Bayes integration of medical image with statistical mechanical approach

2015 Fiscal Year Final Research Report

• Project/Area Number: 25330285
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Soft computing
• Research Institution: The University of Electro-Communications
• Principal Investigator: Shouno Hayaru 電気通信大学, 情報理工学(系)研究科, 教授 (50263231)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: 情報統計力学 / ベイズ推定 / 医用画像 / ノイズ除去

Outline of Final Research Achievements

In this research, we apply a Bayes approach for medical image reconstruction problem. We develop several reconstruction algorithm, from theoretical and practical viewpoints. From the theoretical viewpoint, we solve a image reconstruction problem under the Poisson observation process with a Gaussian Markov random field prior. Also from the practical viewpoint, we apply our Bayes approach to reconstruct the positron emission tomography (PET) for the real device. As the result, we confirm the Bayesian approach is effective when we construct the model based on the fact. On that basis, we should consider several approximation for the effective calculation. The more difficult problem is to assume shape of the prior. Deciding the prior model is the our future work.

Free Research Field

ソフトコンピューティング