Characterizing the Shape of Anatomical Structures with Poisson's Equation

H. Haidar, S. Bouix, J. J. Levitt, C. C. Dickey, R. W. McCarley, M. E. Shenton, J. S. Soul
Proceedings of the 7th International Conference on Medical Image Computing and Computer Assisted Intervention MICCAI
Pages 266-273
2004

Download full paper

Abstract

This paper presents a novel approach to analyze the shape of anatomical structures. Our methodology is rooted in classical physics and in particular Poisson s equation, a fundamental partial differential equation [1]. The solution to this equation and more specifically its equipotential surfaces display properties that are useful for shape analysis. We demonstrate the solution of this equation on synthetic and medical images. We present a numerical algorithm to calculate the length of streamlines formed by the gradient field of the solution to this equation for 3D objects. We used the length of streamlines of equipotential surfaces to introduce a new function to characterize the shape of objects. A preliminary study on the shape of the caudate nucleus in Schizotypal Personality Disorder (SPD) illustrates the power of our method.

Reference

Haidar H, Bouix S, Levitt JJ, Dickey CC, McCarley RW, Shenton ME, Soul JS. Characterizing the shape of anatomical structures with poisson's equation. In Proceedings of the 7th International Conference on Medical Image Computing and Computer Assisted Intervention MICCAI. 2004;266-273.

Grants

NIH/NIMH 2R01 MH01110, NIH RO1 MH50747, NIH/NIMH 2RO1 MH40799, VA Merit Award, Career Development Award, United Cerebral Palsy Foundation

Research areas

segmentation, shapeanalysis
© 2013 Psychiatry Neuroimaging Laboratory | Last updated 04.15.2013