Radonc/BME Medical Image and
Computational Analysis Laboratory
Chia-Lung Chien
MS graduate student, Medical Physics
expected graduation May, 2018

email link
  1. Research Overview

    3D deformable image registration using finite element modeling.

    Background and Significance
    There are many instances hwere deformable image registration requires intelligent modeling of different compressibility of adjacent tissues. The task of registering chest CT scans over time is often thwarted by large differnces in lung volume, and thus large deformation of the lung, while the surrounding chest wall and internat structures of the heart, major vessels, tumor masses and fibrotic lesions all experience negligible deformation or volume change.

    We hypothesis that a finite-lement based approach may be able to account for large changes in lung volume while preserving size and shape of the non-deforming tissues and organs.

    First, the lung is segmented using automatic image processing methods. A synthetic sagittal series of images is generated from teh axial CT series to facilitate generation of contour points on the surface of the diaphragm on multiple slices. A 3D mathematical model of each hemi-lung surface is generated (in a prolate spheroidal coordinate system). From the 3D surface model, a cubic-Hermite finite element model of the lung surface is generated where the curvature of the lung surface at each finite element node is cmoputed directly from the mathematical model. The FEM is then refined by fitting the nodal parameters directly to the original contour points. The FEM is then overlaid onto the original stack of CT slices (in 3D scanner coords). The 3D image space can then be warped analygously to B-spline-based registration techniques by change the FEM nodal positions and deriviatives and optimizing over the similarity of the deformed image to a reference image.

  2. Related MIACALab projects
    1. Myocardial MRI tagging
    2. Lung radiation dose response

  3. Close Research Collaborators
    1. W. O'Dell (Radiation Oncology Research)