2022 Fiscal Year Final Research Report
Tissue diagnosis using quantitative phase imaging: drastic improvement in throughput and recognition accuracy by AI
| Project/Area Number |
20K05362
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 30020:Optical engineering and photon science-related
|
|---|
| Research Institution | Kyushu Institute of Technology |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | 定量位相イメージング / 組織診断 / 分類学習器 / 深層学習 / AI |
|---|
| Outline of Final Research Achievements |
We studied whether AI technologies improve throughput and recognition accuracy in tissue diagnosis using quantitative phase images. We obtained the result that deep neural network used for image resolution enhancement for general images is effective in improving the resolution of quantitative phase images. However, the improvement was marginal and further study is needed in terms of network models and data sets. We also tested breast tissue diagnoisis using multiple markers extracted from quantitative phase images with a support vector machine. A higher recognition rate can be achieved than by using a single marker. In addition, a detailed analysis was performed using principal component analysis to visualize the contribution of each marker.
|
| Free Research Field |
光情報工学
|
| Academic Significance and Societal Importance of the Research Achievements |
本研究によって，定量位相イメージングを用いた組織診断技術とAI技術の融合についての基礎的な知見が得られたことに意義がある．定量位相イメージングを用いた病理診断は従来の病理診断の多くの手間や曖昧さを排除することに貢献するとされるが，AIとの融合によりさらに技術導入への障壁が下がり，発展途上国なども含め，世界中に当該技術が広く普及するきっかけとなることが期待される．
|
