Robust Correspondences for Nonregular Surfaces

Geodesic Distances to Landmarks for Isotropy Invariant Correspondences

[MICCAI 2013]

To address the challenges posed by the inherent complexity (e.g. cortical surface shapes) and variability (e.g. cardiac chambers) evident in many biomedical shapes, we propose the use of geodesic distances to a-priori landmarks as features in the correspondence optimization process. The proposed method minimizes the ensemble entropy based on these features, resulting in isometry invariant correspondences in a very general, flexible framework.


Geodesic Distance & Surface Normal Entropy to Improve Correspondence

[MICCAI 2011]

To resolve the challenges posed by highly nonregular surfaces, we have proposed an efficient method which incorporates Geodesic distances and an intershape penalty based on surface normals to improve correspondences.



Correspondence for Open Surfaces

(w/ J Cates) [MICCAI 2009]

To compute correspondence positions on a set of open surfaces, we define the boundary as the intersection of the surface with a set of geometric primitives. Our goal is to formulate the interactions with these boundaries so that the positions of these constraints has as little influence as possible on the statistical shape model.


Correspondences for Shape Change Over Time

Linear Mixed-effects Model for Shapes

(w/ P Muralidharan, A Kumar) [STIA 2012]

Longitudinal shape changes in anatomy are characterized using a new method that combines point correspondences across shapes with the statistical modeling of individual and population trends via the linear mixed-effects model. This method helps us examine and contrast population trends with individual growth trajectories.


Regression for Brain Structures

The linear regression model was applied to model shape changes in early development of brain structures (left/right hemisphere, cerebellum) and the head shape. This study provides the ability to characterize relative growth of multiple brain structures over a period of time.



Linear Shape Regression

(w/ J Cates) [MICCAI 2009]

Correspondence based on linear regression was incorporated to arrive at a normative model of early head shape development.


Applications in Orthopedics

FAI Analysis

(w/ MD Harris) [JOR 2013] [ORS 2013] [CMBBE 2012]

Cam femoroacetabular impingement (FAI) is characterized by a malformed femoral head that may lead to early hip osteoarthritis. Radiographic measurements are used to diagnose cam FAI and often assume the femur shape to be spherical. Statistical shape modeling (SSM) can be used to compare complex 3D morphology without the need to assume ideal geometry and quantify morphologic differences between control and FAI femurs.

Understanding Short bone Phenotype in Osteochondroma

(w/ K Jones (MD)) [JOR 2013]

Novel statistical methods were developed to study the 'steal phenomenon' caused by multiple osteochondromas in mouse models. Bone lengths and volumes were compared. Metaphyseal volume deviations from normal, as a measure of osteochondroma volumetric growth, were correlated with length deviations.


GP-GPU

GPU Acceleration for Particle Systems

(w/ B Peterson) [SAAHPC 2010]

GPGPU concepts are applied to implement an efficient parallel implementation of the particle distribution process.