Predicting CYP2D6 phenotype from resting brain perfusion images by gradient boosting

详细信息    查看全文

• 作者：Giulio Napolitano^a ; ^{g.napolitano@imbie.uni-bonn.de} ; Julia C. Stingl^b ; Matthias Schmid^a ; Roberto Viviani^c
• 关键词：Machine learning ; Genetic polymorphism ; Image classification ; Enzymatic activity
• 刊名：Psychiatry Research: Neuroimaging
• 出版年：2017
• 出版时间：30 January 2017
• 年：2017
• 卷：259
• 期：Complete
• 页码：16-24
• 全文大小：891 K
• 卷排序：259

文摘

Component-wise gradient boosting is used to classify brain perfusion images. It is the first time component-wise gradient boosting is used to classify phenotypes from fMRI brain images. An extension is proposed to enhance plausibility without affecting performance. CYP2D6 poor and ultrarapid metabolisers are classified with good performance.