Analysis of Reading Function, Preservation and Plasticity at the Temporoparietal Junction in Awake Brain Surgery

2020 Fiscal Year Final Research Report

• Project/Area Number: 19K18407
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 56010:Neurosurgery-related
• Research Institution: Fujita Health University
• Principal Investigator: MUTO Jun 藤田医科大学, 医学部, 講師 (30383839)
• Project Period (FY): 2019-04-01 – 2021-03-31
• Keywords: 白質解析 / 脳機能 / 脳腫瘍 / ベイズ深層学習モデル

Outline of Final Research Achievements

We evaluated the existing neural basis models of language, especially the picture naming task, by combining VBM analysis using cortical and white matter fiber mapping data from craniotomies of brain tumor patients, preoperative and postoperative MRI data, and the results of neurological function tests at different time points, as well as Bayesian deep learning model was used to reproduce the results. The possibility of using deep learning models to quantitatively deal with the visual features of images and the semantic features of language was investigated, and the relationship between the semantic features of the vocabulary produced by the aphasic patients and the pictorial illustrations was examined.
It is now possible to provide quantitative information on semantic and visual illocutionary errors in aphasia.

Free Research Field

脳機能

Academic Significance and Societal Importance of the Research Achievements

現在、臨床現場で用いられている神経心理試験を用いて、その脳内機構を明らかにすることで、脳内の障害部位と神経心理試験の結果をベイズ深層学習モデルで再現し、脳内ネットワークの解明に寄与するという学術的意義がある。さらに、損傷から、リハビリを行い、回復過程を追うことで、脳内の白質繊維の変化と神経心理試験の結果を合わせて解釈し、
ベイズ深層学習モデルで再現を試みることで、リハビリテーションにも寄与するという意義があると考える。