Unencrypted login | home

Program Information

Analysis of Morphology of Normal Liver and Tumor by Breathing Motion Using Deformable Registration Method

no image available
G Gong

G Gong1*, H wang2, Y Yin3, (1) ,,,(2) shandong cancer hospital, Jinan, ,(3) ,Jinan,

SU-E-J-80 Sunday 3:00PM - 6:00PM Room: Exhibit Hall

Purpose: Through combining the free form deformable registration algorithm and four-dimensional computed tomography (4D-CT) to discuss morphology of normal liver and tumor by breathing motion of hepatocellular carcinoma patients.

Methods: 8 hepatocellular carcinoma patients accepted radiotherapy was selected, then acquire 4D-CT simulate positioning images and 4D CT images were segmented into 10-series CT images which were named CT0, CT10 to CT80, CT90 according to the respiratory phases with CT0 at end inspiration and CT50 at end expiration. We chose the CT50 as the reference image. we used the free form deformable (FFD) model based on B-spline grid registration method to deform the other 9 set-image to CT50, this paper also utilized limited-memory Broyden Fletcher Goldfarb Shannon minimization (L-BFGS) to enhance the registration accuracy and speed. Then the subtractions,hausdroff distance and multi-information are used to analyze the influnce of normal liver and tumor targets from breathing motion.

Results: (1)The breathing movement for deformation of normal liver and tumor targets is obvious, and the difference between CT0 and CT50 is biggest. (2)The FFD model based on B-spline grid registration method can deform 4D-CT images effectively. The hausdroff distance between CT0 and CT50 of normal liver and tumors was from 0.0914 to 0.0432 after registered, the average improvement proportion was 54.1%, and the multi-information was from 0.789 to 0.814, the average improvement proportion was 3.17%.

Conclusion:The breathing movement for deformation of normal liver and tumor targets is not ignored, and the FFD model based on B-spline grid registration method can deform 4D-CT images effectively.

Contact Email: