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

A Transfer Learning Approach for Building Super-Resolution Pathology Microscopy

A technology of transfer learning and super-resolution, which is applied in the field of transfer learning to build super-resolution pathological microscopes, can solve problems such as difficulty in guaranteeing transfer effects and difficulties in the third domain, and achieve good transfer results

Active Publication Date: 2022-04-05
ZHEJIANG UNIV
View PDF10 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, it is very difficult to directly find a suitable third domain, and the transfer effect is difficult to guarantee

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
  • A Transfer Learning Approach for Building Super-Resolution Pathology Microscopy
  • A Transfer Learning Approach for Building Super-Resolution Pathology Microscopy
  • A Transfer Learning Approach for Building Super-Resolution Pathology Microscopy

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0037] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be noted that the following embodiments are intended to facilitate the understanding of the present invention, but do not limit it in any way.

[0038] Step 1, using a microscope to collect pathological image pairs.

[0039] Microscopes are used to collect pathological image pairs. During the collection process, high-resolution images and low-resolution images are taken under the same experimental conditions. After shooting, manual corrections are made to keep the content of the images exactly the same, only the difference in image resolution.

[0040] Step 2, divide the source data and target data, and perform normalization preprocessing on the data in the source domain and target domain.

[0041] The pathological image pair composition data set of the target pathological slice type of super-resolution pathological microscope is taken a...

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 present invention provides a migration learning method for constructing a super-resolution pathological microscope, comprising: (1) collecting pathological image pairs of different parts by using a microscope; As the target domain, the pathological image pair of non-target pathological slice type is used as the source domain, and the data of the target domain and the source domain are normalized; (3) the high and low resolution pathological images in each pathological image pair The slice photos are segmented according to a fixed ratio to obtain small-size pathological image pairs, and the segmented small-size pathological image pairs in the target domain are divided into training sets and test sets; (4) using the small-size pathological image pairs in the source domain and target domain (5) Construct a super-resolution model and perform training; (6) After the model is trained, apply the super-resolution model. By using the present invention, a better transfer learning effect can be achieved.

Description

technical field [0001] The invention belongs to the technical field of transfer learning, and in particular relates to a transfer learning method for constructing a super-resolution pathological microscope. Background technique [0002] Deep learning is currently applied to artificial intelligence super-resolution algorithms with good results. However, the use of deep learning models requires a large amount of data. In the case of less data, the effect of deep learning models is difficult to meet expectations. [0003] In super-resolution work, the input image and output image exist in the form of image pairs. Microscopic image super-resolution algorithms based on deep learning use low-resolution images as input to generate high-resolution images. The high-resolution image generated by the low-resolution is learned using the high-resolution image captured by the microscope as a label. However, deep learning requires fast construction and strong generalization, and data se...

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 Patents(China)
IPC IPC(8): G06T7/00G06T3/40G06N20/00
CPCG06T7/0012G06T3/4053G06N20/00G06T2207/10061G06T2207/20081G06T2207/30096
Inventor 吴健陈晋泰刘雪晨
Owner ZHEJIANG UNIV