3D medical image processing using 3D-ACTIT : Automatic Construction of Tree-structural Image Transformation (Computer Vision, Medical Applications and Networked MM)(International Workshop On Advanced Image Technology (IWAIT2004))

概要

論文の詳細を見る
In the field of medical image processing, 3D(three dimensional) image processing is often required to deal with MRI, CT and PET image data. In general, however, it is not easy to construct complex 3D processing procedures manually compared with 2D ones. Thus, we propose a new method named "3D-ACTIT; Automatic Construction of Tree-structural Image Transformation" to make various 3D image processing procedures based upon example learning. ACTIT is the system that we previously proposed for automatic construction of 2D image processing [1][2]. ACTIT learns from several image processing example sets of original images and their target ones, and it approximates a target image processing with a tree whose non-terminals and terminals are typical image filters and input images, respectively according to Genetic Programming. In this paper, we improve the ordinary ACTIT to 3D-ACTIT and apply it to several 3D test data and real PET image data. Experimental results show that various kinds of complex 3D image processing procedures can be constructed automatically by the proposed method.
社団法人電子情報通信学会の論文
2004-01-06