Abdominal CT image analysis and computa assisted diagnonsis with new artificial intelligence

2014 Fiscal Year Final Research Report

• Project/Area Number: 24560497
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: System engineering
• Research Institution: The University of Tokushima
• Principal Investigator: UENO Junji 徳島大学, ヘルスバイオサイエンス研究部, 教授 (60116788)
• Co-Investigator(Kenkyū-buntansha): KONDO Tadashi 徳島大学, 大学院ヘルスバイオサイエンス研究部, 教授 (80205559)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: 人工ニューラルネットワーク / GMDH / マルチスライスCT / コンピュータ支援診断

Outline of Final Research Achievements

In this study, new artificial neural network algorithms which are called as revised Group Method of Data Handling (GMDH)-type neural network algorithms, were proposed and we applied these algorithms to the medical image diagnosis of the liver cancer and the medical image recognition of abdominal multi-organs, and we developed the computer aided diagnosis (CAD) system using abdominal multi-detector row CT images. In this neural network algorithms, the principal component-regression analysis method is used to learn the weights of the neural networks and the neural network architectures are automatically organized using the prediction error criterion defined as Akaike’s Information Criterion (AIC) or Prediction Sum of Squares (PSS) so as to fit the complexity of the medical images. In this study, we applied new neural network algorithms to the medical image diagnosis of liver cancer and it was shown that new neural networks were useful to the medical image diagnosis of liver cancer.

Free Research Field

放射線診断学