| Project/Area Number |
16K19869
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Radiation science
|
|---|
| Research Institution | Akita Cerebrospinal and Cardiovascular Center |
|---|
Principal Investigator |
MATSUBARA Keisuke 秋田県立循環器・脳脊髄センター(研究所), 放射線医学研究部, 研究員 (40588430)
|
|---|
| Research Collaborator |
TAKAHASHI Noriyuki
SHINOHARA Yuki
UMETSU Atsushi
IBARAKI Masanobu
KINOSHITA Toshibumi
|
|---|
| Project Period (FY) |
2016-04-01 – 2019-03-31
|
| Project Status |
Completed (Fiscal Year 2018)
|
| Budget Amount *help |
¥3,900,000 (Direct Cost: ¥3,000,000、Indirect Cost: ¥900,000)
Fiscal Year 2018: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2017: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2016: ¥1,950,000 (Direct Cost: ¥1,500,000、Indirect Cost: ¥450,000)
|
|---|
| Keywords | 機械学習 / 深層学習 / PET / MRI / 脳卒中 / 磁化率強調像 / 脳循環代謝 / 画像診断 |
|---|
| Outline of Final Research Achievements |
We aimed to classify dysfunction of cerebral metabolism in cerebrovascular disease by supervised learning of susceptibility weighted-image acquired from magnetic resonance imaging (MRI). Convolutional neural network (CNN) succeeded to classify the dysfunction for validation data by 97.0% accuracy. However, very low accuracy (61.7%) was observed in test data, which was used in training. These results suggest that the learning of CNN with small data size resulted in overfitting to the training data. Further study with much large data size is required.
|
| Academic Significance and Societal Importance of the Research Achievements |
脳血管の狭窄・閉塞を伴う虚血状態の診断及びそれに対する手術適応の決定において，脳循環代謝機能を測定し，異常所見を捉えることが重要となる．MRI装置で撮像された磁化率強調像(SWI)は脳循環代謝の異常に伴う静脈増強所見を画像上で捉えることができるが，わずかな所見の変化を捉えるには経験を要する．
本研究で目標としたSWIから脳循環代謝異常を予測する分類器は，SWIの読影を支援し得るものであり，実現されれば虚血の診断の精度向上に寄与しうるものである．本研究では残念ながら正確な分類器の作成までに至らなかったが，訓練データ数を増やすことで正確な分類器を作成できる可能性が示された．
|
