13 Computational Biomechanics of the Brain in the Operating Theatre
325
13.2.1.5 Precomputations
Whenever possible, precomputations could be performed pre-operatively to speed
up the resolution during surgery. Precomputations can be purely numerical. When
using a linear elastic model, a common approach is to compute and store the
deformations resulting from elementary nodal displacements [15]. Other strategies
model the nonlinearity of the brain through precomputations [16–18]. The overall
model response to a set of imposed nodal displacements then relies on the
principle of superposition or linear combination of precomputed deformations.
While precomputations could last several hours, the intra-operative inverse problem
is generally straightforward. Finally, numerical optimisations are also possible, for
example, by pre-inversing the stiffness matrix to handle contacts with Lagrange
multipliers [8].
13.2.2 In the Operating Theatre
13.2.2.1 Intra-operative Imaging
Numerous factors can change between the pre-operative planning and the surgery,
such as the patient positioning, location and size of the craniotomy, loss of
cerebrospinal fluid, anaesthesia control, or tissue retraction and resection. All these
factors directly affect how the brain tissues deform. Therefore, simulations relying
only on a pre-operative model and planning cannot be considered to provide reliable
results. It is then mandatory to capture information, in the operating room, about the
current patient tissue configuration and the realised procedure, to specify adequate
loads and boundary conditions of the model. This new information is usually
acquired at least once, typically after dural opening for craniotomy-induced brain
shift compensation, or at the end of a resection procedure to ensure all tumour
tissues were removed. Whenever possible, repeated acquisitions could be used to
track the deformations all along the procedure.
Many intra-operative imaging modalities have been used in neurosurgery [6],
such as MRI (iMRI), ultrasound (iUS), optical systems such as stereo vision CCD
images or Laser Range Systems (LRS), and to a less extent intra-operative X-rays
system like iCT or cone beam CT. A review of the advantages and drawbacks of
these imaging modalities can be found in [1].
13.2.2.2 Processing Intra-operative Images
Intra-operative imaging can be used to capture information about the patient during
surgery. The acquired images must then be processed to extract relevant features that
will be later transferred to the model as boundary conditions or loads. Examples
of structures to segment are points on the exposed cortical surface, contours (e.g.
325
13.2.1.5 Precomputations
Whenever possible, precomputations could be performed pre-operatively to speed
up the resolution during surgery. Precomputations can be purely numerical. When
using a linear elastic model, a common approach is to compute and store the
deformations resulting from elementary nodal displacements [15]. Other strategies
model the nonlinearity of the brain through precomputations [16–18]. The overall
model response to a set of imposed nodal displacements then relies on the
principle of superposition or linear combination of precomputed deformations.
While precomputations could last several hours, the intra-operative inverse problem
is generally straightforward. Finally, numerical optimisations are also possible, for
example, by pre-inversing the stiffness matrix to handle contacts with Lagrange
multipliers [8].
13.2.2 In the Operating Theatre
13.2.2.1 Intra-operative Imaging
Numerous factors can change between the pre-operative planning and the surgery,
such as the patient positioning, location and size of the craniotomy, loss of
cerebrospinal fluid, anaesthesia control, or tissue retraction and resection. All these
factors directly affect how the brain tissues deform. Therefore, simulations relying
only on a pre-operative model and planning cannot be considered to provide reliable
results. It is then mandatory to capture information, in the operating room, about the
current patient tissue configuration and the realised procedure, to specify adequate
loads and boundary conditions of the model. This new information is usually
acquired at least once, typically after dural opening for craniotomy-induced brain
shift compensation, or at the end of a resection procedure to ensure all tumour
tissues were removed. Whenever possible, repeated acquisitions could be used to
track the deformations all along the procedure.
Many intra-operative imaging modalities have been used in neurosurgery [6],
such as MRI (iMRI), ultrasound (iUS), optical systems such as stereo vision CCD
images or Laser Range Systems (LRS), and to a less extent intra-operative X-rays
system like iCT or cone beam CT. A review of the advantages and drawbacks of
these imaging modalities can be found in [1].
13.2.2.2 Processing Intra-operative Images
Intra-operative imaging can be used to capture information about the patient during
surgery. The acquired images must then be processed to extract relevant features that
will be later transferred to the model as boundary conditions or loads. Examples
of structures to segment are points on the exposed cortical surface, contours (e.g.