2021 Fiscal Year Final Research Report
Development of 3D-DSA using by Deep Learning
| Project/Area Number |
17K18291
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Medical Physics and Radiological Technology
Radiation science
|
|---|
| Research Institution | Hiroshima International University |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2017-04-01 – 2022-03-31
|
| Keywords | 血管造影 / DSA / 深層学習 |
|---|
| Outline of Final Research Achievements |
DSA (Digital Subtraction Angiography) technique is used for cerebral angiography and endovascular treatment (IVR). It provides high-definition images and enables diagnosis of blood vessel diseases from various angles of patients. However, DSA has two problems. (1) Since,it is extremely sensitive to the movement of the examinee and it is resulted in comparatively large artifacts, the organ to apply DSA is limited. (2) Increased radiation dose and longer examination time due to before contrast imaging for mask image acquisition. The purpose of this study was to develop a new DSA method that solves the above two problems using deep learning based technique.
In developed method, mask image acquisition is not required. Moreover, artifacts of DSA image are not visualized when patient or organ is moved. Stopping breathing of examinee is no longer needed at DSA study by developed method.
|
| Free Research Field |
医用画像処理
|
| Academic Significance and Societal Importance of the Research Achievements |
本研究ではDSAのマスク画像作成に，深層学習の一種であるCNNを使用する．本研究ではライブ画像からマスク画像を作り出す画像処理に直接，深層学習を使用した．これにより造影前のマスク画像取得が不要になるため，任意の角度でDSAが作成でき被曝と検査時間の短縮に繋がる．本研究成果は，DSAの適用部位を限定することなく，通常の血管造影像やDSAでは観察できなかったあらゆる部位の微細血管までを，立体的に様々な角度から把握することができ，従来の血管造影では明らかにならなかった血管の形態や質，内膜や血栓・石灰化などの三次元的位置を把握でき，診断や手術において非常に有用である，
|
