Method and system for analyzing image

Inventors

JANG, Minje

Assignees

Lunit Inc

Interested in licensing this patent?

MTEC can help explore whether this patent might be available for licensing for your application.

Publication Number

US-12664650-B2

Patent

Publication Date

2026-06-23

Expiration Date


Abstract

An image analysis method and an image analysis system are disclosed. The method may include extracting training raw graphic data including at least one first node corresponding to a plurality of histological features of a training tissue slide image, and at least one first edge defined by a relationship between the histological features and generating training graphic data by sampling the first node of the training raw graphic data. The method may also include determining a parameter of a readout function by training a graph neural network (GNN) using the training graphic data and training output data corresponding to the training graphic data, and extracting inference graphic data including at least one second node corresponding to a plurality of histological features of an inference tissue slide image, and at least one second edge decided by a relationship between the histological features of the inference tissue slide image.

Core Innovation

The invention provides an image-analysis method and an image-analysis system for whole-slide histology that uses a graph neural network (GNN). A tissue slide image of a patient is used to identify at least one node, and class information on histological features is used to determine a region including the at least one node. Graph data are generated based on the region including the at least one node, and inference output data are produced from the graph data through the GNN as a medical prediction for the tissue slide image.

The method updates first feature information of the at least one node based on the first feature information of the at least one node and second feature information of at least one contiguous node. In this way, node feature information is iteratively refined in the graph neural network processing by using information from contiguous nodes, while the medical prediction is output from the graph data.

The disclosed approach emphasizes forming graph data from histological content rather than pixel-wise processing, and it includes region-based node and edge reduction through sampling to reduce graph size. It further includes multiple sampled inference graphs and merges pre-inference output data to improve robustness and accuracy, while maintaining a readout function that outputs the inference predictions from the GNN.

Claims Coverage

The partial claims include two independent claims: a method claim and a system claim. Each independent claim contains the same core set of inventive features, including node identification from a tissue slide image, region determination from class information, graph data generation, medical prediction via a GNN, and updating node features using contiguous-node feature information.

Node identification from patient tissue slide image

Identifying at least one node based on a tissue slide image of a patient.

Region determination from class information on histological features

Determining a region including the at least one node based on class information on histological features of the tissue slide image.

Graph data generation from region nodes

Generating graph data based on the region including the at least one node.

Medical prediction from graph data via GNN

Outputting a medical prediction for the tissue slide image, which is inference output data obtained from the graph data through a graph neural network (GNN).

Node feature updating using contiguous nodes

Updating first feature information of the at least one node based on the first feature information of the at least one node and second feature information of at least one contiguous node.

System configured to perform node, region, graph, prediction, and updating

An image analysis system comprising at least one memory and at least one processor configured to identify at least one node based on a tissue slide image of a patient, determine a region including the at least one node based on class information on histological features of the tissue slide image, generate graph data based on the region including the at least one node, output a medical prediction for the tissue slide image obtained through a graph neural network (GNN), and update first feature information of the at least one node based on first feature information of the at least one node and second feature information of at least one contiguous node.

Across the independent claims, the inventive coverage centers on converting patient tissue-slide histological information into graph data via node identification and region determination from class information, applying a GNN to output inference-based medical prediction, and updating node features using second feature information from contiguous nodes.

Stated Advantages

Documented Applications

No documented applications found

JOIN OUR MAILING LIST

Stay Connected with MTEC

Keep up with active and upcoming solicitations, MTEC news and other valuable information.