Method and system for determining abnormality in medical device

Patent metadata

Field	Value
Publication number	US-12295753-B2
Application number	US-202217973672-A
Country	US
Kind code	B2
Filing date	Oct 26, 2022
Priority date	May 18, 2020
Publication date	May 13, 2025
Grant date	May 13, 2025

Abstract

A method for determining an abnormality in a medical device from a medical image is provided. The method for determining an abnormality in a medical device comprises receiving a medical image, and detecting information on at least a part of a target medical device included in the received medical image.

First claim

What is claimed is:

1. A method for determining an abnormality in a medical device in a medical image, the method being executed by at least one processor and comprising:
receiving a medical image;
detecting information on at least a part of a target medical device in the received medical image by:
extracting, from the received medical image, a fiducial marker associated with the target medical device; and
determining a normal area of the target medical device based on the extracted fiducial marker wherein the normal area is indicative of a region where the medical device is properly located; and

controlling a display to display information on a position of the at least the part of the target medical device and the normal area of the target medical device on the received medical image,
wherein the normal area is displayed on the received medical image in at least one form of a mask, region, contour, line or point.

2. The method according to claim 1 , wherein the detecting comprises detecting the information on the position of the at least the part of the target medical device in the received medical image by using a first machine learning model.

3. The method according to claim 2 , further comprising:
acquiring a plurality of reference medical images including one or more reference medical devices; and
acquiring an annotation for a position of at least a part of the one or more reference medical devices included in the plurality of reference medical images, wherein
the first machine learning model is trained to receive the plurality of reference medical images, and detect information on the one or more reference medical devices included in each of the plurality of reference medical images based on the annotation for the position of the at least the part of the one or more reference medical devices.

4. The method according to claim 2 , wherein the detecting the information on the position of the at least the part of the target medical device comprises:
determining whether or not the target medical device is included in the received medical image by using a second machine learning model; and
based on the target medical device being in the received medical image, detecting the information on the position of the at least the part of the target medical device in the received medical image by using the first machine learning model.

5. The method according to claim 4 , wherein the determining whether or not the target medical device is included in the received medical image by using the second machine learning model comprises determining whether or not a medical device included in the received medical image belongs to the same medical device group as the target medical device, and
the second machine learning model is trained to receive a plurality of reference medical images and output a medical device group to which a reference medical device included in each of the plurality of reference medical images belongs.

6. The method according to claim 1 , wherein the detecting comprises:
determining presence or absence of an abnormality in the target medical device based on the information on the target medical device and the extracted fiducial marker.

7. The method according to claim 6 , wherein the extracting comprises extracting, from the received medical image, the fiducial marker associated with the target medical device by using a third machine learning model.

8. The method according to claim 7 , further comprising:
acquiring a plurality of reference medical images including one or more reference medical devices; and
acquiring an annotation for a reference fiducial marker associated with the one or more reference medical devices included in the plurality of reference medical images, wherein
the third machine learning model is trained to receive the plurality of reference medical images, and extract reference fiducial markers associated with the one or more reference medical devices in the plurality of reference medical images based on the annotation for the reference fiducial marker associated with the one or more reference medical devices.

9. The method according to claim 6 , wherein the determining the presence or absence of the abnormality in the target medical device comprises
determining whether or not the at least the part of the target medical device is positioned in the normal area.

10. A non-transitory computer-readable recording medium storing instructions that, when executed by one or more processors, cause performance of the method according to claim 1 .

11. A method for determining an abnormality in a medical device in a medical image, the method being executed by at least one processor and comprising:
receiving a reference medical image;
determining a normal area associated with a reference medical device in the reference medical image;
generating a first set of training data in which at least a part of the reference medical device is placed in the determined normal area in the reference medical image;
generating a second set of training data in which the at least the part of the reference medical device is placed in an area other than the determined normal area in the reference medical image; and
training a fourth machine learning model for determining presence or absence of an abnormality in the reference medical device based on the first set of training data and the second set of training data.

12. The method according to claim 11 , further comprising:
receiving a medical image; and
determining the presence or absence of the abnormality in a target medical device included in the medical image by using the fourth machine learning model.

13. The method according to claim 11 , wherein the determining comprises:
receiving, from an external device, information on the normal area associated with a position of the at least the part of the reference medical device; and
applying the normal area associated with the position of the at least the part associated with the reference medical device to the reference medical image.

14. The method according to claim 11 , wherein the determining comprises:
receiving, from an external device, information on the reference medical device; and
extracting the normal area associated with the reference medical device in the reference medical image, based on the received information on the reference medical device and the information on the reference medical image.

15. The method according to claim 11 , wherein the fourth machine learning model comprises a binary classification model trained to classify the reference medical image into normal data or abnormal data.

16. An information processing system comprising:
memory storing one or more instructions; and
at least one processor configured to execute the stored one or more instructions to:
receive a medical image;
detect information on at least a part of a target medical device in the received medical image by:
extracting, from the received medical image, a fiducial marker associated with the target medical device; and
determining a normal area of the target medical device based on the extracted fiducial marker wherein the normal area is indicative of a region where the medical device is properly located; and

control a display to display information on a position of the at least the part of the target medical device and the normal area of the target medical device on the received medical image,
wherein the normal area is displayed on the received medical image in at least one form of a mask, region, contour, line or point.

17. The informat

Assignees

• Lunit Inc

Inventors

• Yoo Donggeun
• Lee Sanghyup
• Kim Minchul
• Lee Hanjun
• Park Sunggyun

Classifications

• G06T7/0012

  Biomedical image inspection · CPC title

• A61B6/5217

  extracting a diagnostic or physiological parameter from medical diagnostic data · CPC title

• A61B5/0033

  Features or image-related aspects of imaging apparatus, e.g. for MRI, optical tomography or impedance tomography apparatus; Arrangements of imaging apparatus in a room · CPC title

• A61B5/7264 Primary

  Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems · CPC title

• G06T2207/30061

  Lung · CPC title