Syoji Kobashi

Professor, University of Hyogo

    Syoji Kobashi received Bachelor of engineering in 1995, Master of engineering in 1997, and Doctor of Engineering in 2000, all from Himeji institute of Technology, JAPAN. He was an assistant professor at Himeji Institute of Technology from 2000 to 2004, and am currentlyan associate professor at University of Hyogo, JAPAN, since 2005, and a guest associate professor atOsaka University,JAPAN, WPI immunology frontier research center since 2011. He was a visiting scholar at Department of Radiology, University of Pennsylvania, USA, in 2011-2012. His research interests include medical image understanding and analysis. He published 60 journal papers, 300 conference proceeding papers, and some invited papers.His researcheshave been supported in part by Grants-in-Aid for Scientific Research(KAKENHI), MEXT, JAPAN. He received 14 international awards, including Franklin V. Taylor Memorial Award (IEEE-SMCS, 2009), IEEE-EMBS Japan Young Investigators Competition (EMBS Japan Chapter, 2003). He has been serving on the chair of International Forum on Multimedia and Image Processing in 2012, 2014 and 2016, General co-chair of International conference on informatics, electronics & Vision (ICIEV) in 2016, and other. Moreover, he isa regional editor of intelligent automation &soft computing, and an associate editor of 3 journals including International Journal of Intelligent Computing in Medical Sciences and Image Processing, and a guest editor of some special issues. And, he is organizing many special sessions in international conferences including IEEE SMC and IEEE EMBC. He is the track force chair of IEEE CIS task force on fuzzy logic in medical sciences. He is the senior member of IEEE.

Brain Morphometry Based Developmental RetardationQuantification in Neonatal MRI

    Quantification of brain development retardation is important for diagnosing neonatal cerebral disorders. This study proposes a computer-aided method for estimating brain growth age based on brain morphometry change using T2-weighted MR images.The brain morphometry is characterized using anatomical predetermined landmarks. The estimation system is trained by using a learning dataset.It first applies principal component analysis (PCA)to a set of distances between all combinations of landmarks, and then derive a regression model whose dependent variable is revised age, and independent variables are PC scores.To estimate the brain age of a new subject, it obtains the feature vector as well as the learning procedure, transfers into PC scores using eigenvectors, and then estimates the brain age using the trained regression model.The proposed method was applied to 15 neonates (revised age was 33.7±40.7 days) with normal development and 4 neonates with abnormal development. The error for normalneonates are 15.5±11.8 days in training data and 46.3±37.2 in evaluation data, and the estimated brain age of abnormalsubject was different from their real age.It shows that the method could differentiate the brain growth age between normal and abnormal growing. The most advantage of this method is that it can be easily installed in clinical routine work because it only requires some landmarks.


English is the official language of the conference ( Paper and Presentation)Copyright icetet.org