2019 Fiscal Year Final Research Report
3D Model Simulation of Paralyzed Face and Quantitative Evaluation of Facial Paralysis
Project/Area Number |
17K18243
|
Research Category |
Grant-in-Aid for Young Scientists (B)
|
Allocation Type | Multi-year Fund |
Research Field |
Intelligent informatics
Perceptual information processing
|
Research Institution | Ritsumeikan University |
Principal Investigator |
Seo Masataka 立命館大学, 情報理工学部, 講師 (60725943)
|
Project Period (FY) |
2017-04-01 – 2020-03-31
|
Keywords | Deep Learning / VR / 顔面神経麻痺 / 診断支援 / 人工知能 |
Outline of Final Research Achievements |
The following three research results have been created mainly for the purpose of supporting the diagnosis of facial nerve palsy. (1) Multi-view facial paralysis facial expression database : The database was constructed with the cooperation of 45 patients with facial palsy and 9 healthy subjects. (2) Construction of 3D statistical facial model and facial paralysis simulation : We have built a facial paralysis facial simulator in VR environment using Kinect. Various facial nerve palsy can be simulated on a three-dimensional face created by synthesizing the faces of multiple healthy persons. (3) Creation of facial paralysis score discriminator : We constructed a facial paralysis score discriminator using deep learning. We realized the discrimination with almost the same accuracy as the evaluation by the otolaryngologist.
|
Free Research Field |
人工知能、最適化
|
Academic Significance and Societal Importance of the Research Achievements |
疾患を抱えた顔および頭部の三次元形状表現は計算機診断支援の最先端技術の一つである。本研究では、個人ごとに表情変化動作が異なるために一般的な統計解析手法では再現が困難とされている顔面神経麻痺顔の三次元表情モデル生成を世界に先駆けて実現させた。三次元の顔面神経麻痺顔モデルを作成することで、治療計画立案にも大きく貢献できるとともに、若手医師の学習にも寄与すると考えられる。 また、本研究ではDeep Learningを用いた顔面神経麻痺スコア判別器を構築した。今後は医療機関協力の下、三次元表情モデルとともに診断の現場で有効性の確認を行う予定である。
|