Spatial Transcriptomic Image Encoding for Efficient Clustering

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current data processing technologies for spatial transcriptomic data, such as clustering, consume excessive computing resources and are inefficient due to the need to encode spatial location information effectively.

Innovation Solution

A method involving generating a first image from spatial location and feature information, extracting a feature map, and performing processing directly on this image to encode spatial location information, thereby reducing computational load and improving efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If current clustering technology is used to process spatial transcriptomic data, then clustering accuracy can be achieved by considering spatial location information, but computing resource consumption is excessive and efficiency is low

Engineering Contradiction:
Improveclustering accuracyVSAvoidprocessing efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent segments the spatial transcriptomic data processing into distinct modules: an autoencoder for feature extraction, a graph neural network for spatial relationship modeling, and a clustering module. This segmentation allows each component to specialize in specific tasks, improving overall processing efficiency while maintaining clustering accuracy through coordinated operation of the segmented components.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary autoencoder that transforms high-dimensional gene expression data into compressed latent features before feeding them to the graph neural network. This intermediary representation reduces computational complexity and serves as a bridge between raw data and spatial clustering analysis, significantly lowering computing resource consumption while preserving essential information for accurate clustering.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If spatial location information is encoded using traditional methods, then accurate spatial representation is achieved, but computational load increases

Engineering Contradiction:
Improvespatial location encoding accuracyVSAvoidcomputational load
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent changes the parameter representation of spatial location information by transforming it into a graph structure where cells are nodes and spatial relationships are edges. The graph neural network processes this transformed representation, achieving accurate spatial encoding through message passing mechanisms while reducing computational load compared to traditional coordinate-based methods by leveraging the inherent structure of spatial relationships.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4376010B1Video processing method and apparatus, device, and storage medium
Publication Date: 2025.11.26 TENCENT TECHNOLOGY (SHENZHEN) CO LTD
  • EP4376010B1 patent drawingFigure 1~2
  • EP4376010B1 patent drawingFigure 3~4
  • EP4376010B1 patent drawingFigure 5~7

AI summary

The present application provides a data processing method and apparatus, a device, and a storage medium, which can be applied to the technical fields of artificial intelligence, bioengineering and the like. The method comprises: acquiring data to be processed, and parsing said data to obtain spatial position information and first feature information of each object in N objects; converting said data into a first image on the basis of the spatial position information and the first feature information of each object; from the first image, extracting second feature information of each object in the N objects; and performing preset processing on the second feature information of each object in the N objects to obtain a processing result of said data. According to the method, data to be processed is converted into a first image, the first image is subjected to feature extraction to obtain second feature information of N objects, and spatial position information is encoded into features; the whole data processing process is simple, a small quantity of computing resources are occupied, and the data processing efficiency is high.