Method for recognizing cell division event based on pooling time sequence characteristic representation
A cell division and identification method technology, applied in the field of cell division event identification based on pooled time series feature representation, can solve problems such as difficulty, difficulty in obtaining effective cell data, and inability to effectively describe cell differences, and achieve time domain information. Characterize fine-grained, broadly applicable effects
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0036] Studies have shown that: capturing the change information of the initial feature descriptor between each frame makes the research on the sequence more refined. The embodiment of the present invention proposes a cell division event recognition method based on pooled time series feature representation, see figure 1 , see the description below:
[0037] 101: In the sample database, extract sample-related features, and define the set of all sample features as an initial feature library;
[0038] 102: Each horizontal dimension of the initial feature matrix is a time series, and various pooling operators are applied to the time pyramid structure, and the pooled results are concatenated into a vector as the final representation of the sample;
[0039] 103: Calculate the kernel matrix of the training set and the test set respectively, apply the support vector machine as a classifier, and obtain the final prediction result.
[0040] Wherein, before step 101, the method for id...
Embodiment 2
[0053] The scheme in embodiment 1 is further introduced below in conjunction with specific calculation formulas and accompanying drawings, see the following description for details:
[0054] 201: Collect candidate cell subsequences to form a sample database;
[0055] In this method, the length of candidate subsequences is 23 frames, and the size of each frame is 50*50 pixels, and all candidate subsequences constitute a sample database. The data set C2C12 in the embodiment of the present invention is a population of osteoblast stem cells (ATTC, Manassas, VA), which can differentiate into osteoblasts and muscle cells.
[0056] figure 2 Given a frame of image samples, the extracted samples are segmented from many consecutive frames and stitched in time order. Each sequence in C2C12 contains 1013 images. After obtaining the images, biological researchers use the annotation tool with a graphical user interface to manually annotate cell division events in the image sequence. For...
Embodiment 3
[0093] Below in conjunction with concrete experimental data, accompanying drawing, the scheme in embodiment 1 and 2 is introduced in detail, see the following description for details:
[0094] The culture environment of the C2C12 data set is DMEM cell culture medium, supplemented with 10% bovine fetal serum, 1% penicillin and streptomycin, the ambient temperature is kept constant at 37°C, and the ambient carbon dioxide concentration is 5%. A Zeiss lens (ZeissAxiovert 135TV inverted microscope, 5X, 0.15N.A.) was used to capture a cell image every five minutes during the in vitro culture of stem cells, each image size was 1392×1040 pixels, and the resolution was 1.3 μm / pixel. C2C12 has a total of 16 sequences, and each sequence contains 1013 images.
[0095] The database used in this experiment is the candidate subsequence extracted by step 1). There are two categories, positive and negative. The positive category is a splitting event, and the negative category is not a splittin...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More