis on disease as a result, to instruct patient by studying pathological material Treatment.Due to the high characteristic of the resolution ratio of the whole slide image of complexity and pathology number of pathology itself, disease Reason doctor is taken time and effort to the analysis work of pathological image and accuracy rate is low.Existing Applied Computer Techniques assist pathological image The related work achievement of medical diagnosis mostly concentrates on the non-super high-resolution medical image analysis of small size, and faces superelevation The diagnosis of resolution ratio pathological image and analysis task, often computation complexity is high, and diagnostic result can not be expressed intuitively, fails from this Mitigate pathologist work load in matter.The present invention is using segmentation thought, greatly using the slice extracting method based on piecemeal Reduce computation complexity;Probability probe image is built using the mapping method of slice probability to probability matrix, and is based on probability Probe image carries out full figure prediction using disaggregated model, improves canceration judging nicety rate;Further, the visualization side of use Method obtains the expression of cancerous region probability and the positioning of cancer area is realized in the expression of cancerous region profile, and cancerous issue is enable intuitively to show, The diagnosis efficiency for greatly improving pathologist alleviates the work load of pathologist.
Invention content
The present invention provides a kind of ultrahigh resolution pathological image cancerous region method for visualizing, and this method can be realized to superelevation Resolution ratio pathological image carries out the positioning of cancer area and visualization.
In order to reach above-mentioned technique effect, technical scheme is as follows:
A kind of ultrahigh resolution pathological image cancerous region method for visualizing, includes the following steps:
S1:The slice extraction of interest region:Described image is divided by group using the threshold value optimal algorithm based on pixel RGB values Tissue region and extraneous background region are split tissue regions and extract slice;
S2:Model prediction and probability matrix structure:To biography before the small slice input convolutional neural networks model is carried out The canceration probability for obtaining every small slice is broadcast, gained probability, which is pressed default update method, builds probability matrix;
S3:Full figure is predicted:Full figure prediction is carried out to ultrahigh resolution pathological image based on probability probe image, is divided into sun Property and negative two classes;
S4:Prediction result visualizes:Based on full figure prediction result, probability probe image is mapped to obtain cancerous region probability Expression and the expression of cancerous region profile.
Further, the detailed process of the step S1 is:
Optimal threshold is obtained using the threshold value optimal algorithm of pixel RGB values, based on optimal threshold to ultrahigh resolution pathology Image is detected, and is identified tissue regions and is drawn tissue regions profile boundary rectangle;It is starting with the upper left corner of boundary rectangle Coordinate has the slice that overlappingly interception resolution ratio is 256*256 to preserve to input rank.
Further, in the step S2, if temperature figure resolution ratio corresponds on sample medical image pyramid Level-K, structure Level-K correspond to the complete zero initial matrix Mp of two dimension of dimension, will slice input deep neural network model into Row propagated forward obtains corresponding prediction probability, and the centre coordinate which is mapped to Level-K with corresponding slice is to index more The respective element of new matrix Mp.
Further, in the step S3, a disaggregated model is trained from training set probability probe image extraction feature, from The feature input disaggregated model for the probability probe image extraction same type that test set generates is predicted, by ultrahigh resolution disease Reason image is divided into positive and negative two classes.
Further, in the step S4, probability probe image H is subjected to transparency process with threshold value A lpha, then with it is right The pathological image of the Level-K answered is overlapped processing and generates new images, obtains the expression of cancerous region probability;By probability probe figure Contour detecting is carried out after carrying out binaryzation and corrosion and filling as H, profile coordinate information is mapped to corresponding Level-K's The expression of cancerous region profile is obtained after pathological image.
Further, the slice for having overlapping is intercepted and is preserved, including sliding window interception and tissue regions judge:It utilizes The resolution information of sample to be tested image, sliding window, which adaptively has with corresponding step-length, overlappingly intercepts test sample;To being intercepted Optimal pixel RGB threshold value of the slice based on gained, judge biopsy tissues region accounting whether be more than 30%, if so, should Slice, which preserves and scans next slice, to be judged;If it is not, then directly scanning next slice repeats above-mentioned judgment step.
Further, the identification tissue regions and tissue regions profile boundary rectangle is drawn, process of realizing is based on point Block method:If ultrahigh resolution pathological image is equably divided into stem portion, tissue area is identified to these parts of images respectively Domain simultaneously draws tissue regions profile boundary rectangle.
Further, the probability probe image generated from test set, is generated based on probability matrix, with probability matrix Each element in Mp is that two-dimensional matrix Mp is extended for three-dimensional matrice Mp1 by expansion factor, is based on Mp1 generating probability probe images H, wherein color are deeper, show that the region suffers from that cancer probability is higher, and color is more shallow, and it is lower to show that cancer probability is suffered from the region.
Compared with prior art, the advantageous effect of technical solution of the present invention is:
1, the interest region slice extraction scheme based on piecemeal that the present invention uses, if pathological image is divided into cadre Point, respectively to various pieces identification interest extracted region slice, traditional slice extracting method is compared, the present invention is to non-region of interest Domain has carried out more effective rejecting, improves the operation efficiency of entire positioning and method for visualizing;
2, the slice that the present invention uses arrives the mapping method of probability matrix, by coordinate mapping relations, during prediction Probability matrix is automatically updated, realizes the one-to-one correspondence of position and probability between pathological image corresponding region and canceration probability;
3, the slave probability probe image extraction feature that the present invention uses trains the scheme of a grader progress full figure prediction, Compared to the existing classification based on small slice, pathological image is effectively divided into positive and feminine gender by the present invention, improves prediction Accuracy rate;
4, cancerous region method for visualizing provided by the invention expresses prediction result visual representation for cancerous region probability It is expressed with cancerous region profile, for assisting pathologist to carry out analysis and diagnosis to pathological section, dramatically reduces pathology The work load of doctor improves accuracy rate of diagnosis.
Description of the drawings
Fig. 1 is the method for the present invention flow chart;
Fig. 2 is to be improved to the slice extracting method schematic diagram based on piecemeal from simplicity slice extracting method;
The areas Tu3Wei Ai position and method for visualizing schematic diagram.
Specific implementation mode
The attached figures are only used for illustrative purposes and cannot be understood as limitating the patent;
In order to more preferably illustrate that the present embodiment, the certain components of attached drawing have omission, zoom in or out, actual product is not represented Size;
To those skilled in the art, it is to be appreciated that certain known features and its explanation, which may be omitted, in attached drawing 's.
The following further describes the technical solution of the present invention with reference to the accompanying drawings and examples.
Embodiment 1
As shown in 1-3, a kind of ultrahigh resolution pathological image cancerous region method for visualizing includes the following steps:
S1:The slice extraction of interest region, group is divided into using the threshold value optimal algorithm based on pixel RGB values by described image Tissue region and extraneous background region are split tissue regions and extract slice.
S2:Model prediction and probability matrix structure, to biography before the small slice input convolutional neural networks model is carried out The canceration probability for obtaining every small slice is broadcast, all probability are constituted into probability matrix.
S3:Full figure is predicted, is carried out full figure prediction to ultrahigh resolution pathological image based on probability probe image, is divided into sun Property and negative two classes.
S4:Prediction result visualizes, and is based on full figure prediction result, probability probe image is mapped to obtain cancerous region probability Expression and the expression of cancerous region profile.
Above-mentioned steps specific implementation is as follows:
S1:The slice extraction of interest region
The slice extraction of interest region includes the following steps:
(1) full figure is scanned using the threshold value optimal algorithm based on pixel RGB values, 2 groups of optimal RGB threshold values is calculated, point Not Wei RGB bottom thresholds and RGB upper thresholds, all pixels in the threshold interval will all be determined as tissue regions;
(2) image averaging is divided into 4 parts, quadrant where recording various pieces respectively simultaneously calculates the part and original image Pixel coordinate mapping relations, be based respectively on optimal threshold and each section image be divided into interest region and background area, and point Interest region contour boundary rectangle is not drawn;
(3) according to the resolution information of sample to be tested image, size is the sliding window of 256*256 pixels with boundary rectangle The upper left corner be origin coordinates, adaptively slided with corresponding step-length, sweep test sample image, be based on optimal RGB thresholds Value judges whether sliding window inner tissue region accounting is more than 30%, if so, by image in window and the window center root The coordinate for mapping to artwork according to quadrant mapping relations is preserved as a Patch to input rank;If it is not, then scanning next window Mouthful.
S2:Model prediction and probability matrix structure
Pathological image resolution ratio to be measured is 20480*27904, the corresponding resolution ratio of working level Level-5, Level-5 For 640*872.
Model prediction and probability matrix structure comprise the steps of:
(1) the two-dimentional full null matrix Mp that structure dimension is 640*872;
(2) using 16 Patch of input rank as a Batch, each Batch is standardized, Input persistence deep neural network model carry out propagated forward, obtain every Patch canceration probability, by the probability with The respective element that the centre coordinate that corresponding slice maps to Level-5 is index upgrade matrix Mp.
S3:Full figure is predicted
From training set probability probe image extract feature include:Cancer area accounting, cancer area longest axis, canceration probability are more than 0.9 pixel number, cancer area Maximum Area, mean value, variance, degree of skewness and kurtosis, cancer area perimeter maximum value, mean value, variance, partially Gradient and kurtosis, cancer area profile eccentricity maximum value, mean value, variance, degree of skewness and kurtosis, cancer area profile ductility maximum value, Mean value, variance, degree of skewness and kurtosis, at the same in cancerous region and its minimal convex polygon the maximum value of pixel ratio, mean value, Variance, degree of skewness and kurtosis.By these features for training a support vector machines, and the persistence model is for predicting.
The feature that same type is extracted from probability probe image H inputs the support vector cassification model progress of the persistence Ultrahigh resolution pathological image is divided into positive and negative two classes by prediction.
S4:Prediction result visualizes
Full figure prediction result is the positive, and visualization includes following two steps:
(1) probability probe image H is subjected to transparency process with transparency threshold Alpha=0.7, then with it is corresponding The pathological image superposition of Level-5 generates new images, obtains the expression of cancerous region probability;
(2) probability probe image H is converted into gray-scale map H1, binary image H2 is generated based on gray-scale map H1, to H2 into Contour detecting is carried out after row corrosion and filling processing, profile coordinate information is mapped into corresponding Level-5 with outline form Pathological image obtain cancerous region profile expression.
The same or similar label correspond to the same or similar components;
Position relationship described in attached drawing is used to only for illustration, should not be understood as the limitation to this patent;
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair The restriction of embodiments of the present invention.For those of ordinary skill in the art, may be used also on the basis of the above description To make other variations or changes in different ways.There is no necessity and possibility to exhaust all the enbodiments.It is all this All any modification, equivalent and improvement etc., should be included in the claims in the present invention made by within the spirit and principle of invention Protection domain within.

Claims (8)

1. a kind of ultrahigh resolution pathological image cancerous region method for visualizing, which is characterized in that include the following steps:
S1:The slice extraction of interest region:Described image is divided by tissue area using the threshold value optimal algorithm based on pixel RGB values Domain and extraneous background region are split tissue regions and extract slice;
S2:Model prediction and probability matrix structure:The small slice input convolutional neural networks model is carried out propagated forward to obtain Gained probability is pressed default update method and builds probability matrix by the canceration probability for obtaining every small slice;
S3:Full figure is predicted:Full figure prediction is carried out to ultrahigh resolution pathological image based on probability probe image, be divided into it is positive and Negative two classes;
S4:Prediction result visualizes:Based on full figure prediction result, probability probe image is mapped to obtain the expression of cancerous region probability And cancerous region profile expression.
2. ultrahigh resolution pathological image cancerous region method for visualizing according to claim 1, which is characterized in that described The detailed process of step S1 is:
Optimal threshold is obtained using the threshold value optimal algorithm of pixel RGB values, based on optimal threshold to ultrahigh resolution pathological image It is detected, identifies tissue regions and draws tissue regions profile boundary rectangle;Using the upper left corner of boundary rectangle as origin coordinates, There is the slice that overlappingly interception resolution ratio is 256*256 to preserve to input rank.
3. ultrahigh resolution pathological image cancerous region method for visualizing according to claim 2, which is characterized in that described In step S2, if temperature figure resolution ratio corresponds to the Level-K on sample medical image pyramid, structure Level-K corresponds to dimension It is general to be carried out the corresponding prediction of propagated forward acquisition by the complete zero initial matrix Mp of two dimension of degree for slice input deep neural network model The probability is mapped to the centre coordinate of Level-K as the respective element of index upgrade matrix Mp by rate using corresponding slice.
4. ultrahigh resolution pathological image cancerous region method for visualizing according to claim 3, which is characterized in that described In step S3, a disaggregated model is trained from training set probability probe image extraction feature, the probability probe generated from test set The feature input disaggregated model of image zooming-out same type is predicted, ultrahigh resolution pathological image is divided into positive and cloudy Two classes of property.
5. ultrahigh resolution pathological image cancerous region method for visualizing according to claim 4, which is characterized in that described In step S4, probability probe image H is subjected to transparency process with threshold value A lpha, then with the pathological image of corresponding Level-K It is overlapped processing and generates new images, obtain the expression of cancerous region probability;Probability probe image H is subjected to binaryzation and corrosion With contour detecting is carried out after filling, obtain canceration area after profile coordinate information to be mapped to the pathological image of corresponding Level-K Domain profile expression.
6. ultrahigh resolution pathological image cancerous region method for visualizing according to claim 5, which is characterized in that described The slice for having overlapping intercept and preserve, including sliding window interception and tissue regions judge:Utilize the resolution ratio of sample to be tested image Information, sliding window, which adaptively has with corresponding step-length, overlappingly intercepts test sample;To the slice that is intercepted based on the optimal of gained Pixel RGB threshold values, judge whether biopsy tissues region accounting is more than 30%, if so, the slice is preserved and scans next Slice is judged;If it is not, then directly scanning next slice repeats above-mentioned judgment step.
7. ultrahigh resolution pathological image cancerous region method for visualizing according to claim 6, which is characterized in that described Identification tissue regions and draw tissue regions profile boundary rectangle, realize process be based on method of partition:By ultrahigh resolution disease If managing image uniform Ground Split into stem portion, tissue regions are identified to these parts of images respectively and are drawn outside tissue regions profile Connect rectangle.
8. ultrahigh resolution pathological image cancerous region method for visualizing according to claim 7, which is characterized in that described Slave test set generate probability probe image, based on probability matrix generate, with each element in probability matrix Mp be expand Two-dimensional matrix Mp is extended for three-dimensional matrice Mp1 by the factor, is based on Mp1 generating probability probe image H, wherein color is deeper, shows The region suffers from that cancer probability is higher, and color is more shallow, and it is lower to show that cancer probability is suffered from the region.
CN201810215800.3A 2018-03-15 2018-03-15 A kind of ultrahigh resolution pathological image cancerous region method for visualizing Pending CN108564567A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810215800.3A CN108564567A (en) 2018-03-15 2018-03-15 A kind of ultrahigh resolution pathological image cancerous region method for visualizing

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810215800.3A CN108564567A (en) 2018-03-15 2018-03-15 A kind of ultrahigh resolution pathological image cancerous region method for visualizing

Publications (1)

Publication Number Publication Date
CN108564567A true CN108564567A (en) 2018-09-21

Family

ID=63531809

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810215800.3A Pending CN108564567A (en) 2018-03-15 2018-03-15 A kind of ultrahigh resolution pathological image cancerous region method for visualizing

Country Status (1)

Country Link
CN (1) CN108564567A (en)

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109544507A (en) * 2018-10-18 2019-03-29 清影医疗科技(深圳)有限公司 A kind of pathological image processing method and system, equipment, storage medium
CN110232410A (en) * 2019-05-30 2019-09-13 清华大学 The organ-tissue Pathologic image analysis method and analytical equipment of multi-layer
CN110335284A (en) * 2019-07-11 2019-10-15 上海昌岛医疗科技有限公司 A kind of method of the removal background of pathological image
CN110378313A (en) * 2019-07-26 2019-10-25 玖壹叁陆零医学科技南京有限公司 Cell mass recognition methods, device and electronic equipment
CN110838100A (en) * 2019-10-11 2020-02-25 浙江大学 Colonoscope pathological section screening and segmenting system based on sliding window
CN110866870A (en) * 2019-10-29 2020-03-06 中山大学 Super-resolution processing method for amplifying medical image by any multiple
CN110930369A (en) * 2019-11-01 2020-03-27 中山大学 Pathological section identification method based on group equal variation neural network and conditional probability field
CN111145176A (en) * 2019-04-15 2020-05-12 青岛大学 Method and system for automatically identifying lymph node staining pathological image of gastric cancer based on deep neural network
CN111784637A (en) * 2020-06-04 2020-10-16 复旦大学附属中山医院 Prognostic characteristic visualization method, system, equipment and storage medium
CN112184618A (en) * 2020-08-17 2021-01-05 清华大学 Grape fetus slice image processing method and device based on deep learning
WO2021032062A1 (en) * 2019-08-22 2021-02-25 京东方科技集团股份有限公司 Image processing model generation method, image processing method, apparatus, and electronic device
CN112419253A (en) * 2020-11-16 2021-02-26 中山大学 Digital pathological image analysis method, system, device and storage medium
CN113657401A (en) * 2021-08-24 2021-11-16 凌云光技术股份有限公司 Probability map visualization method and device for defect detection
CN114140445A (en) * 2021-12-06 2022-03-04 上海派影医疗科技有限公司 Breast cancer pathological image identification method based on key attention area extraction
WO2022110525A1 (en) * 2020-11-24 2022-06-02 江苏大学 Comprehensive detection apparatus and method for cancerous region
CN115661815A (en) * 2022-12-07 2023-01-31 赛维森(广州)医疗科技服务有限公司 Pathological image classification method and image classification device based on global feature mapping

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160110632A1 (en) * 2014-10-20 2016-04-21 Siemens Aktiengesellschaft Voxel-level machine learning with or without cloud-based support in medical imaging
CN105975793A (en) * 2016-05-23 2016-09-28 麦克奥迪(厦门)医疗诊断系统有限公司 Auxiliary cancer diagnosis method based on digital pathological images
CN106127255A (en) * 2016-06-29 2016-11-16 深圳先进技术研究院 The sorting technique of a kind of cancer numeral pathological cells image and system
CN107368670A (en) * 2017-06-07 2017-11-21 万香波 Stomach cancer pathology diagnostic support system and method based on big data deep learning
CN107403201A (en) * 2017-08-11 2017-11-28 强深智能医疗科技(昆山)有限公司 Tumour radiotherapy target area and jeopardize that organ is intelligent, automation delineation method

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160110632A1 (en) * 2014-10-20 2016-04-21 Siemens Aktiengesellschaft Voxel-level machine learning with or without cloud-based support in medical imaging
CN105975793A (en) * 2016-05-23 2016-09-28 麦克奥迪(厦门)医疗诊断系统有限公司 Auxiliary cancer diagnosis method based on digital pathological images
CN106127255A (en) * 2016-06-29 2016-11-16 深圳先进技术研究院 The sorting technique of a kind of cancer numeral pathological cells image and system
CN107368670A (en) * 2017-06-07 2017-11-21 万香波 Stomach cancer pathology diagnostic support system and method based on big data deep learning
CN107403201A (en) * 2017-08-11 2017-11-28 强深智能医疗科技(昆山)有限公司 Tumour radiotherapy target area and jeopardize that organ is intelligent, automation delineation method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LE HOU等: "Patch-based Convolutional Neural Network for Whole Slide Tissue Image Classification", 《2016 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)》 *

Cited By (26)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109544507A (en) * 2018-10-18 2019-03-29 清影医疗科技(深圳)有限公司 A kind of pathological image processing method and system, equipment, storage medium
CN111145176A (en) * 2019-04-15 2020-05-12 青岛大学 Method and system for automatically identifying lymph node staining pathological image of gastric cancer based on deep neural network
CN110232410A (en) * 2019-05-30 2019-09-13 清华大学 The organ-tissue Pathologic image analysis method and analytical equipment of multi-layer
CN110335284A (en) * 2019-07-11 2019-10-15 上海昌岛医疗科技有限公司 A kind of method of the removal background of pathological image
CN110378313B (en) * 2019-07-26 2021-05-18 玖壹叁陆零医学科技南京有限公司 Cell cluster identification method and device and electronic equipment
CN110378313A (en) * 2019-07-26 2019-10-25 玖壹叁陆零医学科技南京有限公司 Cell mass recognition methods, device and electronic equipment
US11887303B2 (en) 2019-08-22 2024-01-30 Beijing Boe Technology Development Co., Ltd. Image processing model generation method, image processing method and device, and electronic device
WO2021032062A1 (en) * 2019-08-22 2021-02-25 京东方科技集团股份有限公司 Image processing model generation method, image processing method, apparatus, and electronic device
CN110838100A (en) * 2019-10-11 2020-02-25 浙江大学 Colonoscope pathological section screening and segmenting system based on sliding window
CN110866870A (en) * 2019-10-29 2020-03-06 中山大学 Super-resolution processing method for amplifying medical image by any multiple
CN110866870B (en) * 2019-10-29 2023-04-28 中山大学 Super-resolution processing method for amplifying medical image by any multiple
CN110930369A (en) * 2019-11-01 2020-03-27 中山大学 Pathological section identification method based on group equal variation neural network and conditional probability field
CN110930369B (en) * 2019-11-01 2023-05-05 中山大学 Pathological section identification method based on group et-variable neural network and conditional probability field
CN111784637A (en) * 2020-06-04 2020-10-16 复旦大学附属中山医院 Prognostic characteristic visualization method, system, equipment and storage medium
CN112184618A (en) * 2020-08-17 2021-01-05 清华大学 Grape fetus slice image processing method and device based on deep learning
WO2022100022A1 (en) * 2020-11-16 2022-05-19 中山大学 Digital pathological image analysis method and system, and device and storage medium
CN112419253A (en) * 2020-11-16 2021-02-26 中山大学 Digital pathological image analysis method, system, device and storage medium
CN112419253B (en) * 2020-11-16 2024-04-19 中山大学 Digital pathology image analysis method, system, equipment and storage medium
WO2022110525A1 (en) * 2020-11-24 2022-06-02 江苏大学 Comprehensive detection apparatus and method for cancerous region
US11587231B2 (en) 2020-11-24 2023-02-21 Jiangsu University Comprehensive detection device and method for cancerous region
CN113657401A (en) * 2021-08-24 2021-11-16 凌云光技术股份有限公司 Probability map visualization method and device for defect detection
CN113657401B (en) * 2021-08-24 2024-02-06 凌云光技术股份有限公司 Probability map visualization method and device for defect detection
CN114140445B (en) * 2021-12-06 2022-10-28 上海派影医疗科技有限公司 Breast cancer pathological image identification method based on key attention area extraction
CN114140445A (en) * 2021-12-06 2022-03-04 上海派影医疗科技有限公司 Breast cancer pathological image identification method based on key attention area extraction
CN115661815A (en) * 2022-12-07 2023-01-31 赛维森(广州)医疗科技服务有限公司 Pathological image classification method and image classification device based on global feature mapping
CN115661815B (en) * 2022-12-07 2023-09-12 赛维森(广州)医疗科技服务有限公司 Pathological image classification method and device based on global feature mapping

Similar Documents

Publication Publication Date Title
CN108564567A (en) A kind of ultrahigh resolution pathological image cancerous region method for visualizing
CN109583440B (en) Medical image auxiliary diagnosis method and system combining image recognition and report editing
US10262189B2 (en) Evaluation of co-registered images of differently stained tissue slices
CN113781439B (en) Ultrasonic video focus segmentation method and device
CN101826209B (en) Canny model-based method for segmenting three-dimensional medical image
Garg et al. Unsupervised curvature-based retinal vessel segmentation
Giannini et al. A fully automatic algorithm for segmentation of the breasts in DCE-MR images
CN105513036B (en) The dividing method and device of three-dimensional CT image
CN112132166B (en) Intelligent analysis method, system and device for digital cell pathology image
CN107798679A (en) Breast molybdenum target image breast area is split and tufa formation method
CN112263217B (en) Improved convolutional neural network-based non-melanoma skin cancer pathological image lesion area detection method
CN104732520A (en) Cardio-thoracic ratio measuring algorithm and system for chest digital image
CN113538471B (en) Plaque segmentation method, plaque segmentation device, computer equipment and storage medium
CN111784701A (en) Ultrasonic image segmentation method and system combining boundary feature enhancement and multi-scale information
Chen et al. Identifying metastases in sentinel lymph nodes with deep convolutional neural networks
CN105869169B (en) A kind of automatic division method of the micro- arrangement image of tumor tissues
EP3381010B1 (en) Process for processing medical images of a face for recognition of facial dysmorphisms
Moran et al. Evaluation of quantitative features and convolutional neural networks for nodule identification in thyroid thermographies
Rajaraman et al. Connected components labeling and extraction based interphase removal from chromosome images
CN113689950B (en) Method, system and storage medium for identifying blood vessel distribution pattern of liver cancer IHC staining pattern
Huidrom et al. Automated lung segmentation on computed tomography image for the diagnosis of lung cancer
Kumar et al. Semiautomatic method for segmenting pedicles in vertebral radiographs
Song et al. Liver segmentation based on SKFCM and improved GrowCut for CT images
Jeena et al. Analysis of stroke using texture features
Wang et al. Application of MP-YOLO for Segmentation and Visualization of Ovarian Ultrasound Imaging

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20180921

RJ01 Rejection of invention patent application after publication