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

2022 Fiscal Year Final Research Report

Development of an AI-based freehand 3D system for a 2D ultrasound probe

Research Project

  • PDF
Project/Area Number 19K09624
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Review Section Basic Section 56020:Orthopedics-related
Research InstitutionOkayama University

Principal Investigator

Nakahara Ryuichi  岡山大学, 大学病院, 助教 (30509477)

Co-Investigator(Kenkyū-buntansha) 那須 義久  岡山大学, 大学病院, 助教 (30756101)
竹内 孔一  岡山大学, 自然科学学域, 准教授 (80311174)
Project Period (FY) 2019-04-01 – 2023-03-31
Keywords整形外科 / 超音波 / 人工知能
Outline of Final Research Achievements

To enhance the productivity of ultrasound examinations, we conducted a study to generate three-dimensional representations from images acquired by freehand movements of 2D ultrasound probes. We equipped the 2D probes with optical sensors to physically capture three-dimensional information and proceeded with three-dimensional reconstruction. Regardless of hardware improvements, minor relative shifts may occur, so we explored methods to make fine adjustments using artificial intelligence (AI). During this process, we discovered that it is possible to create pseudo-three-dimensional images using AI alone. We found that by utilizing a dimensionality reduction technique called UMAP (Uniform Manifold Approximation and Projection), we could learn the connections between images as topological spaces. This allows for the detection of identical cross-sections, leading to a significant reduction in examination time.

Free Research Field

整形外科

Academic Significance and Societal Importance of the Research Achievements

CTやMRIは3D画像として取得されるため、事後的に比較が可能であるため、撮像時に診断を行う必要がない。しかし超音波検査はCTやMRIと異なり三次元画像が残らず、また画像の質が検者の技量に依存するため、撮像現場で比較のために同一断面を描出する必要があり、それが検査時間の増大につながっていた。我々の手法は完全な3D化は困難であるが、疑似的な3D画像を通じて同一断面の検出が可能となるため、検査時間の短縮につながる。本手法を発展すれば、検査時間の短縮、ひいては医療費の削減や医療従事者の働き方改革につながる可能性がある。

URL: 

Published: 2024-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi