Computer system for trabecular connectivity recovery of skeletal images reconstructed by artificial neural network through node-link graph-based bone microstructure representation, and method thereof

Patent metadata

Field	Value
Publication number	US-12175665-B2
Application number	US-202217573762-A
Country	US
Kind code	B2
Filing date	Jan 12, 2022
Priority date	Oct 15, 2021
Publication date	Dec 24, 2024
Grant date	Dec 24, 2024

Abstract

Various embodiments relate to a computer apparatus for the bone microstructure connectivity recovery of a skeletal image reconstructed through an artificial neural network using the representations of a node-link graph-based bone microstructure and a method thereof. The computer apparatus and the method may be configured to represent a node-link graph from a bone microstructure of an input skeletal image, reinforce a connectivity of the bone microstructure in the node-link graph, and change the node-link graph into a skeletal image.

First claim

The embodiments of the disclosure in which an exclusive property or privilege is claimed are defined as follows:

1. A method of a computer apparatus, comprising:
representing a node-link graph from a bone microstructure of an input skeletal image;
reinforcing a connectivity of the bone microstructure in the node-link graph; and
changing the node-link graph into a skeletal image,
wherein the changing of the node-link graph into the skeletal image comprises:
searching for bone mineral density (BMD) of each of the plurality of links of the node-link graph; and
representing a thickness of each of trabeculae by using the BMD while representing the trabeculae corresponding to the links, respectively, and
wherein the searching for the BMD comprises searching for BMD of a central portion of a mask, while moving, in a direction perpendicular to each link, the mask having a shape identical with a shape of a link element composed of pixels belonging to the link in the inputted skeletal image.

2. The method of claim 1 , wherein the representing of the node-link graph comprises:
representing trabeculae of the bone microstructure as a plurality of links; and
representing, as a plurality of nodes, points at which the links are connected and an open end of at least one of the links.

3. The method of claim 2 , wherein the representing of the node-link graph further comprises:
obtaining a binarization image by performing image binarization on the inputted skeletal image; and
obtaining a centerline image by performing centerline extraction in the binarization image,
wherein the links are represented based on the centerline image.

4. The method of claim 1 , wherein the computer apparatus reconstructs the skeletal image from the inputted skeletal image reconstructed through an artificial neural network.

5. The method of claim 1 , wherein the reinforcing of the connectivity of the bone microstructure comprises moving a location of an open node to a location of an adjacent element in the node-link graph, and the adjacent element is an adjacent link in the node-link graph, an adjacent node in the node-link graph, or a boundary of the node-link graph.

6. The method of claim 5 , wherein the open node is a node having a node degree of 1.

7. The method of claim 1 , wherein the searching for the BMD comprises:
if pixels adjacent to the link element are present as a plurality of layers,
searching for BMD of a current layer;
searching for BMD of a next layer while setting the BMD of the current layer when the BMD of the current layer is greater than a predetermined value; and
excluding the BMD of the current layer when the BMD of the current layer is equal to or smaller than the predetermined value.

8. A computer apparatus comprising:
a memory; and
a processor connected to the memory and configured to execute at least one instruction stored in the memory,
wherein the processor is configured to:
represent a node-link graph from a bone microstructure of an input skeletal image,
reinforce a connectivity of the bone microstructure in the node-link graph, and
change the node-link graph into a skeletal image,
wherein the processor is configured to:
search for bone mineral density (BMD) of each of the plurality of links of the node-link graph, and
represent a thickness of each of trabeculae by using the BMD while representing the trabeculae corresponding to the links, respectively, and
wherein the processor is configured to search for BMD of a central portion of a mask, while moving, in a direction perpendicular to each link, the mask having a shape identical with a shape of a link element composed of pixels belonging to the link in the inputted skeletal image.

9. The computing apparatus of claim 8 , wherein the processor is configured to:
represent trabeculae of the bone microstructure as a plurality of links, and
represent, as a plurality of nodes, points at which the links are connected and an open end of at least one of the links.

10. The computing apparatus of claim 9 , wherein the processor is configured to:
obtain a binarization image by performing image binarization on the inputted skeletal image,
obtain a centerline image by performing centerline extraction in the binarization image, and
represent the links and the nodes based on the centerline image.

11. The computing apparatus of claim 8 , wherein the processor is configured to reconstruct the skeletal image from the inputted skeletal image reconstructed through an artificial neural network.

12. The computing apparatus of claim 8 , wherein the processor is configured to move a location of an open node to a location of an adjacent element in the node-link graph, and the adjacent element is an adjacent link in the node-link graph, an adjacent node in the node-link graph, or a boundary of the node-link graph.

13. The computing apparatus of claim 8 , wherein the processor is configured to:
if pixels adjacent to the link element are present as a plurality of layers,
search for BMD of a current layer,
search for BMD of a next layer while setting the BMD of the current layer when the BMD of the current layer is greater than a predetermined value, and
exclude the BMD of the current layer when the BMD of the current layer is equal to or smaller than the predetermined value.

14. A non-transitory computer-readable recording medium on which one or more programs for executing a method in a computer apparatus are recorded, the method comprising:
representing a node-link graph from a bone microstructure of an input skeletal image;
reinforcing a connectivity of the bone microstructure in the node-link graph; and
changing the node-link graph into a skeletal image,
wherein the changing of the node-link graph into the skeletal image comprises:
searching for bone mineral density (BMD) of each of the plurality of links of the node-link graph; and
representing a thickness of each of trabeculae by using the BMD while representing the trabeculae corresponding to the links, respectively, and
wherein the searching for the BMD comprises searching for BMD of a central portion of a mask, while moving, in a direction perpendicular to each link, the mask having a shape identical with a shape of a link element composed of pixels belonging to the link in the inputted skeletal image.

Assignees

• Korea Advanced Inst Sci & Tech

Inventors

• Jang Ingwun
• Koh Hyukjin
• Chun Bongju

Classifications

• G06T2207/30008

  Bone · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/10056

  Microscopic image · CPC title

• G06N3/02

  Neural networks · CPC title

• G06V10/457

  by analysing connectivity, e.g. edge linking, connected component analysis or slices · CPC title