Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Fine adjustment method for cell segmentation

A cell and pre-training technology, which is applied in the fields of cell imaging, instance segmentation and model fine-tuning, can solve problems such as limitations and failure to keep up with the pace of general algorithm research, and achieve the effect of improving segmentation performance

Pending Publication Date: 2022-03-29
ZHEJIANG UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, these studies have not kept up with the pace of general algorithm research, and these studies are usually limited to cell nucleus images, and there are a large number of other examples such as nematodes that have segmentation needs

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
  • Fine adjustment method for cell segmentation
  • Fine adjustment method for cell segmentation
  • Fine adjustment method for cell segmentation

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0084] A cell segmentation fine-tuning method, which is divided into two stages of pre-training and fine-tuning, the specific steps are as follows:

[0085] Pre-training phase:

[0086] (1) First use the training set provided by Cellpose, and take the last of every 8 as a verification set to train the Scellseg model;

[0087] (2) Save the obtained model weight file for later use;

[0088] (3) Select a subset from the pre-training data as contrast data, including 7 styles, each style contains 5 pictures, such as Figure 4 shown.

[0089] Fine-tuning stage:

[0090] (1) From the image set generated by the new experiment, mark 10 images as the shot data set, and the remaining images can be used as the query data;

[0091] (2) The Scellseg model reads the pre-trained model weights, and uses shot data, query data, and contrast data to fine-tune the model;

[0092] (3) Save the fine-tuned model weights for future use;

[0093] (4) The fine-tuned model weights can be used for s...

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 discloses a cell segmentation fine tuning method, which comprises the following steps: S1, a pre-training stage: forming a pre-training set by taking known cell image data of different styles as pre-training data, and performing model training by utilizing the training set to obtain a pre-training model; s2, a fine tuning stage: marking at least 8 cell images from the to-be-segmented cell images of the new style as a shot data set, and taking the remaining cell images as a query data set; performing fine tuning and retraining on the pre-training model obtained in the step S1 by utilizing a shot data set to obtain a training model; and S3, an inference stage: inferring and segmenting the cell image in the query data set in the step S2 by using the training model obtained in the step S2. Compared with the prior art, the method has the advantages that cell segmentation of cell images of different styles can be realized, the application scene of a universal algorithm is widened, and the segmentation performance of the universal algorithm is improved.

Description

technical field [0001] The invention relates to the technical fields of cell imaging, instance segmentation and model fine-tuning, in particular to a cell segmentation fine-tuning method. Background technique [0002] Single cell segmentation technology is currently widely used in the fields of cell counting, spatial transcriptomics, high-throughput drug screening, and tumor metastasis detection. However, due to the lack of robust and easy-to-use cell segmentation algorithms, the method of average analysis is still the most commonly used method, which not only leads to information loss, but also may affect downstream analysis and misinterpret the relationship between features. Therefore, it is of great research value to develop a robust single-cell segmentation algorithm with excellent performance. [0003] However, since different cell types, microscopic imaging equipment, processing methods, imaging modes, and staining methods can produce various styles of cell images, it...

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): G06T7/10G06N3/04G06N3/08
CPCG06T7/10G06N3/084G06T2207/20081G06T2207/20084G06T2207/30024G06N3/048G06N3/045Y02T10/40
Inventor 王毅王锐荀德金
Owner ZHEJIANG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products