Single-particle image processing method and device, electronic equipment and storage medium
By developing a single-particle image denoising model that is trained unsupervised on cryo-electron microscopy images, and utilizing CTF convolutional kernels and neural networks to process single-particle images, this model solves the problem of severe noise interference in existing technologies, enabling clearer image observation and more efficient protein structure analysis.
Patent Information
- Authority / Receiving Office
- CN Β· China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BIOMAP (BEIJING) INTELLIGENCE TECH LTD
- Filing Date
- 2023-04-21
- Publication Date
- 2026-07-24
AI Technical Summary
Existing single-particle image processing methods cannot effectively remove noise from cryo-electron microscopy images, making it difficult to clearly observe protein structure and morphology.
An unsupervised training method was adopted, using a set of single-particle images output by cryo-electron microscopy to train the single-particle image denoising model. Image denoising was performed using CTF convolutional kernels and neural networks, including encoding and decoding processes, to learn the three-dimensional structure of sample membrane proteins and generate clear two-dimensional images.
It significantly improves the clarity of cryo-electron microscopy images, enabling clearer observation of particle structure and morphology, and enhancing the effectiveness of single-particle analysis.
Smart Images

Figure CN116453115B_ABST