Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
METHOD AND DEVICE FOR DECIPHERING LESION FROM CAPSULE ENDOSCOPE IMAGE BY USING NEURAL NETWORK
Document Type and Number:
WIPO Patent Application WO/2019/168280
Kind Code:
A1
Abstract:
A neural network learned using a first color space and a second color space is applied such that only a part of images, in which a lesion is suspected to have occurred, can be extracted from among a plurality of captured images from a capsule endoscope image, and the corresponding image can be tagged with information on the lesion. Therefore, a doctor deciphers only a part of the images, extracted by a lesion deciphering device, without deciphering all of the plurality of capsule endoscope images, thereby enabling costs and time to be reduced. Image patterns of frames shown before and after a frame in which the lesion is most clearly seen can be used to decipher the lesion in the frame in which the lesion is most clearly seen. In addition, the frame, which corresponds to the position at which the lesion is suspected is identified by another medical image diagnosis method, among the frames having been received from a capsule endoscope is analyzed so as to enable the deciphering efficiency and accuracy of the lesion to increase.

Inventors:
MOON CHANG MO (KR)
CHOI JANG HWAN (KR)
Application Number:
PCT/KR2019/001600
Publication Date:
September 06, 2019
Filing Date:
February 11, 2019
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
UNIV EWHA IND COLLABORATION (KR)
International Classes:
G16H50/20; A61B1/04; G06T7/30; G16H30/20
Foreign References:
KR20160058812A2016-05-25
KR20170096924A2017-08-25
KR20140018748A2014-02-13
JPH10243247A1998-09-11
KR20140047277A2014-04-22
Attorney, Agent or Firm:
MUHANN PATENT & LAW FIRM (KR)
Download PDF: