Data-Guided Brain Deformation Modeling: Evaluation of a 3-D Adjoint Inversion Method in Porcine Studies

Biomechanical models of brain deformation are useful tools for estimating parenchymal shift that results during open cranial procedures. Intraoperative data is likely to improve model estimates, but incorporation of such data into the model is not trivial. This study tests the adjoint equations meth...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on biomedical engineering 2006-10, Vol.53 (10), p.1893-1900
Main Authors: Lunn, K.E., Paulsen, K.D., Liu, F., Kennedy, F.E., Hartov, A., Roberts, D.W.
Format: Article
Language:English
Subjects:
Adjoints
Algorithms
Animals
Brain
Brain - diagnostic imaging
Brain - physiology
Brain deformation model
Brain modeling
brain shift
Computed tomography
Computer Simulation
Cranial
Data assimilation
Deformable models
Deformation
Displacement
Displacement measurement
Elasticity
Equations
Estimates
Estimating
image-guided neurosurgery
Imaging, Three-Dimensional - methods
Mathematical models
Measurement errors
Models, Biological
Physical Stimulation - methods
Radiographic Image Interpretation, Computer-Assisted - methods
Stress
Stress, Mechanical
Studies
Swine
Testing
Three dimensional
Viscosity
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Biomechanical models of brain deformation are useful tools for estimating parenchymal shift that results during open cranial procedures. Intraoperative data is likely to improve model estimates, but incorporation of such data into the model is not trivial. This study tests the adjoint equations method (AEM) for data assimilation as a viable approach for integrating displacement data into a brain deformation model. AEM was applied to two porcine experiments. AEM-based estimates were compared both to measured displacement data [from computed tomography (CT) scans] and to model solutions obtained without the guidance of sparse data, which we term the best prior estimate (BPE). Additionally, the sensitivity of the AEM solution to inverse parameter selection was investigated. The results suggest that it is most important to estimate the size of the variance in the measurement error correctly, make the correlation length long and estimate displacement (over stress) boundary conditions. Application of AEM shows an average 33% improvement over BPE. This paper represents the first evidence of successful use of the AEM technique in three dimensions with experimental data validation. The guidelines established for selection of model parameters are starting points for further optimization of the method under clinical conditions
ISSN:0018-9294
1558-2531
DOI:10.1109/TBME.2006.881771