本文へジャンプ

成果報告書詳細
管理番号20190000000137
タイトル*平成30年度中間年報 人工知能技術適用によるスマート社会の実現 健康、医療・介護分野 人工知能による脳卒中予防システムの開発・実用化
公開日2019/6/19
報告書年度2018 - 2018
委託先名学校法人慈恵大学東京慈恵会医科大学 学校法人東京理科大学 株式会社マックスネット
プロジェクト番号P18010
部署名ロボット・AI部
和文要約
英文要約Title:Development of Stroke Prediction System by Artificial Intelligence using Medical and Hemodynamic Data (FY2018-FY2019) FY2018 Annual Report

Stroke is a serious cerebrovascular condition, in which brain cells die due to an abrupt blockage of arteries supplying blood and oxygen or due to bleeding in the brain tissue when a blood vessel bursts or ruptures. Stroke incidences can be avoided by a preventive treatment based on the risk of onset. However, since the judgment of the onset risk largely depends on the individual experience and skill of the doctor, a highly accurate prediction method that is independent of the doctor’s experience and skill is the focus of this study. Our goal is to develop a new prediction system for stroke based on artificial intelligence (AI) engine. Toward this end, we first developed an aneurysm rupture prediction system considering medical data and engineering data, such as CFD (Computational Fluid Dynamics)-based hemodynamic data. In this year, we focused on the following research and development.

A. Data correction of aneurysm patients
For machine learning, we corrected medical data and engineering data from a total of 338 aneurysm cases, including 35 ruptured and 303 unruptured aneurysms. We have succeeded to extract 24 medical data and 53 engineering data for each aneurysm. The initial plan aimed at acquiring medical and engineering data for 150 cases until the end of March in 2019, but the result exceeded the target number of cases greatly.

B. Development of AI for aneurysmal rupture prediction
Some typical machine learning methods, such as logistic regression and support vector machine (SVM) were employed to analyze the 338 aneurysms. Out of the 28 important features extracted, 40% belonged to the medical data and the remaining 60% were from the hemodynamic data. Using logistic regression as a classification model, we found the sensitivity of 0.64 and the specificity of 0.85. The results validated the possibility of a highly accurate method of cerebral aneurysm rupture prediction by machine learning using engineering data obtained from mechanical simulation.

C. System development for aneurysmal rupture prediction
We developed a semi-automatically measurement system to obtain aneurysmal geometry data, such as aneurysm-height, -width, -volume, -surface area, neck diameter, and neck area, from DICOM-images. To perform CFD blood flow analysis in aneurysms, we also developed our original CFD solver based on a general LBM (Lattice Boltzmann Method)-based CFD software. As a result of comparing our CFD solver with commercial CFD solver, correlation coefficients of velocity and pressure were R = 0.985 and 0.913, respectively. Additionally, both the velocity and pressure errors on the aneurysm inlet and outlet plane were around 20%.
ダウンロード成果報告書データベース(ユーザ登録必須)から、ダウンロードしてください。

▲トップに戻る