Development of Fast Image Reconstruction Method based on Machine Learning
| Project/Area Number |
15K15214
|
|---|
| Research Category |
Grant-in-Aid for Challenging Exploratory Research
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Medical Physics and Radiological Technology
|
|---|
| Research Institution | Hiroshima International University |
|---|
Principal Investigator |
Okura Yasuhiko 広島国際大学, 保健医療学部, 教授 (80369769)
|
|---|
| Project Period (FY) |
2015-04-01 – 2018-03-31
|
| Project Status |
Completed (Fiscal Year 2017)
|
| Budget Amount *help |
¥3,770,000 (Direct Cost: ¥2,900,000、Indirect Cost: ¥870,000)
Fiscal Year 2017: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Fiscal Year 2016: ¥130,000 (Direct Cost: ¥100,000、Indirect Cost: ¥30,000)
Fiscal Year 2015: ¥2,990,000 (Direct Cost: ¥2,300,000、Indirect Cost: ¥690,000)
|
|---|
| Keywords | 画像再構成 / 機械学習 / ニューラルネットワーク / 医学物理 / 人工知能 / 医用画像処理 / シミュレーション |
|---|
| Outline of Final Research Achievements |
In medical image diagnostic system such as X-ray CT, PET, which give important inspection in medical situation, an image reconstruction method to obtain useful information for diagnosis by constructing images as tomograms from projection data of the patients is a very important technology. However, especially in recent X-ray CT and PET, there are many information to be obtained, so time to calculations necessary for image reconstruction is too long even if newer computer is used for calculation. Therefore, it takes more waiting time to calculate is generated in clinical practice.On the other hand, it is known that the "large-scale neural network" has a relatively light computation load for "inference processing" of obtaining output by inputting data.
In this study, we clarified that high-speed image reconstruction in medical use can be realized by using large-scale neural network.
|
Report
(4 results)
Research Products
(7 results)
