Unlock instant, AI-driven research and patent intelligence for your innovation.

Multi-instance learner for organizing image classification

A tissue image and image analysis technology, applied in the field of digital pathology and image analysis

Pending Publication Date: 2022-07-08
F HOFFMANN LA ROCHE & CO AG
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] However, Applicants have observed that various machine learning techniques that provide good results for the early detection of cancer-associated nodules in mammography images fail to perform well on images of other types of tissue sections, particularly full-field-bearing images. Slide images for classification
[0004] Another problem associated with image classification using existing machine learning methods is that the trained machine learning program is often like a black box

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Multi-instance learner for organizing image classification
  • Multi-instance learner for organizing image classification
  • Multi-instance learner for organizing image classification

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0265] figure 1 A method flow diagram according to an embodiment of the present invention is shown. This method can be used to classify tissue images of patients. For example, the classification can be performed to predict patient attribute values ​​such as, for example, biomarker status, diagnosis, treatment outcome, microsatellite status (MSS) for a particular cancer (such as colorectal or breast cancer), lymph node micrometastases, and Pathological complete response (pCR) in diagnostic biopsy. The prediction is based on classification of digital images of histological slides using a trained MIL procedure that takes into account model uncertainty. exist figure 1 In the following description of the figure 2 , 15 and 16 elements.

[0266]The method 100 can be used to identify hitherto unknown predictive histological features and / or to classify tissue samples with high accuracy.

[0267] In a first step 102, the image analysis system 200 (eg, refer to figure 2 descr...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to a method for classifying tissue images. The method comprises:-receiving (102) a plurality of digital tissue images; -splitting (104) each received image into a set of image blocks; -for each of said blocks, extracting (106) a feature vector from said block; providing (108) a multi-instance learning (MIL) program configured to use a model to classify the input image as members of one of at least two different categories based on feature vectors extracted from all blocks of any image; calculating (110), for each of the blocks, a certainty value indicative of a certainty of the model with respect to a contribution of a feature vector of the block to a classification of the image; for each of the images, aggregating a feature vector of the image or a predictor calculated from the feature vector of the image into an aggregated predictor by the MIL program using (114) a pooling function based on deterministic values as a function of the deterministic values of the block of the image; and-classifying (116) each of the images as a member of one of the categories based on the aggregated predictors.

Description

technical field [0001] The present invention relates to the field of digital pathology, and more particularly to the field of image analysis. Background technique [0002] Several image classification methods are known for classifying digital pathology images into different categories, such as "healthy tissue" or "cancerous tissue". For example, Sertan Kaymaka et al. in "Breast cancer imageclassification using artificial neural networks", Procedia Computer Science, Vol. 120, 2017, pp. 126-131, describe a method using backpropagation neural networks (BPPN) Methods for automatic classification of breast cancer diagnostic images. [0003] Applicants have observed, however, that various machine learning techniques that provide good results for early detection of cancer-related nodules in mammography images fail to perform well on images of other types of tissue sections, particularly full-field Slide images are classified. [0004] Another problem associated with using existi...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06T7/00
CPCG16H50/70G16H30/40G16H30/20G06V20/695G06V20/698G06V10/454G06V10/82G06F18/2414G06F18/214G16H50/20G06V10/7715G06V10/764G06N3/084G06T7/0012G06T2200/24G06T2207/20081G06T2207/20084G06T2207/30024
Inventor E·克莱曼J·吉尔登布拉特I·B·肖尔
Owner F HOFFMANN LA ROCHE & CO AG