Localization and Alignment of Non-rigid Organs by Exhaustive Search of Position, Orientation,Scale and Deformation Parameters

2016 Fiscal Year Final Research Report

• Project/Area Number: 26330212
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Perceptual information processing
• Research Institution: Ritsumeikan University
• Principal Investigator: Xu Gang 立命館大学, 情報理工学部, 教授 (90226374)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: 画像認識 / 医用画像処理 / 3次元 / アラインメント / 変形 / ボリュームデータ / CT / MRI

Outline of Final Research Achievements

This research shows that against the intuition of impossibility of exhaustive search of a relatively large number of parameters, by a number of techniques to reduce computational cost, we succeeded in localization and alignment of livers in MR volume data by exhaustive search of 15 parameters including 3 for position, 3 for orientation, 1 for scale, and 8 for deformation. Deformation is represented as a linear combination of principal components of a set of some 50 aligned 3D sample models.
Using GPU for acceleration, we were able to find livers withour any human intervention at a speed of 1.5 minutes, which is now manually done by human operators at a speed of half a day.

Free Research Field

三次元画像認識