2021 Fiscal Year Final Research Report
Development of videofluoroscopic swallowing studies diagnostic assist system using machine learning
Project/Area Number |
18H03017
|
Research Category |
Grant-in-Aid for Scientific Research (B)
|
Allocation Type | Single-year Grants |
Section | 一般 |
Review Section |
Basic Section 57080:Social dentistry-related
|
Research Institution | The Nippon Dental University |
Principal Investigator |
|
Co-Investigator(Kenkyū-buntansha) |
川口 孝泰 東京情報大学, 看護学部, 教授 (40214613)
井出 吉昭 日本歯科大学, 生命歯学部, 准教授 (70409225)
|
Project Period (FY) |
2018-04-01 – 2021-03-31
|
Keywords | 嚥下造影検査 / 誤嚥 / 嚥下障害 / 人工知能(AI) / 特徴量 / 画像解析 / 嚥下動態 / サルコペニア |
Outline of Final Research Achievements |
This study attempts to automate the diagnosis of pathological conditions observed by videofluoroscopic swallowing studies(VESS) using image analysis technology and machine learning. As a result, it was possible to detect organic changes in the pharynx and dynamic changes in pharyngeal function in patients with sarcopenia. In order to perform machine learning of swallowing dynamics, it is necessary to automatically extract feature points from moving images, and then machine learning these feature points to improve diagnostic efficiency.When machine learning was implemented, the accuracy of aspiration was 70.5%. On the other hand, there was an error in detecting or predicting the position of the feature point directly from the VESS movingimage. Based on the above, further studies were needed to apply the image acquisition technology and diagnostic efficiency obtained in this study to clinical applications.
|
Free Research Field |
高齢者歯科学
|
Academic Significance and Societal Importance of the Research Achievements |
画像取得技術、診断効率は、臨床応用に適応するにはさらなる検討が必要であったものの、一定レベルまで自動データ収集と診断効率を得ることが可能であることを示せた意義は大きい。患者が嚥下する際に体幹が動くことや膨大なデータ量を自動追尾する方法の課題は明らかになった。嚥下造影検査は嚥下機能評価におけるゴールドスタンダードといわれているが、診断手順が煩雑であることから臨床現場では必要な検査が行われない場合が多く、患者にとって不利益が生じている現状である。本技術の確立は、医療現場における診断の効率化を通じて患者に大きな福音となる。
|