PreprintPDF Available

Backflow mitigation in stepped catheters

Authors:
Preprints and early-stage research may not have been peer reviewed yet.

Abstract

Background: Backflow of infusate along the catheter shaft can lead to undesirable distributions of drug during convection-enhanced delivery into the brain parenchyma. Stepped catheter geometries have been suggested to limit backflow. This paper contains (i) experimental results on the relation between the size of the step and the mitigation of backflow; and (ii) a theoretical appendix in which we examine possible explanations of the effect (and reject some).
1
Backflow mitigation in stepped catheters
Raghavan R.1, Brady M.L.1, Grabow B.3,
and Block W.2,3,4
1Therataxis LLC, Baltimore, MD 21218, USA;
Departments of 2Medical Physics, 3Bioengeneering , University of Wisconsin, Madison, WI
53715, USA; 4inseRT MRI, Madison, WI
CORRESPONDING AUTHOR:
Raghu Raghavan
Therataxis, LLC
1101 E. 33rd Street, Ste. B305, Baltimore, MD 21218
(443) 451-7154 (phone); (443) 451-7157 (fax)
raghu@therataxis.com
SHORT TITLE:
Stepped catheters
KEY WORDS:
Convection-enhanced delivery, backflow, intraparenchymal infusion, magnetic resonance
imaging, reflux-resistant, catheters, prestress.
2
ABSTRACT
Background: Backflow of infusate along the catheter shaft can lead to undesirable distributions
of drug during convection-enhanced delivery into the brain parenchyma. Stepped catheter
geometries have been suggested to limit backflow.
Objective: The principal objective of this study was to compare two different-sized step catheter
designs, SmartFlow™ 16- (SF16)and 14-gauge (SF14) catheters, for their backflow
characteristics at various flow rates.
Methods: The experiments were conducted in live porcine brains, with catheters were inserted
into the thalamus. In-vivo infusions, monitored continuously, of diluted gadodiamide at flow
rates increasing from 2.5 L/min to 10 L/min were conducted to observe the backflow in the SF
catheters which were placed at the same location in either thalamus.
Results: The smaller SF16 was far more likely to restrain backflow at the 3mm step than the
larger SF14, particularly at the lower flow rates. This led to lower average backflow (3.9±2.0
mm for the 16-gauge vs. 9.0±5.0 mm for the 14-gauge at 2.5 μl/min).
Conclusions: The SF16 stepped catheter proved more likely to restrict backflow at the step. A
simple theoretical estimate, based on our previous study of backflow, suggests that prestress in
the tissue is an additional effect on backflow –compared with the effects accounted for in the
usual theories of backflow -- that may account for the superior performance. However, the
SF16 has then two features which reduce backflow in comparison with the SF14: the smaller
diameter of the tip, and the larger width of the step. The experiments could not disentangle the
effects due to each, and unclear whether the smaller diameters of the step and shaft, or the
larger step size are primarily responsible for the difference. Further studies would be valuable
in clarifying the causes of backflow reduction.
3
INTRODUCTION
MATERIALS AND METHODS
All experiments were performed at the University of Wisconsin, Madison.
Animal model: All infusions were targeted in the thalamus of adolescent swine, as it is a large
enough volume to contain multiple high-volume infusions during experiments at high flow rates.
The swine was chosen over NHP models for its similarity of head size to humans, appropriate
use of animal resources, lack of disease transmission risks, and substantial reduction in
required animal care personnel during the procedure. The swine has also been extensively
utilized elsewhere for investigations involving CED procedures in the brain [1-3].
Surgery and Targeting Method.
We utilized a platform for MR-guided neurological surgery that was first implemented at UW-
Madison [4-6] and is now being extended by a commercial entity (inseRT MRI, Madison, WI).
The platform accesses the scanner hardware through a commercial portal termed RTHawk,
(HeartVista, Palo Alto, CA) [4], that permits the design and execution of image-guided
procedures using high level imaging plug-ins to define a procedural workflow. The platform
allows the interventionalist to manipulate the brain ports used to orient catheters with real-time
feedback similar to a stereotactic OR suite, rather than the iterative “shoot and scoot”
manipulation provided commercially in MR image guidance today.
Through a bidirectional link with RTHawk, image data acquired in real-time was displayed by
Vurtigo (Visual Understanding of Real-Time Image Guided Operations, Sunnybrook Health
Sciences Centre; Toronto, Canada), an open-source visualization platform that allows
simultaneous display and interaction with multiple 3D and 2D datasets [7]. Changes to the
desired acquisition geometry made in Vurtigo were sent to RTHawk and then on to the scanner,
allowing visualization of dynamically acquired 2D planes overlaid on previously acquired 3D
volumes. This link was implemented using a “geometry server” software program that
communicates with RTHawk and Vurtigo.
Real-time scan control and visualization was conducted on a high-performance external
workstation with two quad-core Intel Xeon E5620 2.4 GHz CPUs, 12 GB of memory, an NVIDIA
GF100 Quadro 4000 graphics card, and gigabit Ethernet controller, running 64-bit Linux.
Scanner interface was via an internal Ethernet switch. Visualization display was available on a
screen in the control room, which was placed in the scanner room window so that an operator
could lean into the bore and reposition the MR-visible fluid-filled alignment guide to the optimal
trajectory angle
The device targeting capability is based on a real-time implementation of prospective stereotaxy
[8]. The NavigusTM brain port consists of a MR-visible fluid-filled alignment guide seated in a
ball-and-socket pivot joint with two degrees of rotational freedom. The visible tip of the guide
was positioned at the center of the pivot joint, and the body of the guide extended away from the
skull along the trajectory of device insertion. We identified the desired target point in the brain
and the location of the pivot point by first acquiring a high resolution 3D T1-weighted “roadmap”
volume with inversion recovery preparation. After identifying location markers of the alignment
guide pivot point and the Ce target point, the trajectory alignment software tool calculated an
“aiming point” outside the skull that was co-linear with the target and alignment guide pivot
4
points. Real-time imaging of a plane perpendicular to and centered on the aiming point provided
dynamic feedback during the alignment step, allowing the operator to incrementally move the
alignment guide until the distal end of the alignment guide overlapped with the software-
displayed aiming point.
When the trajectory angle (antero-postrerior, medio-lateral direction) of the fluid-filled alignment
guide was confirmed to be on target, the alignment stem was locked into position. The fluid-filled
alignment guide was removed, the remote introducer was fastened to the stem, and the guiding
insert was placed in the guiding stem. The catheter for the infusion was threaded through the
remote introducer and the guiding insert, and was fastened to the remote introducer by a locking
mechanism.
Catheters
Two sizes of SmartFlow™ multi-stepped catheter (MRI Interventions, Inc., Memphis, TN, USA)
were compared in this study. The two catheters, shown in Fig. 1, differ primarily in the outer
diameters of their three sections. Both catheters have a 3mm length tip with open end port,
followed by at step to an 18mm long central shaft of larger diameter, which then tapers to a 16-
or 14- gauge main shaft. Table 1 lists the outer diameters of tip and central shaft, and the
resulting computed step width (half the difference between shaft and tip diameters).
Infusion.
Silica infusion lines were used to connect the catheter to 5cc. Hamilton gas-tight syringes. The
syringes were driven by an MRI-compatible syringe pump (PHD 2000, Harvard Apparatus, Inc.,
Holliston, MA, USA) placed at approximately the same height as the catheter tip. A solution of
gadodiamide (2 mmol/l; Omniscan™, GE Healthcare Inc., Princeton, N.J., USA) in phosphate-
buffered saline was infused. For each catheter placement, backflow measurements were
obtained at multiple flow rates. In three of the four animals, the initial flow rate was 2.5 μl/min,
while the fourth was started at 5.0 μl/min. The infusion was run at the initial flow rate for 5
minutes, and then stopped for 8-9 minutes while a high resolution 3D MRI image was acquired.
This process was repeated at flow rates increasing by 2.5 μl/min at each iteration. The flow rate
was incremented in this manner until either the backflow reached a CSF boundary (at which
point no additional backflow is expected) or a maximum flow rate of 10.0 μl/min was reached.
MRI Acquisition.
All procedures were monitored with a 3-Tesla GE SIGNA (MR 750) MRI scanner (GE
Healthcare, Waukesha, WI). During infusion, the infusion was monitored with real-time 2D
imaging. A single slice aligned with the catheter was acquired at 6 sec intervals (TR = 33 ms,
TE = 4.5 ms, flip angle = 50 o, FOV = 140 x 105 mm, matrix = 256x192, slice thickness = 2.5
mm). After each infusion, a 3D fast spoiled gradient echo (FSPGR) scan was acquired (TR =
9.1 ms, TE = 3.9 ms, in-plane FOV = 180 mm, matrix = 256x224, slice thickness = 0.8 mm, 248
contiguous slices spaced at 0.4 mm intervals.)
Backflow Measurement
Backflow was identified and monitored during each five-minute infusion using in real time MRI.
Backflow distance was measured in the 3D T1-weighted scan acquired just after the flow was
stopped. The distance was measured from the catheter tip to the furthest extent of visible tracer
along the catheter shaft using custom 3D image processing software. Measures were recorded
to 0.5mm precision, which was limited by the resolution of the MR imaging.
5
Statistical Analysis
Backflow distances were compared using a two-sample t-test, assuming equal variances. Two-
tailed tests were used, with statistical significance defined as p < 0.05. Grouped data are
presented in the test as M ± SD.
RESULTS
Four pigs were used for a total of 16 infusion sites. Each pig therefore served for four infusion
sites which consisted of paired sites in each hemisphere, with two sites in the thalamus in each
hemisphere. Tables 2A and 2B summarize the set of backflow distances measured in the
study.
Most infusions started at 2.5 μl/min, and all reached at least 5.0 μl/min before the backflow
reached a CSF boundary. At both 2.5 and 5.0 μl/min, the average backflow of the 16-gauge
catheter was less than that of the 14-gauge, reaching statistical significance at 2.5 μl/min
(p=0.040) and just failing at 5.0 μl/min (p=0.051). However, it should be noted that most (6 out
of 8) of the 14-gauge infusions reached a CSF boundary at 5.0 μl/min, which limited their
backflow distance. The remaining two reached CSF at 7.5 μl/min. In contrast, none of the 16-
gauge infusions reached CSF at 5.0 μl/min and only two at 7.5 μl/min.
Statistical analysis of the backflow distances may not be appropriate in this study, due to the low
number of samples and high standard deviation of the results. Furthermore, the backflow
distance measure does not appear to be normally distributed. In every case, the backflow was
observed to reach at least to the catheter step, 3 mm from the tip. Furthermore, in the 16-gauge
catheter backflow distances were clustered around the 3mm value. Five out of six backflows at
2.5 μl/min and five of eight at 5.0 μl/min are in the range of 3-4mm. The remaining four
backflow that overran the step had widely distributed backflows, from 6 – 12 mm. By contrast,
only one of the 14-gauge catheter infusions, at 2.5 μl/min, had a backflow in the 3-4 mm range.
Images of the full set of eight infusions with each catheter at 5.0 μl/min are shown in Fig. 2. T1-
weighted 3D MR data sets acquired after completion of the five-minute 5.0 μl/min infusion were
rotated so that the catheter appears in-plane in each image. The infusate appears to be
concentrated around the tip in most of the 16-gauge infusions (upper row), while the 14-gauge
infusions are distributed around a greater length of the central shaft (lower row).
DISCUSSION
It appears from this study that the step in the 16-gauge catheter does tend to restrict backflow.
The barrier is not absolute. A fraction of the infusions did backflow past the step, and this
fraction increased with the flow rate. For those infusions that did overrun the step, the backflow
varied widely. Typically, a large increase in backflow was seen at the lowest flow rate to
overrun the step, and much smaller increases with flow rate thereafter. Furthermore, among
those infusions at a given flow rate that had overrun the step, there was little difference in
between the 16- and 14-gauge catheters; i.e., their difference lies mainly in the tendency to
overrun the step, not in the performance for backflow beyond the step.
It is not unexpected that a smaller catheter would display lower backflow. Theoretical models
and experiment have demonstrated a relation between catheter radius and backflow distance in
straight open end port catheters: the simplest such model is described in [9]. For more
numerical and recent approaches, see [12], [13]. However, the mechanism underlying the
backflow-restricting properties of stepped designs is not well understood. It is important to note
that the 16-gauge catheter is not a proportionally scaled down version of the 14-gauge. While
6
the tip and shaft are both smaller than their counterparts in the 14-gauge design, the size of the
step is in fact larger by nearly a factor of two. It is not clear from the data itself whether the
larger step or the smaller radii play a greater role in the improved performance.
In order to better visualize the effect of the step, we undertook some experiments in an Agarose
gel, which we describe here.
Materials and Methods
All experiments were performed at the Engineering Resources Group in Hialeah, FL, USA. The
gel used was a 0.6% agarose gel prepared in a standard way [10]. A catheter with a step, with
the same dimensions as the SF16 was then used in the experiment.
Infusion.
The images shown were taken when the catheter was inserted at a relatively fast insertion (5
mm/sec). This is to ensure that it encounters sliding friction when inserted, and not stick-slip
friction which will obtain at slow insertion speeds. The latter would allow a punch to develop
elastically and then tear, resulting in reflux due to gel damage rather than backflow which is due
to response within the domain of elastic response and porous flow.
Results
It is seen that the flow has difficulty penetrating a small region in front of the step, which
therefore acts as a barrier. It is also interesting to compare this with the behavior reported in
[11] where the shaft of a cylindrical catheter with no step is shown to have accumulated tissue in
front of it at slow insertion speeds, and this has the opposite effect that the step does: namely it
seems to increase the reflux due to the tissue damage caused.
In the Appendix, we show a simple calculation that indicates that prestress in addition to the
conventional factors that affect backflow may be sufficient to explain the phenomena observed.
Further discussion of the issues that need to be examined is discussed there.
ACKNOWLEDGEMENTS
This research was supported by the Kinetics Foundation.
7
REFERENCES
[1] Martin L. Brady, Raghu Raghavan, Deep Singh, P J Anand, Adam S. Fleisher, Jaime Mata,
William C. Broaddus, and William L. Olbricht. In-vivo performance of a micro-fabricated catheter
for intraparenchymal delivery. Journal of Neuroscience Methods. 229: 76-83, 2014.
[2] Martin Brady, Raghu Raghavan, Zhi jian Chen, and William C. Broaddus. Quantifying fluid
infusions and tissue expansion in brain. IEEE Transactions on Biomedical Engineering,
58:2228-2237, 2011.
[3] A. Jones, A. Bienemann, N. Barua, P. J. Murison, and S. Gill, "Anaesthetic complications
pigs undergoing MRI guided convection enhanced drug delivery to the brain: a case series,"
Veterinary anaesthesia and analgesia, vol. 39, pp. 647-52, 2012.
[4] Emborg ME, Joers V, Fisher R, Brunner K, Carter V, Ross C, et al. Intraoperative
intracerebral MRI-guided navigation for accurate targeting in nonhuman primates. Cell
transplantation. 2010;19(12):1587-97. doi: 10.3727/096368910X514323. PubMed PMID:
20587170; PubMed Central PMCID: PMC3278961.
[5] Brodsky EK, Block WF. Intraoperative device targeting using real-time MRI. Biomedical
Sciences …. 2011:6-9.
[6] Grabow B, Block W, Alexander AL, Hurley S, CD R, Sillay K, et al., editors. Extensible real-
time MRI platform for intraoperative targeting and monitoring. Society of Brain Mapping and
Therapeutics; 2012; Toronto.
[7] Radau, P. E. et al. VURTIGO : Visualization Platform for Real-Time , MRI-Guided Cardiac
Electroanatomic Mapping. 244-253 (2012).
[8] Truwit, C.L and Liu, H. Prospective stereotaxy: a novel method of trajectory alignment using
real-time guidance. Journal of Magnetic Resonance Imaging, vol 13, 452 – 457 (2001).
[9] Raghu Raghavan, Samuel Mikaelian, Martin Brady, and Zhi-Jian Chen. Fluid infusions from
catheters into elastic tissue I: azimuthally symmetric backflow in homogeneous media. Physics
in Medicine and Biology, 55:281-304, 2010.
[10] Chen ZJ1, Gillies GT, Broaddus WC, Prabhu SS, Fillmore H, Mitchell RM, Corwin
FD, Fatouros PP. A realistic brain tissue phantom for intraparenchymal infusion studies. J
Neurosurg. 2004 Aug;101(2):314-22.
[11] Casanova F, Carney PR and Sarntinoranont M. Influence of needle insertion speed on
backflow for convection-enhanced delivery J Biomech Eng. 2012 134(4):041006.
[12] G. A. Orozco, J. H. Smith, and and J. J. García. Backflow length predictions during flow-
controlled infusions using a nonlinear biphasic finite element model. Medical &
Biological Engineering & Computing, 52:841–849, 2014. doi:10.1007/s11517-014-1187-1.
[13] J. J. García, A. B. Molano, and J. H. Smith. Description and validation of a finite
element model of backflow during infusion into a brain tissue phantom. Journal of
Computational and Nonlinear Dynamics, 8:011017, 2013. doi:10.1115/1.4007311.
8
Figure 1: A side view photograph of the SFh (top) and SFa (bottom) catheters.
Catheter Tip diameter
(mm)
Shaft diameter
(mm)
Step width
(mm)
16
gauge 0.36 0.76 0.2
14
gauge 0.665 0.877 0.106
Table 1. The measured dimensions of the catheters
9
Pig Infusion
Backflow (mm)
2.5
μl/min
5.0
μl/min
7.5
μl/min
10.0
μl/min
18 3 3.0 4.0 10.0
18 4 3.0 7.0 9.0
19 3 3.0 12.0 13.0
19 4 3.0 3.0 3.0 7.0
20 3 3.0 4.0 8.0 13.0
20 4 8.0 11.0 12.0
25 2 3.5 6.0 8.0 9.0
25 3 3.0 4.0 6.5 8.0
Mean ± SD 3.9 ± 2.0 5.8 ± 3.7 7.7 ± 3.5 9.3 ± 2.0
Table 2A. Backflow in the 16-gauge SmartFlow catheter.
Pig Infusion
Backflow (mm)
2.5
μl/min
5.0
μl/min
7.5
μl/min
10.0
μl/min
18 1 5.0
18 2 11.0
19 1 8.0 8.0
19 2 6.5 10.0
20 1 18.0 18.0
20 2 3.0 5.0 13.0
25 1 9.0 9.0 9.0
25 4 10.0 17.5
Mean ± SD 9.0 ± 5.0 10.4 ± 5.0
Table 2B. Backflow in the 14-gauge SmartFlow catheter.
10
Figure 2: T1-weighted MR imaging of eight paired infusions of the SmartFlow 16- and 14-gauge
catheters showing the backflow at 5 L/min. The top row shows the backflow around the 16-
gauge catheter, while the bottom row shows the infusion from the 14-gauge catheter. The 16
gauge catheter has a smaller tip diameter as well as a larger step width than the 14-gauge one.
11
(a) (b) (c)
(d) (e) (f)
(g) (h) (i)
(j) (k) (l)
Figure 3: A series of “time lapse” images of a dye infusion within gel from a stepped catheter.
Appendix: pre-stress and ow redirection in stepped catheters
Raghu Raghavan
A. The effect of prestress from a step
The step catheter was rst claimed by its inventors to reduce backow by redirecting the
ow laterally. However, we surmise that the eect of prestress is more likely to be eective
for this purpose. We use the simple theory of backow developed earlier [9] to estimate
the reduction of backow due to pre-stress from a step catheter. In a second subsection,
we consider the eect of ow redirection, which can be added to the design of the tests,
but the eect is negligible for the designs being considered. The more important eect
is due to the prestress. The calculations here are at best order of magnitude estimates:
more detailed work, necessarily numerical, may be called for when moreso studies are
done to investigate the phenomenon. We should also mention that, since this work was
completed, we have studied another catheter designed to have a so-called bullet nose that
has a maximum width considerably larger [1]: correspondingly it seems to limit backow
even more strongly, giving further evidence for our pre-stress hypothesis.
We make the following assumptions to simplify the treatment here:
1. The microtip at the end of either the 14 gauge (OD: 0665 mm) or 16 gauge (OD:
036 mm) SmartFlow(TM) catheter tears thetissueapartsothatthereisnostress
onthetissueforthelength(3mm) of this microtip. These tips are the same
length (3mm) in both (though unfortunately not at all of the same diameter which
complicates understanding of the phenomenon as well as the analysis).
2. The wider cannula that has a step (radial extent 01mmfor the 14 gauge and 02mm
for the 16 gauge) that applies a stress or a deformation to the tissue and does not
tear it: we assume the deformation is entirely radial.
3. We will need two kinds of quantities to estimate a third, given the above data on
the catheters. Namely, we need to know the ow rate ref that has a backow just
1
Appendix: pre-stress and flow redirection in stepped catheters 2
equal to the tip length (3mm) in each catheter. Since the catheters are of dierent
diameters, from the assumption made in 1 above, we will get dierent numbers for
the two catheters. These ow rates were not measured, so we will assume such
a distance for one catheter, and estimate it for the other from theory (see below).
The other quantity we need is the ow rate 0(
ref ) which is just stopped by the
step in question. Then we can estimate the ow rate that is just prevented by the
other catheter.
4. Theory does allow us to estimate these ow rates directly, but at the cost of knowing
tissue parameters, which we shall avoid in this preliminary analysis.
5. We thus need two quantities which we have to estimate. We can take these to be
ref [14] and 0[14] which are the ow rates for the 14 gauge catheter at which the
backow has reached 3mm, and at which the step size of 01just prevents further
backow. We shall then estimate the corresponding quantities for the 16 gauge
catheter, the quantity 0[16] being of the most interest.
Of course, had we had identical tip diameters for both catheters, and only a varying width
for the step, the comparison would involve fewer parameters to be tted, but unfortunately
this was not the case. As stated above, we shall make use of the formulas derived in [9].
The formulas are only approximate but they are simple, leading to simple rough criteria.
It is shown there that uid pressure in the annulus between catheter of outer radius and
soft tissue will shear away a tissue—free width
()'()
2()(1)
“Soft” here means that the shear modulus of tissue is much less than the compressional
modulus. In the formula above it is assumed there is a uid pressure ()at a point
adjacent to the catheter, along its length, and measured back from its tip. Then, if the
ow rate is =0 , along with the corresponding to pressure rise at catheter tip to =0,
Appendix: pre-stress and flow redirection in stepped catheters 3
we know from the previous work that
=1
3µ5000×3
=0 ×4×4
19231
5
;()==0 ³1
´;()==0 µ1
2
3
(2)
The relationship between the pressure =0 at the tip of the catheter and the ow rate
=0 there is also known (see previous paper for the approximations involved: in par-
ticular =0 is not quite the pump ow rate, but can be computed from it if a rened
estimate is needed):
=0 =25
=0 µ1203
41
5
=: 25
045(3)
where depends on uid and tissue parameters but not on the radius of the tip, nor on
the ow rates. Now we have at the step of a catheter, which is up a distance 1,say,
from the tip, the ow rate and pressure are givenbyequation(2),with=1.Ifwe
then increase the catheter radius at 1to a larger value +, then we assume that there
is an extra radial pre-stress at this point in accordance with (1):
=
2(4)
see [9]. If
(1)
(5)
we assert that there is no further backow. In other words, we have the equation for the
critical pressure at the tip
=0 µ11
02
3
=2
(6)
Letusassumethespecialconditionthatwehavecalibratedtheexperimentsothatatthe
ow rate ref , the tip length is just the backow length for the given tip radius. This
means
1
0
=µref
=0 35
(7)
Appendix: pre-stress and flow redirection in stepped catheters 4
Then we nally get, using (3), (4), and (6),
=25
015Ã1µref
035!2
3
=15³35
=0 35
ref ´2
3(8)
that the critical step size that just prevents backow given an input ow rate =0
is given by the right hand side of (8). Consider the two SmartFlow catheters, we note
that [16][14] is about 19from the measurements (see Table 1). Similarly, we nd
[16][14] 185. Given all the other errors we have we neglected the eect of the
dierent tip radii in the ratio of the step lengths that stop the corresponding backow.
Unfortunately, we do not have good data on the other quantities involved, either but let
us take ref [14] 1Lmin (which means that this ow rate causes a backow of 3mm,
just up to the step in this catheter). The scaling of backow lengths with radii would give
ref [16] 25Lmin for the ow rates that reach the step for the smaller diameter
catheter. Using the values from Table 1 then, we have
³35
=0[16] 35
ref [16]´2
3
³35
=0[14] 35
ref [14]´2
3
=192 (9)
We now need one more number, say, =0[14],whichistheow rate that is just stopped
by the 14 gauge catheter step. Let us assume this is 2Lmin, in rough accordance
with the discussion above. Then we nd
³35
0[16] 2535´2
3
¡2351¢2
3
=192 (10)
which gives the ow rate that is stopped by the 16 gauge catheter as 0[16] = 66Lmin.
It should be noted that a more rened analysis (which is available) of backow results
in a reduction of the quantity ref[16], given the two catheter diameters. However, we
postpone such renements and their consequences to a later date. The conclusion of this
back-of-the-envelope calculation is to indicate that prestress may be sucient to explain
Appendix: pre-stress and flow redirection in stepped catheters 5
the behavior of the step in resisting backow. Clearly more studies are needed: the
theory can easily be extended and rened to support accurate measurements and test this
hypothesis more rigorously.
It should be noted that absent pre-stress — as is well known, and also evident from
equation (2) — larger diameter catheters are subject to greater backow. Thus, once a
ow makes it past a step, the larger diameter will allow greater backow than a catheter
with the narrow tip diameter throughout. Is there then an optimal step size so that ows
make it past do not result in catastrophic backow? Wecanusetheaboveformulasto
analyze this problem as well, but it would be an academic exercise, since one should design
pre-stress inducing geometries to forbid backow at the maximum ow rates envisaged,
and not design them to fail to do so within the range of ow rates we wish to use.
B. Flow redirection is ineffective in reducing backflow at a step
In the discussion above, we did not allow for the ow rate entering the step to be greater
than the ow rate attempting to leave it and cause further backow. As stated, people
originally speculated that this might be the cause for backow reduction. While it is
obvious that an innite sized step would stop backow, it is also intuitive that the small
steps in question should not have a signicant eect on the ow rate. We conrm this
with a calculation here. The reduction is essentially a ‘radial backow’ phenomenon,
which we can develop in full analogy to the axial backow of [1]. We recall that we
(i) assume fully developed Poiseuille ow in the circular annulus between the step of the
catheter and a putatively sheared away tissue, and (ii) compute an approximation to the
Darcy ow that results from uid leaving this annulus and entering the tissue due to the
reduction in uid ux that results from the Poiseuille ow. (The paper referred to may
be consulted for details.) First we compute radial Poiseuille ow. Let the ow enter the
space at =and let the annuli be be between =and =.Inotherwords2
is the unknown width of the channel, and will be a function of , the radial distance. We
use cylindrical coordinates. Assuming the lossless, incompressible ow is fully developed
Appendix: pre-stress and flow redirection in stepped catheters 6
for 
,wehave
=0
1
 ()=0
Stokes ow obeys

 =µ
 µ1
 ()+2
2
The no slip conditions at the walls are
=0 =± 
The ow rate into the annulus is
=Z
2 =: 4
which denes if is known. Let us assume that the zero of pressure occurs at a distance
, the same used in the NIH backow model, loc.cit. Thus, now calling 2=: ,the
width of the annulus, we get
()=6 ()
3ln
for constant , which does depend on because it is 2in contrast to 1Poiseuille ow
through a cylindrical annulus where the ow does not depend on axial distance. Also,
=3
8
1
µ1³
´2
This completes the Poiseuille ow half of what we need. The second half is to compute the
pressure required to sustain Darcy ow given the ow 0()leaving the tissue—free circular
annulus. This problem is identical to that of the well known electrostatic potential due
to a charged annulus. However, we further simplify everywhere, we shall not use it for
thesamereasonthatweusethesimplied model of an “innite cylinder” in the usual
Appendix: pre-stress and flow redirection in stepped catheters 7
backow model. So we will use the on-axis result, since that will allow us not to deal with
the polar coordinate and retain cylindrical symmetry. We can begin with the on-axis
potential of the ring between and +, see any electrosatics text or the web:
 =
40p2+2
If the charge density is a constant, then we have
(= 0)=2 Z2
1

p2+2=
20µq2+2
2q2+2
1
We ch e ck that as 10
2→∞,
=
20
+const
giving the correct constant electric eld. In order to proceed with the backow calculation,
we need to identify and 0.,asbefore,is
=0()
2
while we should identify 0with . So, on axis we have
(  =)= 0()
4 µp2+2q2+2
1
=: 0()
We ignore since it will be much smaller than even 1.So
4
to a good approximation.
We have
()=6 ()
3ln
;()=
4
0()
Appendix: pre-stress and flow redirection in stepped catheters 8
We also need the relation between or and . From dimensional analysis, we set
=
for now, where is the radius of the host catheter. So
4()=63()
3ln
(11)
So we have
4
0()
=µ63
31
4µln
1
4
1
4()
Integration yields
Z
1
1
4 =4
3
3
4
04
33
4=µ63
31
44
×"2
8×21
4
Γµ1
42ln
+1
22µln
1
4#
1
where Γis the incomplete gamma function. The right hand side evaluates to 2431 8 ×
106. To immediately see how little dierence this makes, let us convert from cgs to
 min so that we get
3
4
03
48×103(12)
with the ow rates on the left being expressed in  min.Thusthelossofis completely
negligible over the distance of (this) step for any clinical ow rate (of the order of 120
 min), since the right hand side is less than two orders of magnitude smaller than the
lowest ow rates used. Only when the ow rates approach 1100  min will this ow
redirection be of any eect: at those ow rates of course there is no measureable backow
no matter what catheter one uses. We conclude that ow redirection is of no eect in
reducing backow.
... As noted, our backflow model was an approximation to only one geometry of catheter-the cylindrical endport-and did not account for pre-stress. Both of these have very significant effects on backflow (see, for example, [32,46]). In addition, porous catheters, deemed useful for delivery over a large area with advantages of bridging over sinks [47], require a quite different model for outflow from the catheter [48]. ...
Article
Full-text available
Intra-parenchymal injection and delivery of therapeutic agents have been used in clinical trials for brain cancer and other neurodegenerative diseases. The complexity of transport pathways in tissue makes it difficult to envision therapeutic agent distribution from clinical MR images. Computer-assisted planning has been proposed to mitigate risk for inadequate delivery through quantitative understanding of infusion characteristics. We present results from human studies and simulations of intratumoral infusions of immunotoxins in glioblastoma patients. Gd-DTPA and 124I-labeled human serum albumin (124I-HSA) were co-infused with the therapeutic, and their distributions measured in MRI and PET. Simulations were created by modeling tissue fluid mechanics and physiology and suggested that reduced distribution of tracer molecules within tumor is primarily related to elevated loss rates computed from DCE. PET-tracer on the other hand shows that the larger albumin molecule had longer but heterogeneous residence times within the tumor. We found over two orders of magnitude variation in distribution volumes for the same infusion volumes, with relative error ~20%, allowing understanding of even anomalous infusions. Modeling and measurement revealed that key determinants of flow include infusion-induced expansion and loss through compromised BBB. Opportunities are described to improve computer-assisted CED through iterative feedback between simulations and imaging.
Conference Paper
Full-text available
Guidance of electrophysiological (EP) procedures by magnetic resonance imaging (MRI) has significant advantages over x-ray fluoroscopy. Display of electroanatomic mapping (EAM) during an intervention fused with a prior MR volume and DE-MRI derived tissue classification should improve the accuracy of cardiac resynchronization therapy (CRT) for ventricular arrhythmias. Improved accuracy in the spatial localization of recorded EP points will produce an EAM to constrain and customize patient-specific cardiac electroanatomic models being developed for understanding the patterns of arrhythmogenic slow conduction zones causing reentry circuits and treatment planning. The Vurtigo software presented here is a four dimensional (3D+time) real-time visualization application for guiding interventions capable of displaying prior volumes, real-time MRI scan planes, EAM (voltage or activation times), segmented models, and tracked catheters. This paper will describe the architecture and features of Vurtigo followed by the application example of guiding percutaneous cardiac electroanatomic mapping in porcine models.
Article
Full-text available
The technique of direct infusions into brain tissue of therapeutic molecules that would otherwise not adequately cross the blood-brain barrier (BBB) continues to be used in clinical trials. As part of our research into understanding the transport of fluids and molecules in brain tissue, we performed infusions of a saline solution of the magnetic resonance (MR) marker Gadodiamide( Omniscan) into porcine brains. We use quantitative concentration measurements of contrast reagents from MR images to both measure the distribution profile of the infusate and to elucidate important determinants of fluid flow during infusions into brain parenchyma. Based on this, and from other MRI data collected during infusion, we give preliminary results for the quantification of the expansion of the volume fraction of the interstitium particularly in white matter regions of brain during infusion-induced edema. We claim this expansion, rather than an anisotropy of fluid conductivity, makes white matter tracts a preferred pathway for flow. We also comment briefly on other determinants that are currently being pursued such as the influence of the cerebrospinal fluid and perivascular spaces that may be elucidated with quantitative tracking of tracer, but which need further studies.
Article
Full-text available
During in vivo intracerebral infusions, the ability to perform accurate targeting towards a 3D-specific point allows control of the anatomical variable and identification of the effects of variations in other factors. Intraoperative MRI navigation systems are currently being used in the clinic, yet their use in nonhuman primates and MRI monitoring of intracerebral infusions has not been reported. In this study rhesus monkeys were placed in a MRI-compatible stereotaxic frame. T1 MRIs in the three planes were obtained in a 3.0T GE scanner to identify the target and plan the trajectory to ventral postcommisural putamen. A craniotomy was performed under sterile surgical conditions at the trajectory entry point. A modified MRI-compatible trajectory guide base (Medtronic Inc.) was secured above the cranial opening and the alignment stem applied. Scans were taken to define the position of the alignment stem. When the projection of the catheter in the three planes matched the desired trajectory to the target, the base was locked in position. A catheter replaced the alignment stem and was slowly introduced to the final target structure. Additional scans were performed to confirm trajectory and during the infusion of a solution of gadoteridol (ProHance, Bracco Diagnostics; 2 mM/L) and bromophenol blue (0.16 mg/ml) in saline. Monitoring of the pressure in the infusion lines was performed using pressure monitoring and infusion pump controller system (Engineering Resources Group Inc.) in combination with a MRI-compatible infusion pump (Harvard). MRI during infusion confirmed successful targeting and matched postmortem visualization of bromophenol blue. Assessment of the accuracy of the targeting revealed an overall 3D mean ± SD distance error of 1.2 ± 0.6 mm and angular distance error of 0.9 ± 0.5 mm. Our results in nonhuman primates confirm the accuracy of intraoperative MRI intracerebral navigation combined with an adaptable, pivot point-based targeting system and validates its use for preclinical intracerebral procedures.
Article
Full-text available
Directly injecting therapeutics into brain tissue has been investigated both experimentally and theoretically. Paul Morrison and others from the National Institutes of Health pointed out the importance of backflow along and outside a catheter inserted into the tissue, once steady state conditions have been reached. Here we investigate and extend their model. We begin with a reformulation of their results and demonstrate an exact solution that exhibits the scaling behavior of the model where the surrounding tissue medium is homogeneous and isotropic. We report on experimental tests of our predictions in agarose gels. We describe the limitations of the assumptions used and the utility of our reformulation. Extensions of the model, including improvements on some of its crude assumptions and generalizations to inhomogeneous media, will be submitted separately.
Article
An axisymmetric biphasic finite element model is proposed to simulate the backflow that develops around the external boundary of the catheter during flow-controlled infusions. The model includes both material and geometric nonlinearities and special treatments for the nonlinear boundary conditions used to represent the forward flow from the catheter tip and the axial backflow that occurs in the annular gap that develops as the porous medium detaches from the catheter. Specifically, a layer of elements with high hydraulic conductivity and low Young's modulus was used to represent the nonlinear boundary condition for the forward flow, and another layer of elements with axial hydraulic conductivity consistent with Poiseuille flow was used to represent the backflow. Validation of the model was performed by modifying the elastic properties of the latter layer to fit published experimental values for the backflow length and maximum fluid pressure obtained during infusions into agarose gels undertaken with a 0.98-mm-radius catheter. Next, the finite element model predictions showed good agreement with independent experimental data obtained for 0.5-mm-radius and 0.33-mm-radius catheters. Compared to analytical models developed by others, this finite element model predicts a smaller backflow length, a larger fluid pressure, and a substantially larger percentage of forward flow. This latter difference can be explained by the important axial flow in the tissue that is not considered in the analytical models. These results may provide valuable guidelines to optimize protocols during future clinical studies. The model can be extended to describe infusions in brain tissue and in patient-specific geometries. [DOI: 10.1115/1.4007311]
Article
A previously proposed finite element model that considers geometric and material nonlinearities and the free boundary problems that occur at the catheter tip and in the annular zone around the lateral surface of the catheter was revised and was used to fit a power-law formula to predict backflow length during infusions into brain tissue. Compared to a closed-form solution based on linear elasticity, the power-law formula for compliant materials predicted a substantial lower influence of the shear modulus and catheter radius on the backflow length, whereas the corresponding influence for stiffer materials was more consistent with the closed-form solution. The finite element model predicted decreases of the backflow length for reduction of the shear modulus for highly compliant materials (shear modulus less than 500 Pa) due to the increased area of infusion and the high fluid fraction near the infusion cavity that greatly increased the surface area available for fluid transfer and reduced the hydraulic resistance toward the tissue. These results show the importance of taking into account the material and geometrical nonlinearities that arise near the infusion surface as well as the change of hydraulic conductivity with strain for a proper characterization of backflow length during flow-controlled infusions into the brain.
Article
Convection-enhanced delivery (CED) is currently the only effective clinical technique to deliver biological therapeutic agents that would otherwise not cross the blood-brain barrier. Despite the promise of CED, several technical problems have limited its effectiveness. Brain infusions into a large mammal (pig) were performed with a catheter that was fabricated using micro-electro-mechanical systems (MEMS) technology (Olbricht et al., 2010). The performance of the catheter was evaluated for infusions at increasing infusion rates. Magnetic resonance (MR) images were acquired in real time to examine the distribution of infused tracers in the parenchyma. Both backflow and the distribution of CED of infusates into a variety of cytoarchitectures in porcine brain were quantified. Concentration profiles were determined for several MR contrast reagents as well as a fluorescent dye that are the sizes of small molecules, therapeutic proteins and an adeno-associated virus (AAV). The reagents can serve as surrogates for assessing the convective distribution of active molecules. Infusion rates up to 20μL/min were attained without evidence of backflow along the catheter. The device performed well in terms of both backflow and infusion, superior to that of many studies reported in the literature on other catheters. All infused molecules had comparable ratios of distribution to infusion volumes. The catheter described in this report appears able to target tissue structures with precision, deliver therapeutics at high infusion rates, and resist backflow that can compromise the efficacy of CED therapy. The technology allows development of "smart" catheters for future applications.
Article
Observations: A total of 13 intracerebral infusions were performed at approximately 1 month intervals in three NIH miniature pigs over the age range of 31-59 weeks. Pigs received azaperone and ketamine premedication to allow venous cannulation and propofol induction of anaesthesia. Anaesthesia was maintained with isoflurane throughout cranial surgery and MRI scanning. Physiological monitoring during surgery consisted of blood pressure, pulse, temperature and oxygen saturation monitoring, ECG and capnography. Analgesia consisted of meloxicam and morphine. However, during MRI scanning blood pressure and ECG monitoring had to be discontinued. Anaesthetized pigs underwent intermittent intraputamenal convection enhanced delivery (CED) of gadolinium with real-time magnetic resonance imaging. Progressive tachycardia was consistently observed in all pigs during CED with a mean ± SD maximum increase of 41 ± 22 beats minute(-1) from a baseline heart rate of 96 ± 9 minute(-1) . The heart rate remained elevated until recovery. A mean reduction in body temperature of 2.8 ± 0.6 °C from the start of anaesthesia was also observed during the period of MRI scanning. All pigs recovered from anaesthesia smoothly and heart rates returned to normal during the recovery period. Conclusions: Hypothermia is common in pigs undergoing this sedation and anaesthesia protocol. Convection enhanced delivery of drugs in healthy anaesthetized pigs may result in tachycardia.
Article
To describe prospective stereotaxy, a novel method of trajectory alignment that works in real-time. Prospective stereotaxy was used in minimally-invasive neurosurgical procedures in 74 patients since February 1999. This methodology differs from framed and frameless stereotaxy, both of which are based on retrospective data. Rather, prospective stereotaxy uses real-time MR images to align a surgical trajectory. Phantom tests and clinical procedures in all patients were successfully performed using prospective stereotaxy. In all cases, surgical targets were accessed, and the diagnostic yield of neurobiopsy using prospective stereotaxy was 100%. Prospective stereotaxy is applicable to all cross-sectional imaging, and is particularly useful for MR- and CT-guided interventions. The method is simple, reproducible, and accurate in surgical targeting for neurobiopsy and electrode placement. It does not require cumbersome stereotactic frames or expensive optical detectors, and it offers immediate entry into the field of interventional MR with cylindrical MR scanners.
Article
Fluid flow back along the outer surface of a needle (backflow) can be a significant problem during the direct infusion of drugs into brain tissues for procedures such as convection-enhanced delivery (CED). This study evaluates the effects of needle insertion speed (0.2 and 1.8 mm/s) as well as needle diameter and flow rate on the extent of backflow and local damage to surrounding tissues. Infusion experiments were conducted on a transparent tissue phantom, 0.6% (w/v) agarose hydrogel, to visualize backflow. Needle insertion experiments were also performed to evaluate local damage at the needle tip and to back out the prestress in the surrounding media for speed conditions where localized damage was not excessive. Prestress values were then used in an analytical model of backflow. At the higher insertion speed (1.8 mm/s), local insertion damage was found to be reduced and backflow was decreased. The compressive prestress at the needle-tissue interface was estimated to be approximately constant (0.812 kPa), and backflow distances were similar regardless of needle gauge (22, 26, and 32 gauge). The analytical model underestimated backflow distances at low infusion flow rates and overestimated backflow at higher flow rates. At the lower insertion speed (0.2 mm/s), significant backflow was measured. This corresponded to an observed accumulation of material at the needle tip which produced a gap between the needle and the surrounding media. Local tissue damage was also evaluated in excised rat brain tissues, and insertion tests show similar rate-dependent accumulation of tissue at the needle tip at the lower insertion speed. These results indicate that local tissue damage and backflow may be avoided by using an appropriate insertion speed.