• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to project page

2020 Fiscal Year Final Research Report

Development of a Medical Image Reconstruction Method Using Deep Learning of Time Series Signals from Biometric Measurements

Research Project

  • PDF
Project/Area Number 18K18357
Research Category

Grant-in-Aid for Early-Career Scientists

Allocation TypeMulti-year Fund
Review Section Basic Section 90110:Biomedical engineering-related
Research InstitutionThe University of Tokyo

Principal Investigator

Tomii Naoki  東京大学, 大学院医学系研究科(医学部), 助教 (00803602)

Project Period (FY) 2018-04-01 – 2021-03-31
Keywords医用画像再構成 / 深層学習 / 超音波計測 / 心電図計測 / 不整脈
Outline of Final Research Achievements

This study aims to expand the application of simple biomedical measurement methods, such as ultrasound and bioelectrical measurements, to the precise diagnosis of various diseases, and to construct an image reconstruction method that is robust against inhomogeneities in the body by applying pattern recognition based on machine learning. As a result of training a deep neural network that reconstructs medical images from measurement signals with large-scale training data using numerical simulations for both ultrasound and ECG measurements, it was found that it is possible to reconstruct highly accurate medical images from limited measurement signals with higher precision than conventional methods.

Free Research Field

生体信号処理

Academic Significance and Societal Importance of the Research Achievements

本研究を通じて、深層学習によって従来より高画質な超音波計測が実現できる可能性が示された。これにより将来的に、現状では画質が限られる超音波画像診断をより精密な診断に応用できる可能性が開かれた。さらに、カテーテルを用いた心内心電図信号から、心臓内に発生する電気的興奮波を、興奮回復特性まで含めて従来よりも精密に可視化できる可能性が示された。これにより現状では治療の難しい複雑な不整脈に対し、正確な興奮状態の把握に基づく焼灼等の精密治療の可能性が開かれた。

URL: 

Published: 2022-01-27  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi