136
K. Miller et al.
Surgical simulation systems are used to provide visual and haptic feedback
to a surgeon or trainee. Various haptic interfaces for medical simulation are
especially useful for training surgeons for minimally invasive procedures
(laparoscopy/interventional radiology) and remote surgery using tele-operators.
These systems must compute the deformation field within a soft organ and the
interaction force between a surgical tool and the tissue to present visual and haptic
feedback to the surgeon. Haptic feedback must be provided at frequencies of at
least 500 Hz. From a solid-mechanics perspective, the problem involves large
deformations, non-linear material properties and non-linear boundary conditions.
Moreover, it requires extremely efficient solution algorithms to satisfy the stringent
requirements of the frequency of haptic feedback. Thus, surgical simulation is a
very challenging problem in solid mechanics.
When a simulator is intended to be used for surgeon training, a generic model
developed from average organ geometry and material properties can be used in
computations. However, when the intended application is for operation planning,
the computational model must be patient-specific. This requirement adds to the
difficulty of the problem – the question of how to rapidly generate patient-specific
computational models still awaits a satisfactory answer [1].
6.1.2 Image Registration in Image-Guided Neurosurgery
One common element of most new therapeutic technologies, such as gene therapy,
stimulators, focused radiation, lesion generation, nanotechnological devices, drug
polymers, robotic surgery and robotic prosthetics, is that they have extremely
localised areas of therapeutic effect. As a result, they have to be applied precisely in
relation to the patient’s current (i.e. intra-operative) anatomy, directly at the specific
location of anatomic or functional abnormality [2]. Nakaji and Speltzer [3] list the
‘accurate localisation of the target’ as the first principle in modern neurosurgical
approaches.
As only pre-operative anatomy of the patient is known precisely from medical
images, usually magnetic resonance images (MRIs), it is now recognised that the
ability to predict soft organ deformation (and therefore intra-operative anatomy)
during the operation is the main problem in performing reliable surgery on soft
organs. In the context of brain surgery, it is very important to be able to predict
the effect of procedures on the position of pathologies and critical healthy areas in
the brain. If displacements within the brain can be computed during the operation,
then they can be used to warp pre-operative high-quality MR images so that they
represent the current, intra-operative configuration of the brain; see Fig. 6.1.
The neuroimage registration problem involves large deformations, non-linear
material properties and non-linear boundary conditions, as well as the difficult issue
of generating patient-specific computational models. However, it is easier than the
surgical simulation problem discussed above in two important ways. Firstly, we
are interested in accurate computations of the displacement field only. Accuracy of
K. Miller et al.
Surgical simulation systems are used to provide visual and haptic feedback
to a surgeon or trainee. Various haptic interfaces for medical simulation are
especially useful for training surgeons for minimally invasive procedures
(laparoscopy/interventional radiology) and remote surgery using tele-operators.
These systems must compute the deformation field within a soft organ and the
interaction force between a surgical tool and the tissue to present visual and haptic
feedback to the surgeon. Haptic feedback must be provided at frequencies of at
least 500 Hz. From a solid-mechanics perspective, the problem involves large
deformations, non-linear material properties and non-linear boundary conditions.
Moreover, it requires extremely efficient solution algorithms to satisfy the stringent
requirements of the frequency of haptic feedback. Thus, surgical simulation is a
very challenging problem in solid mechanics.
When a simulator is intended to be used for surgeon training, a generic model
developed from average organ geometry and material properties can be used in
computations. However, when the intended application is for operation planning,
the computational model must be patient-specific. This requirement adds to the
difficulty of the problem – the question of how to rapidly generate patient-specific
computational models still awaits a satisfactory answer [1].
6.1.2 Image Registration in Image-Guided Neurosurgery
One common element of most new therapeutic technologies, such as gene therapy,
stimulators, focused radiation, lesion generation, nanotechnological devices, drug
polymers, robotic surgery and robotic prosthetics, is that they have extremely
localised areas of therapeutic effect. As a result, they have to be applied precisely in
relation to the patient’s current (i.e. intra-operative) anatomy, directly at the specific
location of anatomic or functional abnormality [2]. Nakaji and Speltzer [3] list the
‘accurate localisation of the target’ as the first principle in modern neurosurgical
approaches.
As only pre-operative anatomy of the patient is known precisely from medical
images, usually magnetic resonance images (MRIs), it is now recognised that the
ability to predict soft organ deformation (and therefore intra-operative anatomy)
during the operation is the main problem in performing reliable surgery on soft
organs. In the context of brain surgery, it is very important to be able to predict
the effect of procedures on the position of pathologies and critical healthy areas in
the brain. If displacements within the brain can be computed during the operation,
then they can be used to warp pre-operative high-quality MR images so that they
represent the current, intra-operative configuration of the brain; see Fig. 6.1.
The neuroimage registration problem involves large deformations, non-linear
material properties and non-linear boundary conditions, as well as the difficult issue
of generating patient-specific computational models. However, it is easier than the
surgical simulation problem discussed above in two important ways. Firstly, we
are interested in accurate computations of the displacement field only. Accuracy of