engineering models of decompressive craniectomy · decompressive craniectomy can be used to control...

Engineering models of decompressive craniectomy

Funded by EPSRC and MRC/NIHR Supported by RESCUEicp

1. Division of Neurosurgery, Addenbrooke's Hospital & University of Cambridge2. Department of Engineering, University of Cambridge

Fletcher TL12, Kolias AG1, Timofeev I1, Corteen EA1, Sutcliffe MPF2, Hutchinson PJ1

Contents

– Introduction

– Finite element modelling approaches

– Image analysis

Background

Decompressive craniectomy can be used to control refractory intracranial hypertension post TBI.

Associated with a number of complications

– Brain herniation

– Deformation

Engineering models may provide insight into the deformation and possible regions of damage.

Suboptimum hemi-craniectomy

Computational finite element (FE) model developed of a similar situation.

– Large curvature near bone margin

– Max displacement ~15 mm

Schematic of suboptimum hemicraniectomy.Wagner, S. et al., 2001. J Neurosurg, 94(5), pp.693-6

Finite element model to mimic the schematic of Wagner et. al.

Deformation profile for small craniectomy shown to be detrimental to patient outcome

The finite element model

Create an overall model

Split the model into small elements (finite elements)

Create a set of partial differential equations....

...Numerically solve

Use commercially available Abaqus FE software.

FE model of craniectomy

Gao and Ang model of craniectomy size against pressure.

Trade-off between bulge and pressure drop.

We are currently focusing on the possible tissue damage caused by craniectomies of varying size.

Craniectomy size vs maximum displacement from Gao, C. & Ang, B., 2008. Acta Neurochir Suppl, pp.279-282.

Max

imum

di

spla

cem

ent

(mm

)

Elements in subcranial area

Research plan

Focus on simple spherical axi-symmetric models

Explore the influence on size of 'at risk region' of two geometrical

parameters:

– craniectomy size

– bone edge fillet radius (or sharpness)

Despite simplification the real world trends are still relevant.

Spherical brain model - schematic

r '=rR

Defining the 'at risk region'

Elkin and Morrison 2007

– Rat cortical brain experiments

– Varying rates/speeds

– Varying strains (stretch)

Cell death quantified and thresholds for damage hypothesised.

– 20% Lagrangian strain.

• Large deformation stretch ratio

Strain and rate vs cell death for day 4 post injury from Elkin, B.S. & Morrison, B., 2007. Stapp car crash J, 51(Oct), pp.127-38.

(s -1)

How to use the threshold

Using computational models:

– Output strain for models with varying parameters

– Create a contour of threshold strain

– Output areas/volumes of regions of strain below this (negative) threshold

See example, top, showing FE output

Bottom showing first at risk region

Output for single fillet radius

Displacement vs at risk region area

Maximum 'at risk region' when r' is 0.3

Does not directly correlate with maximum displacement, which occurs with r' = 0.2

Output

At risk region for shear strain greater than 20% for varying craniectomy size and bone edge fillet radius.

Bone edge fillet radius has little effect on size of at risk region

These models must be validated against patient data

– Analysis of CT scans

CT scan analysis

CT scans - Introduction

Overview

– Pre to post-op registration

– Analysis of brain shape

Analysis...

– Curvature

– Deformation Δy

– Contusion volume

– Midline shift

– Volume changes

CT scans - Method

Conclusions

Models produced show promise in developing understanding of brain deformation post DC

– Potential to assist in determining the optimal size and location of DC

Correlate with radiological and microdialysis data.

Further work will be undertaken to improve the model:

– Poro-elastic materials and time dependence

– Geometrical improvements

– Influence of blood volume on ICP during DC

Sample images