US20170140528A1 - Automated histological diagnosis of bacterial infection using image analysis - Google Patents

Automated histological diagnosis of bacterial infection using image analysis Download PDF

Info

Publication number
US20170140528A1
US20170140528A1 US15/113,712 US201515113712A US2017140528A1 US 20170140528 A1 US20170140528 A1 US 20170140528A1 US 201515113712 A US201515113712 A US 201515113712A US 2017140528 A1 US2017140528 A1 US 2017140528A1
Authority
US
United States
Prior art keywords
image
specimen
slide
images
decision support
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.)
Abandoned
Application number
US15/113,712
Other languages
English (en)
Inventor
Amir Aharon Handzel
Isaac Daniel Buzaglo
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.)
Individual
Original Assignee
Individual
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 Individual filed Critical Individual
Priority to US15/113,712 priority Critical patent/US20170140528A1/en
Publication of US20170140528A1 publication Critical patent/US20170140528A1/en
Abandoned legal-status Critical Current

Links

Images

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
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/51Indexing; Data structures therefor; Storage structures
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/24Classification techniques
    • G06F18/241Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G06K9/4628
    • G06K9/627
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/20Image preprocessing
    • G06V10/26Segmentation 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
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/42Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/44Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/40Extraction of image or video features
    • G06V10/44Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
    • G06V10/449Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters
    • G06V10/451Biologically inspired filters, e.g. difference of Gaussians [DoG] or Gabor filters with interaction between the filter responses, e.g. cortical complex cells
    • G06V10/454Integrating the filters into a hierarchical structure, e.g. convolutional neural networks [CNN]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/70Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/751Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/70Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/60Type of objects
    • G06V20/69Microscopic objects, e.g. biological cells or cellular parts
    • G06V20/695Preprocessing, e.g. image segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/60Type of objects
    • G06V20/69Microscopic objects, e.g. biological cells or cellular parts
    • G06V20/698Matching; Classification
    • G06K2209/05
    • 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
    • G06T2207/20084Artificial 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
    • G06T2207/30092Stomach; Gastric
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00Indexing scheme relating to image or video recognition or understanding
    • G06V2201/03Recognition of patterns in medical or anatomical images

Definitions

  • the invention relates generally to the field of image processing, and more particularly relates to automated optical inspection, template matching segmentation, classification, features extraction, pattern recognition techniques and machine learning of digitized microscope slides.
  • the invention relates also to histology analyses, and in particular, to decision support systems for diagnostic use by pathology professionals for the determination of the presence of bacteria in tissue biopsies, in general, and presence of Helicobacter pylori (“HP”) in histological specimens of the stomach, in particular.
  • HP Helicobacter pylori
  • Histology is a branch of science which studies the microscopic structures of organic tissues. Histological analysis is performed by examining a thin slice of a microscopically sectioned tissue biopsy under a microscope, in order to determine the possible presence of cells of a certain cellular type, or to recognize a variety of microbial components or to recognize architectural structures of the tissue. For example, a histological analysis may be performed on a tissue sample slide to detect the presence of HP in a tissue biopsy.
  • HP is one of the most common bacterial pathogens; it infects the gastric mucosa in humans and it is estimated that by adulthood over half the world population is infected.
  • This bacterium is Gram negative, spiral-shaped, and motile by polar flagella.
  • a certain proportion of the infected population will develop acute gastritis or gastric or duodenal ulcers. Some persons may develop gastric cancers. Because of strong association between gastric cancer and HP infection this bacterium is classified as a bacterial carcinogen.
  • Gastric biopsies are often extracted during endoscopic examination to rule out HP infection by histologic evaluation. Histology is considered to be the reference (“gold standard”) in the direct diagnosis of HP gastritis. Before observation, slices are subjected to staining in order to obtain the prepared pieces to be inspected and enhance contrast. An experienced pathologist is needed to inspect the histological slice in order to determine the possible presence of HP. This procedure is exhausting and prone to a high false negative rate due to the tedium and fatigue suffered by the pathologist.
  • the slide specimen includes many complex arrangements and architectures, including overlapping tissue components, cell boundaries and nuclei corrupted by noise which is not present in a blood sample for example. Some structures, such as cells, may appear connected, blurred or occluded by other tissue elements in the image. Furthermore, the cell architecture and bacteria may be presented in various 3D orientations due to the method of slicing that exacerbate the challenge of image analysis. Those complexities have made histological slides difficult to extract cell regions and bacteria by traditional image segmentation approaches. When attempting to segment and recognize bacteria in tissue all the above challenges exist and are made significantly more complicated because of the small size of bacteria in histological images both in absolute measure and by their much smaller size relative to the surrounding tissue features. Identification of the boundaries of a bacterium is especially difficult on such background.
  • the present invention relates to a method, a system and an apparatus for automated diagnosis and decision support of histology slides through automated detection of bacteria in general, and more specifically of HP.
  • the invention is a method for automated analysis and diagnosis of the presence of a bacterium in a histological specimen, comprising: providing a specimen-stained slide; obtaining a color scanned image of the slide with a color digital camera (such as a charge-coupled device (“CCD”) or complementary metal-oxide semiconductor (“CMOS”) digital camera); associating identification data with the slide (which may be done using manual data entry, a barcode reader or optical character recognition, for example); transforming the color scanned image into a single band image; searching parts of the single band image that match a pattern identified by gray level values, wherein parts matching the pattern are candidate object components; classifying the candidate object components with a subsequent additional machine language classifier (preferably a CNN classifier); creating from the CNN classifier an automatic decision support diagnosis report of the presence or absence of the bacterium in the specimen; and storing images with associated metadata and the decision support diagnosis report in a computer readable medium.
  • a color digital camera such as a charge-coupled device (“CCD”) or complementary metal-oxid
  • the method includes receiving diagnosis information (for example from a pathologist making a determination based on the support diagnosis report) and including the diagnosis information in the CNN classifier (or other machine language classifier) going forward.
  • diagnosis information for example from a pathologist making a determination based on the support diagnosis report
  • CNN classifier or other machine language classifier
  • the method omits the steps of pattern identification and relies completely on a CNN that has been trained using a database of images to produce an automatic support diagnosis report.
  • the CNN having previously been trained with a database of images, has acquired the ability to recognize and report with some precision the presence of bacteria in the tissue image.
  • the necessary information is implicitly stored in its large set of internal parameters whose values have been gradually adjusted during the presentation of the images in the database using an appropriate training algorithm.
  • the invention is a system for automated analysis and diagnosis of the presence of a bacterium in a histological specimen, comprising: a specimen-stained slide; an automated microscope for enlarging the images of the slide; a color digital video camera (CCD or CMOS for example) for creating a digital image from the slide; a mechanical stage control for automatically scanning the specimen; a slide feeder; one or more computer(s) running an image processing classifier for analyzing the digital image and providing a support diagnosis report of the presence or absence of the bacterium in the specimen.
  • the system further comprises an external database for images and report storage.
  • the one or more computer(s) run an image processing classifier adapted to search parts of a single band image obtained from the digital image that match a pattern of a bacterium identified by gray level values, wherein parts matching the pattern are identified as candidate object components.
  • the computer(s) are adapted to run an additional CNN classifier to classify the candidate object components to obtain an automatic decision support diagnosis report of the presence or absence of the bacterium in the specimen, which is then stored in the external database.
  • a microscope for enlarging the images of the gastric, intestinal or other tissue, such as a commercially available automated histology microscope with a color CCD or CMOS digital video camera, having a manually or automatically controlled stage for scanning a slide.
  • external illumination system(s) provide visible light illumination.
  • the microscope is further provided with a mechanical stage having the ability to be controlled externally by a computer for automatic scanning of a slide using an automated slide feeder.
  • the CCD or CMOS digital video camera is adapted to obtain 3D images of a histological specimen slide for analysis.
  • An example of a 3D imaging method comprises optical sectioning of the tissue slide, using emitted in-focus and out-of-focus light, to create multiple layers of 2D images.
  • the invention includes an initial module comprising template matching methods.
  • This initial classifier identifies objects in images that have image characteristics of bacteria, in particular of HP.
  • the output of this module is a set of candidates of such image objects.
  • a candidate object component is a part of an image, which may be as small as a pixel, which is considered more likely to be within HP.
  • These are fed to a subsequent classifier that further processes the candidates by algorithms of pattern recognition, such as machine learning and CNN, and confirms or rejects their identity as the target object.
  • the information about confirmation or rejection by each pathologist may then be communicated back to a central processor.
  • the data thus obtained may be evaluated and maintained on a central server(s) such that the data can be reintroduced into the CNN.
  • This feedback information is accumulated from multiple, possibly independent users of the system who are located in diverse and distant locations. It is used to further calibrate and fine-tune the CNN for improved performance.
  • Another objective of the present invention is to provide a system to enable the storage of large quantity of data and images that are accumulated. It is proposed that the system can include local storage or network storage linked in a network structure.
  • the images and the data will be accessible by a computer via a communication network.
  • the system provides an editable summary screen or screens showing relevant images from the automatic diagnosis containing bacteria, or specifically HP findings; in this application, the pathologist will diagnose the slide based on the automated support system.
  • FIG. 1 shows a schematic representation of an alternate embodiment of an automated screening apparatus and system in accordance with the methods of the invention.
  • FIG. 2 shows a flow chart summarizing the method of one embodiment of the invention, illustrating the steps from the biopsy procedure to the final diagnosis by a pathologist.
  • FIG. 3 shows a schematic illustration of a characteristic HP pattern, matching the mask template of multiple angular lines for the detection of the principal edges of the object.
  • FIG. 4 shows a block diagram of an embodiment of a CNN according to the present invention.
  • FIG. 5 shows a schematic representation of the Continuous calibration system. This is a feedback cycle created for the improvement of the diagnosis system engine situated at the client location.
  • FIG. 6 shows a schematic representation of data visualization flow chart performed by the examined person with a mobile device or a computer.
  • FIG. 1 illustrates a schematic representation of an automated screening apparatus and system according to an embodiment of the invention.
  • Image acquisition system 101 includes visible light illumination 102 , X-Y automated stage 103 equipped with a motor controller for stage movement, slide feeder 104 , microscope 105 , and a color RGB CCD or CMOS camera 106 to acquire digital images.
  • the images obtained from image acquisition system 101 are transferred to image processing module 107 for automated classification decision.
  • Slide data is sent to the data storage module 108 in the local or network computers, waiting for the pathologist 109 to upload suspected images with HP for diagnosis review.
  • FIG. 2 illustrates a schematic representation of an embodiment of a method of the invention.
  • an endoscopic examination is performed to select microscopic biopsies from the stomach 201 .
  • These tissue biopsies are sectioned, mounted on a glass slide and stained to enhance contrast in the microscopic image.
  • the histological specimen slices may be stained, by H&E or Giemsa staining to enhance certain features in the image.
  • the slide is then loaded on an automated screening device 202 for the acquisition of color image or images of the histology slide specimen.
  • Metadata 203 are collected by a barcode reader or an OCR reader, or entered manually.
  • Metadata includes information relevant to the specimen, such as, without limitation, indexing information, source identification of the specimen, how the image was obtained, and the like.
  • a slide may be looked up in the system by specifying unique slide identification metadata, and data associated with the specimen may be associated with the identification data.
  • the color RGB image or images 204 are transformed into a single band image (gray level) which is then normalized (for example by a weighted normalized superposition of the three bands, or by applying Mahalanobis color segmentation).
  • the resultant gray scale image is used for the template matching algorithm 205 .
  • Template matching techniques are used to search small parts of the image that match the HP template pattern (or “mask”). Two specific template matching methods are described here.
  • initial edge detection and completion are performed resulting in the segmentation of the image and identification of areas of closed objects in the image.
  • Spatial two dimensional second moments are calculated for each closed object.
  • the eigenvectors and eigenvalues of the image covariance matrix are calculated.
  • For each closed object conditions are tested for the total area and for the covariance eigenvalues; if these parameters correspond to those of images of HP then the closed object is declared a candidate HP.
  • the image is scanned pixel by pixel, and each pixel is tested for being in the interior and near the central area of a bacterium in the image.
  • This analysis uses a sub-image “moving window” centered at each pixel as it is being analyzed. Once the analysis of a pixel is completed the sub-image window moves to the next pixel; the procedure continues until the image is fully scanned. A detailed description of the analysis procedure for each pixel is given below in connection with FIG. 3 .
  • the above template matching process feeds the subsequent machine learning module with a reduced set of HP template candidates.
  • Pattern recognition module 206 uses supervised machine learning to provide definitive identification of bacteria candidates in the tissue.
  • the machine learning architecture is a CNN.
  • the CNN performs feature extraction and classification within the same architecture, preserving the two-dimensional spatial structure of the input image.
  • the input layer of the supervised machine learning is a 36 ⁇ 36 sub-image centered at each of the centroids previously found by the template matching algorithm. It will be appreciated by the person of ordinary skill in the art that the CNN may be implemented with a variety of equivalent structures.
  • the output layer generated by the supervised CNN network has two classification labels: positive or negative HP found in the sub-image.
  • an automated diagnosis output report 207 is produced and appended to the metadata of the slide.
  • the report summarizes the findings of HP, by mapping the positions of the HP found into an additional file for each relevant image or for all the images of the slide.
  • the slide images, the metadata and the automated diagnosis report are transferred to a local storage or network storage 208 for storing and retrieving.
  • a pathologist loads the slide images and diagnosis for validation.
  • the software provides image annotation by displaying and highlighting or marking on the screen small regions with HP findings, to be validated by the pathologist. Those small regions can be expanded, moved to any location within the image or zoomed in or out.
  • the pathologist can add or remove image annotation and findings.
  • the pathologist's diagnosis is stored with the slide data, and possibly printed 211 for delivering a report. It can also be linked to other forms of reporting, such as Electronic Medical Records, in the healthcare center or system, thus providing an end-to-end system.
  • Information related to the pathologist's diagnosis and subsequent processing, is collectively referred to herein as “diagnosis information.”
  • a specific embodiment of a method of the invention is to perform 206 , the CNN, without 205 , the Template Matching classifier.
  • a surrounding rectangle of 36 ⁇ 36 pixels is fed as input to the machine learning engine in order to classify it as positive or negative HP finding.
  • a template mask is used.
  • a “mask” may have fixed pattern elements as well as additional dynamic constructive elements, analogous to finite impulse filters in classic one-dimensional classical signal processing.
  • FIG. 3 depicts a template mask construction for a characteristic HP bacterium pattern 301 .
  • the reference template mask is composed of 8 rays of length MaxLineLength 302 , all emanating from the tested pixel and situated at 45 degrees angle between adjacent ones. For images optically magnified by 60 ⁇ MaxLineLength equals 10 pixels. The following conditions are then computed and tested; all need to be satisfied for the pixel to be declared a candidate for presence of HP:
  • the tested pixel is added to the list of HP candidate pixels.
  • centroid For each group of associated pixels the centroid is calculated; each centroid then represents a candidate HP bacterium image object.
  • the above example refers to HP bacteria which have spiral shape.
  • Other bacteria of different shape families are detected by the same template matching model but with different values for MaxLineLength of the 8 rays, and different relations among the resulting LineLength values of the rays in the specified conditions.
  • the template matching conditions for the detection of the Shigella dysenteriae bacteria is the same as shown in FIG. 3 example, except for condition number four:
  • the two largest LineLength values are checked for being of rays at a relative angle (without orientation) of 180 degrees.
  • the template matching conditions to detect the Coccidioides immitis bacteria are different except for the condition 1 (which is the same), as follows:
  • the fourth condition is omitted.
  • the CNN architecture of the network FIG. 4 detects the HP bacteria in tissue, but can also be used to detect other bacteria in tissue.
  • the CNN contains two convolutional layers ( 401 , 403 ), two pooling layers ( 402 , 404 ), and three fully-connected layers ( 405 , 406 , 407 ), but any other CNN of appropriate architecture could work.
  • the input layer is a 36 ⁇ 36 single band sub-image of the HP candidate.
  • 8 first order Gabor filters of size 9*9 are convolved separately with the input data resulting in 8 28 ⁇ 28 sub-images.
  • the second layer performs 2 ⁇ 2 pooling of the data of the first layer resulting in 8 14*14 sub-images.
  • the third layer is a Convolutional layer convolving 4 kernels of size 5 ⁇ 5 with the second layer resulting in 32 10 ⁇ 10 sub-images.
  • the Fourth layer performs 2 ⁇ 2 pooling of the data resulting in 32 5 ⁇ 5 sub-images.
  • the CNN In order for the CNN to perform the classification task, first it has to be trained, i.e. its internal parameters have to be adjusted based on real data of images. To train the network, a set of images is shown to the CNN whose weights and biases are calibrated by using gradient descent on the output error, e.g. through the Backpropagation algorithm, or Restricted Boltzmann Machine Learning, the specifics of which are well known in the art. Variations to the training algorithm can be used, all relying on the core principle of using a database of images which are presented to the CNN during training. During training the CNN is provided with labels of categories for the presented images, though not necessarily for all of them. Relevant labels for an image are “HP bacteria present” and “HP bacteria absent”; other labels are possible according to the same principle.
  • the template matching can be implemented in Matlab software package R2012b version 8.2.0.701
  • Gabor filters can be implemented using the Signal Processing and Image Processing toolboxes in Matlab
  • the CNN can be implemented using the CNN function class available on Mathworks File Exchange (http://www.mathworks.com/matlabcentral/fileexchange/24291-cnn-convolutional-neural-network-class).
  • Mathworks File Exchange http://www.mathworks.com/matlabcentral/fileexchange/24291-cnn-convolutional-neural-network-class.
  • the analysis may be performed by individually analyzing each of the 2D images extracted from the slide. The results of the analysis of all the layers are then combined such that all identified bacteria in all the layers are identified and reported
  • FIG. 5 illustrates the basic information feedback cycle for the improvement of the automated diagnosis engine.
  • the information about confirmation or rejection of the automated diagnosis by a pathologist 501 is then communicated back to a central processor or computer server 503 that is managed as part of the proprietary system 502 .
  • This feedback information is accumulated from multiple, possibly independent users of the system who are located in diverse and distant locations.
  • the accumulated data is examined for quality control (QC) and the high quality data are added to the existing curated database 504 .
  • the added data is used to further calibrate and fine-tune the CNN for improved performance and generalization 505 .
  • the training process of the CNN model is based on increasing amounts of data with growing accuracy and performance parameters, improving the CNN system.
  • the resulting CNN model is then tested 506 and released back to the client sites 507 .
  • FIG. 6 illustrates an additional embodiment of the invention based on the data accumulated from remote locations 601 and stored on a computer server 602 as part of system 603 .
  • the patient data is uploaded to a mobile device or PC, through the company application form 604 , and contains a slide viewer 605 and a diagnosis report viewer 606 .
  • the application enables data sharing of the pathologist report and the slides for a second opinion diagnosis 607 .

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Investigating Or Analysing Biological Materials (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)
US15/113,712 2014-01-25 2015-01-26 Automated histological diagnosis of bacterial infection using image analysis Abandoned US20170140528A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US15/113,712 US20170140528A1 (en) 2014-01-25 2015-01-26 Automated histological diagnosis of bacterial infection using image analysis

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US201461931572P 2014-01-25 2014-01-25
US15/113,712 US20170140528A1 (en) 2014-01-25 2015-01-26 Automated histological diagnosis of bacterial infection using image analysis
PCT/US2015/012809 WO2015112932A1 (fr) 2014-01-25 2015-01-26 Diagnostic histologique automatisé d'infection bactérienne à l'aide d'une analyse d'image

Publications (1)

Publication Number Publication Date
US20170140528A1 true US20170140528A1 (en) 2017-05-18

Family

ID=53679507

Family Applications (2)

Application Number Title Priority Date Filing Date
US15/113,712 Abandoned US20170140528A1 (en) 2014-01-25 2015-01-26 Automated histological diagnosis of bacterial infection using image analysis
US14/604,848 Expired - Fee Related US9684960B2 (en) 2014-01-25 2015-01-26 Automated histological diagnosis of bacterial infection using image analysis

Family Applications After (1)

Application Number Title Priority Date Filing Date
US14/604,848 Expired - Fee Related US9684960B2 (en) 2014-01-25 2015-01-26 Automated histological diagnosis of bacterial infection using image analysis

Country Status (2)

Country Link
US (2) US20170140528A1 (fr)
WO (1) WO2015112932A1 (fr)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020096889A1 (fr) * 2018-11-05 2020-05-14 Medivators Inc. Évaluation de dommages de canal d'endoscope à l'aide d'une analyse vidéo à intelligence artificielle
US11195313B2 (en) * 2016-10-14 2021-12-07 International Business Machines Corporation Cross-modality neural network transform for semi-automatic medical image annotation

Families Citing this family (38)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2014235494A (ja) * 2013-05-31 2014-12-15 富士ゼロックス株式会社 画像処理装置及びプログラム
US9760990B2 (en) * 2014-12-14 2017-09-12 International Business Machines Corporation Cloud-based infrastructure for feedback-driven training and image recognition
CN106156807B (zh) * 2015-04-02 2020-06-02 华中科技大学 卷积神经网络模型的训练方法及装置
US9940577B2 (en) * 2015-07-07 2018-04-10 Adobe Systems Incorporated Finding semantic parts in images
CN105022836B (zh) * 2015-08-14 2018-07-03 中国科学技术大学 基于sift嵌入的紧凑的深度cnn特征索引方法
US9739783B1 (en) 2016-03-15 2017-08-22 Anixa Diagnostics Corporation Convolutional neural networks for cancer diagnosis
CN107239785A (zh) * 2016-03-29 2017-10-10 北京市商汤科技开发有限公司 一种在智能设备中对图片进行分类的方法和系统
US9589374B1 (en) * 2016-08-01 2017-03-07 12 Sigma Technologies Computer-aided diagnosis system for medical images using deep convolutional neural networks
CN106407891B (zh) * 2016-08-26 2019-06-28 东方网力科技股份有限公司 基于卷积神经网络的目标匹配方法及装置
CN107844487A (zh) * 2016-09-19 2018-03-27 中国科学院大连化学物理研究所 用于谱分析光电子/低能电子显微镜的自动数据采集方法
US10387740B2 (en) * 2016-10-10 2019-08-20 Gyrfalcon Technology Inc. Object detection and recognition apparatus based on CNN based integrated circuits
US10366328B2 (en) * 2017-09-19 2019-07-30 Gyrfalcon Technology Inc. Approximating fully-connected layers with multiple arrays of 3x3 convolutional filter kernels in a CNN based integrated circuit
CN106570467B (zh) * 2016-10-25 2019-05-24 南京南瑞集团公司 一种基于卷积神经网络的人员离岗检测方法
JP2019537157A (ja) 2016-12-01 2019-12-19 バークレー ライツ,インコーポレイテッド マイクロ流体デバイスによる微小物体の自動検出及び再配置
CN108509826B (zh) * 2017-02-27 2022-03-01 千寻位置网络有限公司 一种遥感影像的道路识别方法及其系统
US11164082B2 (en) 2017-02-28 2021-11-02 Anixa Diagnostics Corporation Methods for using artificial neural network analysis on flow cytometry data for cancer diagnosis
US10360499B2 (en) 2017-02-28 2019-07-23 Anixa Diagnostics Corporation Methods for using artificial neural network analysis on flow cytometry data for cancer diagnosis
US9934364B1 (en) 2017-02-28 2018-04-03 Anixa Diagnostics Corporation Methods for using artificial neural network analysis on flow cytometry data for cancer diagnosis
JP7055818B2 (ja) * 2017-04-13 2022-04-18 シーメンス・ヘルスケア・ダイアグノスティックス・インコーポレーテッド 畳み込みニューラルネットワークを用いたhiln評価方法及び装置
JP6947841B2 (ja) * 2017-06-13 2021-10-13 グーグル エルエルシーGoogle LLC 病理学用の拡張現実顕微鏡
CN107329233B (zh) * 2017-08-25 2019-10-11 西南大学 一种基于神经网络的微滴式pcr仪自动对焦方法
WO2019043458A2 (fr) * 2017-08-30 2019-03-07 Bioaxial Sas Procédés de métrologie à super résolution basés sur des distributions singulières et un apprentissage profond
TW201922174A (zh) * 2017-10-30 2019-06-16 公益財團法人癌症研究會 圖像診斷支援裝置、資料收集方法、圖像診斷支援方法及圖像診斷支援程式
CN108121805B (zh) * 2017-12-25 2021-09-14 广东三维家信息科技有限公司 基于深度学习实现快速珠宝图片检索的方法
CN108846312B (zh) * 2018-05-07 2021-03-12 深圳市恒扬数据股份有限公司 一种细菌有效活动区域的识别方法、装置及终端设备
WO2019222135A1 (fr) 2018-05-16 2019-11-21 Benevis Informatics, Llc Systèmes et procédés pour l'examen de la détection assistée par ordinateur d'une pathologie dans des images
DE112019002848T5 (de) * 2018-06-06 2021-04-01 Cognex Corporation System und verfahren zum auffinden und klassifizieren von mustern in einem bild mit einem bildverarbeitungssystem
WO2019243897A2 (fr) * 2018-06-19 2019-12-26 Metasystems Hard And Software Gmbh Système et procédé de détection et de classification d'objets d'intérêt dans des images de microscope par apprentissage machine supervisé
CN109063766B (zh) * 2018-07-31 2021-11-30 湘潭大学 一种基于判别性预测稀疏分解模型的图像分类方法
KR20200030806A (ko) * 2018-09-13 2020-03-23 삼성전자주식회사 반도체 제조 공정을 모니터링하기 위해 인공 신경망에 기초하는 이미지 변환 모델을 포함하는 비일시적 컴퓨터 판독가능 매체 그리고 반도체 웨이퍼의 이미지를 변환하는 방법
EP3988988A4 (fr) * 2018-09-28 2023-09-13 Evident Corporation Système de microscope, unité de projection et procédé de projection d'image
JP7150866B2 (ja) 2018-09-28 2022-10-11 株式会社エビデント 顕微鏡システム、投影ユニット、及び、画像投影方法
JP2022506135A (ja) * 2018-10-30 2022-01-17 アレン インスティテュート ヒトの寄与を組み込む反復的深層学習フローを使用した顕微鏡画像内の3d細胞間構造のセグメント化
US10902590B2 (en) 2018-11-28 2021-01-26 International Business Machines Corporation Recognizing pathological images captured by alternate image capturing devices
US11592657B2 (en) 2018-12-18 2023-02-28 Hemotech Cognition, LLC Method and system for identifying objects in a blood sample
WO2020154698A1 (fr) 2019-01-25 2020-07-30 Otonexus Medical Technologies, Inc. Apprentissage machine pour diagnostic d'otite moyenne
US11557033B2 (en) 2019-08-09 2023-01-17 International Business Machines Corporation Bacteria classification
JP6782825B1 (ja) * 2019-09-12 2020-11-11 三菱電機インフォメーションシステムズ株式会社 通知判定装置、通知判定方法及び通知判定プログラム

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6819790B2 (en) * 2002-04-12 2004-11-16 The University Of Chicago Massive training artificial neural network (MTANN) for detecting abnormalities in medical images
US20090028403A1 (en) * 2006-03-03 2009-01-29 Medic Vision - Brain Technologies Ltd. System and Method of Automatic Prioritization and Analysis of Medical Images
US20120328178A1 (en) * 2010-06-25 2012-12-27 Cireca Theranostics, Llc Method for analyzing biological specimens by spectral imaging
US8488877B1 (en) * 2009-12-02 2013-07-16 Hrl Laboratories, Llc System for object recognition in colorized point clouds
US20130272587A1 (en) * 2010-08-25 2013-10-17 Siemens Corporation System and method for interactive segmentation on mobile devices in a cloud computing environment
US8879813B1 (en) * 2013-10-22 2014-11-04 Eyenuk, Inc. Systems and methods for automated interest region detection in retinal images
US20150213302A1 (en) * 2014-01-30 2015-07-30 Case Western Reserve University Automatic Detection Of Mitosis Using Handcrafted And Convolutional Neural Network Features
US20170244908A1 (en) * 2016-02-22 2017-08-24 GenMe Inc. Video background replacement system

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
AU5783900A (en) * 1999-07-02 2001-01-22 Hypermed, Inc. Integrated imaging apparatus
AU2002322033A1 (en) * 2001-06-04 2002-12-16 Ikonisys Inc. Method for detecting infectious agents using computer controlled automated image analysis
US7894664B2 (en) * 2006-03-22 2011-02-22 University Of Washington Conditional shape model for image processing
WO2009142758A1 (fr) * 2008-05-23 2009-11-26 Spectral Image, Inc. Systèmes et procédés d’imagerie médicale hyperspectrale
JP5478618B2 (ja) * 2008-07-03 2014-04-23 エヌイーシー ラボラトリーズ アメリカ インク 上皮層検出器およびそれに関連する方法
KR20140104946A (ko) 2011-10-05 2014-08-29 시레카 테라노스틱스, 엘엘씨 스펙트럼 이미징에 의해 생물학적 표본을 분석하는 방법 및 시스템
EP2602608B1 (fr) * 2011-12-07 2016-09-14 Imec Analyse et tri des cellules biologiques dans un écoulement

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6819790B2 (en) * 2002-04-12 2004-11-16 The University Of Chicago Massive training artificial neural network (MTANN) for detecting abnormalities in medical images
US20090028403A1 (en) * 2006-03-03 2009-01-29 Medic Vision - Brain Technologies Ltd. System and Method of Automatic Prioritization and Analysis of Medical Images
US8488877B1 (en) * 2009-12-02 2013-07-16 Hrl Laboratories, Llc System for object recognition in colorized point clouds
US20120328178A1 (en) * 2010-06-25 2012-12-27 Cireca Theranostics, Llc Method for analyzing biological specimens by spectral imaging
US20130272587A1 (en) * 2010-08-25 2013-10-17 Siemens Corporation System and method for interactive segmentation on mobile devices in a cloud computing environment
US8879813B1 (en) * 2013-10-22 2014-11-04 Eyenuk, Inc. Systems and methods for automated interest region detection in retinal images
US20150213302A1 (en) * 2014-01-30 2015-07-30 Case Western Reserve University Automatic Detection Of Mitosis Using Handcrafted And Convolutional Neural Network Features
US20170244908A1 (en) * 2016-02-22 2017-08-24 GenMe Inc. Video background replacement system

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11195313B2 (en) * 2016-10-14 2021-12-07 International Business Machines Corporation Cross-modality neural network transform for semi-automatic medical image annotation
WO2020096889A1 (fr) * 2018-11-05 2020-05-14 Medivators Inc. Évaluation de dommages de canal d'endoscope à l'aide d'une analyse vidéo à intelligence artificielle

Also Published As

Publication number Publication date
US20150213599A1 (en) 2015-07-30
WO2015112932A1 (fr) 2015-07-30
US9684960B2 (en) 2017-06-20

Similar Documents

Publication Publication Date Title
US9684960B2 (en) Automated histological diagnosis of bacterial infection using image analysis
US11927738B2 (en) Computational microscopy based-system and method for automated imaging and analysis of pathology specimens
US20220076411A1 (en) Neural netork based identification of areas of interest in digital pathology images
US8600143B1 (en) Method and system for hierarchical tissue analysis and classification
Quinn et al. Deep convolutional neural networks for microscopy-based point of care diagnostics
US10692209B2 (en) Image processing apparatus, image processing method, and computer-readable non-transitory recording medium storing image processing program
DK2973397T3 (en) Tissue-object-based machine learning system for automated assessment of digital whole-slide glass
AU2014237346B2 (en) System and method for reviewing and analyzing cytological specimens
US10345218B2 (en) Automated slide assessments and tracking in digital microscopy
CN109815945B (zh) 一种基于图像识别的呼吸道检查结果判读系统和方法
US20240079116A1 (en) Automated segmentation of artifacts in histopathology images
CN114945941A (zh) 用于支持肿瘤检测和分析的非肿瘤分割
EP4196889A1 (fr) Modèles d'apprentissage par machine pour la localisation de cellules et classement appris par codage de rejet
Kanwal et al. Quantifying the effect of color processing on blood and damaged tissue detection in whole slide images
WO2016189469A1 (fr) Procédé de dépistage médical et système associé
Hossain et al. Tissue artifact segmentation and severity assessment for automatic analysis using wsi
US20220138939A1 (en) Systems and Methods for Digital Pathology
CN113450305A (zh) 医疗图像的处理方法、系统、设备及可读存储介质
EP3818468A1 (fr) Procédé et appareil d'entraînement de classificateur à réseau neuronal pour classifier une image représentant un ou plusieurs objets d'un échantillon biologique
US20220260825A1 (en) Scanning/pre-scanning quality control of slides
Ramarolahy et al. Classification and generation of microscopy images with plasmodium falciparum via artificial neural networks using low cost settings
US20230230242A1 (en) Correcting differences in multi-scanners for digital pathology images using deep learning
Aulia et al. A Novel Digitized Microscopic Images of ZN-Stained Sputum Smear and Its Classification Based on IUATLD Grades
Alimi et al. Quantitative Approach to Automated Diagnosis of Malaria from Giemsa-Thin Blood Stain using Support Vector Machine

Legal Events

Date Code Title Description
STCB Information on status: application discontinuation

Free format text: ABANDONED -- FAILURE TO RESPOND TO AN OFFICE ACTION