226
V. Kurtcuoglu et al.
originally suggested by Du Boulay [60]. While it is not entirely clear how vascular
deformation is translated to CSF motion, the third ventricle is no longer considered
the primary source [32, 61]. In 2010, Cheng et al. investigated with an MRI-derived
CFD model of the third ventricle and aqueduct of Sylvius how the position of the
interthalamic adhesion affects CSF flow [62]. They imposed flow at the foramina
of Monro based on published values, set constant pressure at the distal end of the
aqueduct and assumed rigid ventricle walls.
In 2008, Howden and co-workers presented the first anatomically accurate
3D CFD model of the entire ventricular space [63]. Instead of using constant
pressure boundary conditions at the domain outlets (the foramina of Luschka and
Magendie), the authors applied transient pressure. Pressures were estimated using a
geometrically simplified 3D model of the combined cranial and spinal CSF spaces
that featured a pulsatile velocity inlet at the choroid plexus and constant pressure
at the arachnoid granulations. While such a staggered approach is, in principle,
preferable to using constant pressure, it is difficult to carry out in practice. This
is because choroid plexus pulsation is not the sole origin of CSF oscillation,
and the relative importance of the choroid plexus, ventricles and subarachnoid
space as potential contributors to pulsatile flow is unknown. The origin of CSF
motion in the ventricular space was addressed by Siyahhan et al. in 2014 [53],
focusing on the effect of ependymal cilia, which are believed to play a major role
in CSF propulsion in mice. Ciliary action was taken into account by momentum
sources. The computations showed that near the ventricular surface, ependymal cilia
contribute substantially to the local flow conditions.
Changes in cerebral ventricle size are characteristic of several pathologic conditions, most notably hydrocephalus, which is usually treated by ventriculo-peritoneal
CSF shunting [64]. Gholampour et al. investigated CSF dynamics using a fluidstructure interaction model in the ventricular space of hydrocephalus patients with
aqueduct stenosis before and after shunting, focusing on changes in flow and
pressure [65]. Another approach for treating hydrocephalus is endoscopic third
ventriculostomy (ETV) [66], which Gholampour addressed with the same modelling
framework [67]. Farnoush et al. investigated how CSF dynamics are affected by
ETV in the presence and absence of aqueduct stenosis [68]. CFD was performed in
the third ventricle, with the ventriculostomy modelled as a 5 mm diameter hole. The
results showed a threefold higher reduction in ventricular pressure in the case with
stenosis and a temporal shift in pressure conditions in the stenosis-free setup.
The continuous increase in computing power and advances in magnetic resonance technology have allowed for a gradual expansion of the investigated CSF
domain sizes. By considering multiple CSF compartments at once, the number of
required boundary conditions can be decreased. For example, if the ventricular space
and the cranial subarachnoid space are modelled together, no boundary conditions
are needed at the foramina of Luschka and Magendie. Such descriptions including
both ventricular and other CSF spaces are included in the following sections.
V. Kurtcuoglu et al.
originally suggested by Du Boulay [60]. While it is not entirely clear how vascular
deformation is translated to CSF motion, the third ventricle is no longer considered
the primary source [32, 61]. In 2010, Cheng et al. investigated with an MRI-derived
CFD model of the third ventricle and aqueduct of Sylvius how the position of the
interthalamic adhesion affects CSF flow [62]. They imposed flow at the foramina
of Monro based on published values, set constant pressure at the distal end of the
aqueduct and assumed rigid ventricle walls.
In 2008, Howden and co-workers presented the first anatomically accurate
3D CFD model of the entire ventricular space [63]. Instead of using constant
pressure boundary conditions at the domain outlets (the foramina of Luschka and
Magendie), the authors applied transient pressure. Pressures were estimated using a
geometrically simplified 3D model of the combined cranial and spinal CSF spaces
that featured a pulsatile velocity inlet at the choroid plexus and constant pressure
at the arachnoid granulations. While such a staggered approach is, in principle,
preferable to using constant pressure, it is difficult to carry out in practice. This
is because choroid plexus pulsation is not the sole origin of CSF oscillation,
and the relative importance of the choroid plexus, ventricles and subarachnoid
space as potential contributors to pulsatile flow is unknown. The origin of CSF
motion in the ventricular space was addressed by Siyahhan et al. in 2014 [53],
focusing on the effect of ependymal cilia, which are believed to play a major role
in CSF propulsion in mice. Ciliary action was taken into account by momentum
sources. The computations showed that near the ventricular surface, ependymal cilia
contribute substantially to the local flow conditions.
Changes in cerebral ventricle size are characteristic of several pathologic conditions, most notably hydrocephalus, which is usually treated by ventriculo-peritoneal
CSF shunting [64]. Gholampour et al. investigated CSF dynamics using a fluidstructure interaction model in the ventricular space of hydrocephalus patients with
aqueduct stenosis before and after shunting, focusing on changes in flow and
pressure [65]. Another approach for treating hydrocephalus is endoscopic third
ventriculostomy (ETV) [66], which Gholampour addressed with the same modelling
framework [67]. Farnoush et al. investigated how CSF dynamics are affected by
ETV in the presence and absence of aqueduct stenosis [68]. CFD was performed in
the third ventricle, with the ventriculostomy modelled as a 5 mm diameter hole. The
results showed a threefold higher reduction in ventricular pressure in the case with
stenosis and a temporal shift in pressure conditions in the stenosis-free setup.
The continuous increase in computing power and advances in magnetic resonance technology have allowed for a gradual expansion of the investigated CSF
domain sizes. By considering multiple CSF compartments at once, the number of
required boundary conditions can be decreased. For example, if the ventricular space
and the cranial subarachnoid space are modelled together, no boundary conditions
are needed at the foramina of Luschka and Magendie. Such descriptions including
both ventricular and other CSF spaces are included in the following sections.