US 11,967,136 B2
Landmark detection using machine learning techniques
Shanhui Sun, Lexington, MA (US); Yikang Liu, Cambridge, MA (US); Xiao Chen, Cambridge, MA (US); Zhang Chen, Brookline, MA (US); and Terrence Chen, Lexington, MA (US)
Assigned to Shanghai United Imaging Intelligence Co., Ltd., Shanghai (CN)
Filed by Shanghai United Imaging Intelligence Co., Ltd., Shanghai (CN)
Filed on Dec. 21, 2021, as Appl. No. 17/557,984.
Prior Publication US 2023/0196742 A1, Jun. 22, 2023
Int. Cl. G06V 10/774 (2022.01); G06T 7/00 (2017.01); G06V 10/82 (2022.01)
CPC G06V 10/7747 (2022.01) [G06T 7/0012 (2013.01); G06V 10/82 (2022.01); G06T 2207/30004 (2013.01)] 20 Claims
OG exemplary drawing
 
1. An apparatus, comprising:
one or more processors configured to:
determine a plurality of hypothetical landmarks in one or more medical images of an anatomical structure;
determine a graph representation of the plurality of hypothetical landmarks, wherein the graph representation includes multiple nodes and multiple edges, each of the nodes represents a hypothetical landmark among the plurality of hypothetical landmarks, and each of the edges represents a relationship between a pair of hypothetical landmarks among the plurality of hypothetical landmarks; and
identify, using a graph neural network (GNN), one or more hypothetical landmarks among the plurality of hypothetical landmarks as true landmarks or one or more hypothetical landmarks among the plurality of hypothetical landmarks as false landmarks, wherein the GNN includes:
an encoder network trained to extract respective features from the nodes of the graph representation and the edges of the graph representation;
a core network trained to estimate respective states of the nodes and edges of the graph representation based on the features extracted by the encoder network; and
a decoder network trained to indicate the true landmarks or the false landmarks based on the respective states of the nodes and edges estimated by the core network.