Medical Engineering & Physics 36 (2014) 13731380

Contents lists available at ScienceDirect

Medical Engineering & Physics

journal homepage: www.elsevier.com/locate/medengphy

Editors Comment: Reports on medical and clinically related applications of additive manufacturing techniques appear regularly in the
Journal. In this Communication by Herrmann and colleagues, the authors present their experiences of using a fused-deposition technique
to fabricate MRI compatible plastic components. The article details the design and fabrication of a small-animal xation device for use in
a clinical MRI scanner. In the on-line supplement to this article, the authors provide the CAD datasets on which their models were based
for readers to download (in the industry standard STL format). With additive manufacturing systems becoming more readily available,
and at relatively low cost, the technology is likely to nd ever increasing application in biomedical engineering research.

Communication

3D printing of MRI compatible components: Why every MRI research

group should have a low-budget 3D printer
Karl-Heinz Herrmann a, , Clemens Grtner a,b , Daniel Gllmar a , Martin Krmer a ,
Jrgen R. Reichenbach a
a
Medical Physics Group, Institute for Diagnostic and Interventional Radiology I, Center of Radiology, Jena University Hospital, Friedrich-Schiller-University,
Philosophenweg 3, Building 5, 07743 Jena, Germany
b
Ernst-Abbe-Fachhochschule Jena, Jena Carl-Zeiss-Promenade 2, 07745 Jena, Germany

a r t i c l e

i n f o

Article history:
Received 5 February 2014
Received in revised form 30 April 2014
Accepted 15 June 2014
Keywords:
3D printer, MRI, CAD, xation
Small animal imaging

a b s t r a c t
Purpose: To evaluate low budget 3D printing technology to create MRI compatible components.
Material and methods: A 3D printer is used to create customized MRI compatible components, a loop-coil
platform and a multipart mouse xation. The mouse xation is custom t for a dedicated coil and facilitates head xation with bite bar, anesthetic gas supply and biomonitoring sensors. The mouse xation
was tested in a clinical 3T scanner.
Results: All parts were successfully printed and proved MR compatible. Both design and printing were
accomplished within a few days and the nal print results were functional with well dened details
and accurate dimensions ( < 0.4 mm). MR images of the mouse head clearly showed reduced motion
artifacts, ghosting and signal loss when using the xation.
Conclusions: We have demonstrated that a low budget 3D printer can be used to quickly progress from
a concept to a functional device at very low production cost. While 3D printing technology does impose
some restrictions on model geometry, additive printing technology can create objects with complex
internal structures that can otherwise not be created by using lathe technology. Thus, we consider a 3D
printer a valuable asset for MRI research groups.
2014 IPEM. Published by Elsevier Ltd. All rights reserved.

1. Introduction
Commercial clinical MRI systems are usually equipped with a
variety of MR-compatible accessories, for instance to provide positioning support and comfort for patients and staff. However, during
research and initial sequence testing custom-made phantoms are
frequently used, which may not t into the standard xation accessories. Make-shift arrangement solutions to xate and position
phantoms using cushions and sandbags are often sub-optimal, par-

Corresponding author at: Medical Physics Group, IDIR I, Philosophenweg 3, MRT

Gebude Am Steiger, 07743 Jena, Germany. Tel.: +49 3641 935365;
fax: +49 3641 935081.
E-mail addresses: Karl-Heinz.Herrmann@med.uni-jena.de, khh@khherrmann.de
(K.-H. Herrmann).
http://dx.doi.org/10.1016/j.medengphy.2014.06.008
1350-4533/ 2014 IPEM. Published by Elsevier Ltd. All rights reserved.

ticularly as these items can, in our experience, slowly move or shift

under their own load and under the inuence of scanner vibrations
during the experiments. Small animal MRI is another application
increasingly performed on clinical scanners [1,2], which requires,
apart from using special dedicated animal coils, sufcient yet careful immobilization of the animals [3,4]. Specically, due to the
required sub-millimeter imaging resolution, even small motions
can result in imaging artifacts [5,6].
Several of these limitations could be overcome by specially
designed appliances, which exactly t the experimental setup.
However, special unique xation devices or holders frequently
require individual manufacturing and the production with turning lathes or CNC lathes is usually time consuming and expensive
as this manufacturing process is not well suited to produce
unique specimen. For this reason, rapid prototyping has become an

1374

K.-H. Herrmann et al. / Medical Engineering & Physics 36 (2014) 13731380

Fig. 1. In (a) the Ultimaker 3D printer is shown. The red PLA material (arrow 1) is pushed by a stepper motor (arrow 2) through the tube into the print head (arrow 3). The
print head melts the PLA material while it moves in the xy plane (metal rods) and the build platform (blue surface, arrow 4) moves downwards (z-direction). The object
(arrow 5) is printed by adding PLA material layer by layer. The printer is controlled by an Arduino board (mounted under the printer) and the optional UltiController (arrow 6)
provides an interface to operate the printer without a controlling Personal Computer. In (b) a screenshot of the preprint visualization of the slicer program Cura is presented
where the print process is frozen at a single layer. The pre visualization includes all details of the object and all additional supporting structures like the skirt line to start
the material ow (1), the object with inll of 40% (2) and overhang support structures (3, 4). In (c) a snap shot during the actual printing process is shown: The skirt line (1),
object with inll (2), support structures (3, 4). The heated print head (5) keeps the temperature at approximately 210 C, the nozzle extrudes the material with a diameter of
0.4 mm (6). (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)

integral part in research and development departments where

the recent advances in 3D printing technology [7,8] enable direct
manufacturing of functional, detailed and complex prototypes in a
fast and efcient way.
Recent hard- and software developments have led to a range
of designs for 3D printers using fused deposition modeling (FDM)
[7], which are meanwhile available as self-assembly kits in the price
range of 5002000 D [9,10]. The design of these 3D printers is based
on a print head that acts like an xyz plotter. The raw material,
most often acrylonitrile butadiene styrene (ABS) or biodegradable
polylactic acid (PLA), is molten in the print head and deposited layer
by layer on the build table, creating the 3D object [7,8].
The aim of this work was to design, create and test MRI compatible holders as well as a xation device for mouse imaging on a
clinical MRI system by using a low budget 3D printer [11,12]. The
process from the digital representation to the nal printed object
is described and illustrated.

2. Materials and methods

The main steps in creating a 3D printed object are to design
and dene the object with the aid of computer aided design (CAD)
software prior to conversion into a toolpath description for the 3D
printer, which is accomplished by so called slicer programs.
2.1.1. 3D printer
All objects in this work were printed on an Ultimaker 3D printer
(Ultimaking Ltd., 4191PL Geldermalsen, Netherlands), which was
assembled from a kit containing all necessary parts and tools

(Fig. 1a). The additional UltiController kit allows to operate the

printer in stand alone mode without a connected PC.
The Ultimaker printer is equipped with a single print head
(extruder) allowing use of only one material at a time. The build
table moves vertically along the z-axis, while the print head moves
horizontally in x and y direction. The available build volume is
limited to 210 mm 210 mm 205 mm. Although the positioning of the print head can theoretically achieve a resolution of
0.0125 mm, the actually printable structures are limited by the nozzle diameter of 0.4 mm. The layer thickness can be set to 0.1 mm
for ne details and good print quality.
Without a heated build table there are known warping issues
[13] that are most severe for elongated, rectangular shaped objects.
The warping is caused by a temperature gradient while the print
layers are deposited and cool down and manifests itself primarily at
the contact surface of the printed object and the build table. Vertical
structures and the upper side are usually not affected by warping.
Since the geometry of the object plays an important role choosing the most suitable printing direction and reducing the thermal
stress by adapting the object design can signicantly reduce warping issues [1315].
The quality of the nal print result depends also strongly on the
slicer program, which divides the object into digital cross-sections
so the printer is able to build it layer by layer (see Fig. 1b and c)
and has to take care of all physical and hardware related issues,
e.g., nding optimal toolpaths for an efcient print process, creating additional support structures or enforcing minimal cool down
time for the last printed layer. We used the software Cura1 (version

http://software.ultimaker.com/.

K.-H. Herrmann et al. / Medical Engineering & Physics 36 (2014) 13731380

13.06) developed by the printer manufacturer (Ultimaking Ltd.) and

provided as open source software2 . Before printing the 3D object
the generated toolpath can be veried in Cura (Fig. 1b). The red and
green lines mark the solid walls of the object while the brown grid
(Fig. 1b, arrow 2) is the inll, i.e., the structures inside the enclosing walls. This toolpath visualization can be quickly tracked and
veried for all layers of the object. The slicer program also detects
critical overhangs, which would have to be printed without any
underlying supporting structure to deposit the new layer on. The
software Cura automatically creates support structures (arrow 3
and 4 in Fig. 1b and c) for these overhangs.
2.2. MRI compatible 3D printed items
In the following we present two examples of 3D printed support
equipment, which were designed for use with a clinical 3T MRI
system (TIM Trio, Siemens Healthcare, Erlangen, Germany).
2.2.1. Validating MRI compatibility
To be MRI safe the material has to fulll several criteria as
described by [16,17]. With regard to the static magnetic eld PLA
can be considered MR safe, since PLA is not ferro magnetic. Additionally, PLA can be considered quite safe with respect to gradient
or rf heating since the volume resistivity was determined to be
> 5 108  cm, which is two orders of magnitude larger then that
of other common MR safe materials like PE or Polyester [18]. Beyond
the pure safety aspects, to ascertain MR compatibility and artifact
free imaging the susceptibility of PLA must be close to biological tissue. To assess susceptibility artifacts caused by PLA a multi
echo gradient echo experiment with 10 echoes from TE = 5.2 ms
to TE = 70 ms, an isotropic voxel size of (0.42 mm)3 and TR = 78 ms
was performed, comparing differently colored PLA and ABS samples
with a diameter of 2.9 mm and an air lled tube placed perpendicular to B0 in a bowl of water.
2.2.2. Flat platform for small loop coil
The rst object is a large at platform to provide a smooth, stable surface for use with a vendor supplied small loop coil (Siemens
Healthcare, L4). The loop coil ts into the opening of the platform,
which has the same height as the coil, so that anything placed on
the platform will be in plain contact with the coil. The coils cable
is covered by the platform to provide a single large at surface
area. The platform was designed using the web application Tinkercad (tinkercad.com), which works by Boolean combination of basic
geometric shapes. In contrast to other more complex and full featured CAD programs Tinkercad is easy to use and offers a shorter
learning curve.
2.2.3. Mouse xation
For small animal MRI we use a dedicated, cylindrical linear
Litzcage coil with an inner diameter of 38 mm and a length of 50 mm
(Doty Scientic Inc., Columbia, SC, USA) [19,20]. The coil is mounted
on a long, narrow platform which provides basic support but no
further xation for the animal.
To improve mouse brain imaging a support bed and head xation were designed, which tted onto the coil platform and into
a cylindrical mounting block (inner diameter 20 mm), which was
added to the platform with a screw mechanism to lock the mouse
bed. Since the anesthetic gas has to be delivered to the animal
and physiological parameters like respiration, heart rate and body
temperature have to be monitored the xation provides internal
channels to equip the animal with sensors as described in [21] without obstructing the experiments setup or the animal exchange.

https://github.com/daid/Cura/.

1375

The individual parts of the mouse xation are shown in Fig. 2.

The cylinder shown in Fig. 2a connects to the mouse bed with a 90
twist-lock and is used to clamp the assembly to the coil platform.
Furthermore, this cylinder is itself hollow providing a channel to
insert the bite bar (Fig. 2c), which can be moved back and forth
by the attached nger grips to adjust the bite bar position for the
mouse. The bite bar is again hollow to deliver the anesthetic gas.
The mouse bed shown in Fig. 2b is the central part accommodating
the biomonitoring sensors, providing mouse support and a hinge
to insert and rotate the head xation mask (Fig. 2d). In the hinged
down, closed position the head mask and the mouse bed are locked
together by the ratchet teeth (Fig. 2d) and are tted into the coil
with a small tolerance, which prevents the mask from opening and
therefore keeps the ratchet locked during measurements. The Ttongue at the far end of the mouse bed (Fig. 2b) is inserted into a
T-groove on the coil platform, thereby stabilizing the mouse bed
and reducing vibrations.
For this appliance the CAD software SolidWorks 2012 (DassaultSystmes SolidWorks Corp., Waltham, MA, USA) was used to create
the complex 3D models. In contrast to simpler CAD software like,
e.g., openscad or Tinkercad, more complex CAD programs like SolidWorks offer a large array of complex design features e.g. allowing
for extrusion, lofting and sweeping of complex and curved geometries.
2.2.4. Mechanical measurements
To evaluate the accuracy of the printed parts the dimensions
were measured with a digital caliper (Burgwchter, proscale Precise PS 7215).
2.3. In vivo MRI validation experiment
To validate the improvements in image quality by using the
mouse bed experimental scans of an adult mouse (C57Bl/6) were
performed using a Litzcage coil and a 3D, T2-weighted, turbo-spin
echo sequence (SPACE sequence) [20,2224] with isotropic resolution of 0.2 mm, TE/TR = 287 ms/2500 ms, bandwidth of 145 Hz/pixel
and a rectangular eld of view with matrix size 256 184. The
mouse imaging was repeated once without the holder, placing the
mouse head inside the Litz coil, and once using the PLA mouse xation. The resulting images are compared and a ROI based analysis
is used to calculate the SNR of brain tissue, cerebrospinal uid and
fat tissue outside the brain. The SNR value is estimated by dividing
the mean of a signal region by the standard deviation of a homogeneous background region, which is a reasonable estimate for
relative comparisons in the case of unaccelerated single channel
acquisitions.
3. Results
All designed parts were successfully printed on the Ultimaker
3D printer. The loop coil platform (Fig. 4) was designed and printed
overnight, demonstrating one major advantage of having a 3D
printer available in the lab. The more complex mouse bed (Fig. 5)
took longer to design and to print. A summary of all the parts, their
weight and print time is given in Table 1.
3.1. MRI compatibility validation
The gradient echo image acquired at TE = 70 ms is shown in Fig. 3.
Only for this very long echo time the susceptibility artifacts, caused
by the PLA samples, become visible. The susceptibility difference
between PLA and water creates signal voids, which increase the
diameter of the PLA sample from 2.9 mm to apparent 4.0 mm. The
apparent diameter of the ABS sample is 3.7 mm and that of the air
lled tube 3.54.5 mm, depending on the direction.

1376

K.-H. Herrmann et al. / Medical Engineering & Physics 36 (2014) 13731380

Fig. 2. Individual parts of the mouse bed rendered in the CAD software Solidworks 2012 (www.solidworks.com). To assemble the mouse xation the two locking pins of the
cylinder (a) are locked with a 90 turn into the mouse bed (b). The bite bar (c) is inserted into the cylinder and is itself hollow to supply the anesthetic gas to the mouse. The
nger grips are clipped into both sides of the bite bar and are then used to position the bite bar. The xation mask (d) is inserted into the hinge of the mouse bed (b) and,
after the mouse is placed on the bed, this mask is hinged down to xate the mouses head. The tube of the pressure sensor, which monitors physiological parameters, and
the glass ber temperature sensor are inserted in the mouse bed (b).

3.2. Flat platform for loop coil

Since the platform has a large at contact area with the build
table, holes were designed into the bottom of the 3D model (Fig. 4,
right), which reduced the thermal stress and warping. Only in the
outermost bottom corners an upward bend of approximately 1 mm
was observed. The weight, and therefore material cost, of the platform was reduced by using a low inll percentage of only 20% (see
Table 1). The additional solid walls and the internal structuring of
the holes also made the platform stiff and stable.
3.3. Mouse xation
The mouse bed, cylinder, bite bar and mask were printed
upright, i.e. the largest dimension of each part was printed in zdirection, which made warping issues negligible due to the small
foot print area in the xy plane. However, the connection to the
build table should not become too small or the object might come
loose during printing. The Slicer software Cura allows to add additional grid structures, so called rafts or brims, to create a larger

adhesion area to keep the object stable during the print. This feature was utilized for some parts (see Table 1). All these more or
less cylindrical parts were designed with rounded edges and as few
overhangs as possible to minimize the necessity of additional support structures. As a result some of the complex shaped parts, like
the entire mouse bed, could be printed without any support structures, leading to smooth surfaces right off the build platform, as can
be seen in Fig. 5a and b. Even ne structures like the locking teeth
on the mask and the mouse bed were printed accurately and did
not require any manual post processing.
The channel to accommodate the pressure pad tube for respiratory monitoring and especially the curved channel housing the
glass ber temperature sensor were ready for use directly after the
print with no additional clean-up procedures necessary. The parts
were designed to t tightly into each other or into the coil. For
example the mouse bed was designed for 37 mm outer diameter to
t into the 38 mm inner diameter coil and the cylinder clamp was
designed for 19.8 mm to t the 20 mm clamp. The caliper measurements of the printed mouse bed and mask were 36.8 0.4 mm and
the cylinder was 19.6 0.1 mm.

Table 1
Summary of all printed parts with details on some print parameter, the weight of each part and the required printing time. The cost of the used PLA material is approximately
33 D per kg.
Part name

Supports

Raft

Loop coil platform

No

No

Internal ll
20%

Mouse bed
Bed
Guide rail platform
Mask
Cylinder
Bite bar
Fingergrip left
Fingergrip right

No
No
Yes
Yes
No
Yes
Yes

No
No
No
Yes
Yes
No
No

40%
20%
40%
100%
100%
100%
100%

Part weight

Print time

120 g

10 h

Total 135.1 g

Total 18.5 h

49 g
39 g
8.8 g
29 g
7.3 g
1.0 g
1.0 g

7.25 h
3.5 h
2.5 h
4h
45 min
15 min
15 min

K.-H. Herrmann et al. / Medical Engineering & Physics 36 (2014) 13731380

1377

Fig. 3. Shown is the last echo of a multi echo gradient sequence at TE = 70 ms. The
phantom consists of 4 PLA samples (transparent, transparent blue, opaque red and
white), two ABS samples (opaque black and red) as well as an air lled PE tube
all with an outer diameter of 2.9 mm immersed in deionized water. No differences
between the colors could be observed. Both ABS and PLA show no MRI signal even
for the shortest acquired echo time of 5.2 ms.
Table 2
In vivo evaluation of image SNR with and without the mouse bed xation.
ROI area

Deep brain
Bulbus olfactorius
Cortical gray matter
Ventricle right
Ventricle left
Fat tissue

SNR
No holder

Mouse bed

11.4
15.3
13.3
32.0
29.7
7.1

21.4
27.8
24.7
71.1
72.2
28.6

3.4. In vivo MRI validation

Fig. 6 shows the results of two MRI scans conducted with (a) and
without (b) the xation. Although the mouse was anaesthetized,
without xation the subtle respiratory motion caused signicant
motion artifacts and reduced signal in tissue (Fig. 6b), which is also
reected in the SNR values given in Table 2. The image quality is
substantially improved with the mouse bed and xation (Fig. 6a,
Table 2). Without head xation (Fig. 6b) some neck tissue regions
show distinct signal loss and signal bands outside the mouse that
increase the noise in the images, which are otherwise identically
windowed. The only identiable fat tissue region in Fig. 6b has

Fig. 5. The printed and fully assembled mouse bed. (a) The mouse bed with the ber
optic temperature sensor, pressure pad for respiratory and cardiac monitoring, the
bite bar with its little hole for the anesthetic gas supply, the opened head xation
mask with the locking ratchet and the T-groove platform, which supports the tail
end side of the mouse bed. (b) The assembly is shown with mouse and closed xation
mask, positioned just before insertion into the Doty coil. (c) The complete assembly
of the 3D printed mouse xation (all red parts) mounted on the original coil platform
with the added mounting block (white, 5). The main parts are the T-groove platform
(1), which supports the tail end of the mouse bed (2). The head xation mask is
positioned inside the Doty coil (3). The cylinder (4) is screw locked by the white
mounting block (5). Inside the cylinder (4) the bite bar can be moved using the nger
grips (6) and the anesthetic gas is supplied through the bite bar. All three leads, the
ber optic temperature sensor (7), the pressure pad tube (8) and the gas supply tube
(9), are connected from the front of the coil assembly, which eases handling. (For
interpretation of the references to color in this gure legend, the reader is referred
to the web version of this article.)

Fig. 4. The printed loop coil platform together with the loop coil. The left image shows a view from top, the right image from the bottom exposing the additional holes, which
were added to reduce warping effects.

1378

K.-H. Herrmann et al. / Medical Engineering & Physics 36 (2014) 13731380

Fig. 6. Orthogonal planes of 3D T2-weighted MR images of an in vivo mouse head acquired with the Doty coil on a clinical 3T scanner with (a) and without (b) using the
mouse bed xation. Without the head xation (b) the visually perceived signal to noise ratio of the images is lower and some of the soft tissue in the neck and head suffer
from severe signal loss due to motion. Ghosting artifacts are also much more pronounced without using the xation. Also note the straighter head position in (a, top) and
the more S-shaped neck in (b, top).

a factor 4 lower SNR (Table 2) compared to the images acquired

with the mouse bed. Note also the improved straight head and
neck alignment with the xation, thereby facilitating reproducible
positioning in longitudinal studies.

4. Discussion
The possibilities of current 3D printers using FDM print technology [10] are quite intriguing, especially with respect to MRI as
the materials used are inherently MR compatible. The versatility of
the 3D printing process enables a wide variety of applications, from
holders and appliances, as shown here, to phantoms [25] and even
printing of implantable scaffolds for tissue regeneration [26]. Simple geometries, like the loop coil platform, can be quickly designed
and printed overnight. More complex designs like the multipart
mouse bed require more time for the actual conception and creation of the CAD models but can be realized within several days. One
major advantage of a locally available 3D printer is a very fast and
easy print, test and redesign cycle until all parts are optimized and
t together. The nal print of all mouse bed parts was completed
within 2 work days.

The mouse bed represents an excellent demonstration of the

capabilities of 3D printing, since even tiny details, like the saw
tooth ratchet, or internal structures, like the temperature sensor
canal with 3 mm diameter, could be printed successfully. In particular, the possibility to create objects with complex shaped internal
hollow structures is a major advantage of additive manufacturing
techniques compared to traditional, subtractive methods, like CNC
lathes [8]. The parts of the mouse xation were mostly printed
upright, with a small circular footprint on the build table, thereby
rendering the warping problem negligible and resulting in accurate
dimensions ( < 0.4 mm) for all parts.
Since our version of the Ultimaker 3D printer (delivery date
March 2013) is not equipped with a heated build table, temperature gradients in the printed object can cause warping [1315].
However, printing all parts of the mouse bed vertically rendered all
warping issues negligible. The loop coil platform and the T-groove
platform for the mouse bed suffer from some warping, however,
in the case of the T-groove platform this does not matter as it only
affects the outer edges while the guide rail is mounted on the 6
posts. Also, the stability of the loop coil platform does not suffer by
the upward bend of the outer most edge. If warping poses a serious
problem, a heated build table can the installed, which reduces the

K.-H. Herrmann et al. / Medical Engineering & Physics 36 (2014) 13731380

origin of the warping, i.e. the temperature gradient between build

table and rst print layers. Also to determine optimal print parameters some initial test prints were required. Since one parameter
can inuence the optimal settings of others, particularly the object
size, print speed and layer thickness had to be optimized together
with the print temperature for optimal results [27].
The readily available printing material PLA consists of polymerized lactic acid, which is biocompatible [28] and was already
successfully used as biodegradable scaffold for implants [26,29].
For clinical use the material is resistant to common disinfectants
but cannot be sterilized in an autoclave as PLA will become very
soft above 60 C. In dry conditions and at room temperature printed
parts are stable over years [28]. Also, due to its biological origin, the
susceptibility of PLA is very close to biological tissue, which was veried by the susceptibility sensitive gradient echo sequence (Fig. 3).
In order to design and build customized appliances quickly and
cheaply, 3D printing technology is a viable choice. The manufacturing costs, after an initial investment of approximately 1500 D for the
printer, are only the costs for the raw material, which for PLA is currently 33 D per kg. For example, the material costs of the loop coil
platform or the complete mouse xation were less than 5 D each.
In contrast, commercially available MRI appliances, or more generally clinical and laboratory equipment, are often expensive. Taking
these price differences into account, the 3D printer itself might be
amortized within very few projects [12].
Choosing 3D printing technology introduces some design
constraints but also opens design possibilities only achievable
with additive manufacturing techniques. For example, using the
classic turning lathe approach, all parts will be rotationally symmetric, leaving less room for individual, anatomical adaptations.
Furthermore, CNC lathes will also encounter their particular limits
once internal details or structures within an object are necessary
or desired.
One advantage of the CADCAM concept is also the fact that a
generic digital 3D CAD model is created rst. This can be realized
by any number of manufacturing processes including a local 3D
printer. Here a locally available 3D printer facilitates a rapid iterative cycle of development, creation, testing and renement of the
CAD models. If the nal print quality on a low budget printer should
indeed prove insufcient due to limitations of the particular 3D
printer or the print technology, the same CAD le can still be used
to create the models with other, more sophisticated manufacturing
techniques. The number of companies offering 3D print services
is steadily growing with a much wider variety of MR safe materials (e.g. nylon, ceramics, etc.) and print technologies like stereo
lithography or laser sintering, yielding much higher accuracy and
isotropic resolution [7,8].
Another advantage of the low-end 3D printing technology arises
from its origin in free and open source software and hardware
development [12]. There exists a large community that is supportive in modifying and improving the printers. Furthermore, a vast
number of CAD models are available under open creative commons
licenses which can be used as a starting point, be improved and
again shared with the community. For example, there are a growing number of 3D models available for optical laboratories [12]. Our
designs for the loop coil and the mouse xation are also available
as supplementary material to this article.
In conclusion, our experiences with a low-budget 3D printer
are quite encouraging, especially for applications in an MRI lab. The
materials are not only cheap, but entirely MR safe, invisible to standard MRI, and can therefore be used without constraints in combination with phantoms, animals or volunteers. Although design and
printing both require time, the possibility to create individual appliances directly in the lab within reasonably short time periods constitutes an immense advantage including cost effectiveness compared to outsourcing both the design and manufacturing process.

1379

Ethical approval
This work did not involve any humans, so no ethical approval is
required. All animal experiments were performed in accordance to
the European Convention for Animal Care and Use of Laboratory
Animals and with approval of the local animal welfare commitee
and the Thringer Landesamt fr Verbraucherschutz.
Acknowledgment
This work was nancially supported by the Carl Zeiss Foundation with a dissertation fellowship (MK).
Appendix A. Supplementary data
Supplementary material related to this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.medengphy.
2014.06.008.
Conict of interest statement
All authors declare that they have no competing interests.
References
[1] Yamamoto A, Sato H, Enmi J, Ishida K, Ose T, Kimura A, et al. Use of a clinical MRI
scanner for preclinical research on rats. Radiol Phys Technol 2009;2:1321.
[2] Haenold R, Herrmann KH, Schmidt S, Reichenbach JR, Schmidt KF, Lowel S, et al.
Magnetic resonance imaging of the mouse visual pathway for in vivo studies
of degeneration and regeneration in the CNS. Neuroimage 2012;59:36376.
[3] Lemieux SK, Glover GH. An infrared device for monitoring the respiration
of small rodents during magnetic resonance imaging. J Magn Reson Imaging
1996;6:5614.
[4] Minard KR, Wind RA, Phelps RL. A compact respiratory-triggering device
for routine microimaging of laboratory mice. J Magn Reson Imaging
1998;8:13438.
[5] Kokuryo D, Kimura Y, Obata T, Yamaya T, Kawamura K, Zhang MR, et al. A small
animal holding xture system with positional reproducibility for longitudinal
multimodal imaging. Phys Med Biol 2010;55:411930.
[6] Boretius S, Kasper L, Tammer R, Michaelis T, Frahm J. MRI of cellular layers in
mouse brain in vivo. Neuroimage 2009;47:125260.
[7] Upcraft S, Fletcher R. The rapid prototyping technologies. Assem Autom
2003;23:31830.
[8] McMains S. Layered manufacturing technologies. Commun ACM 2005;48:506.
[9] Knig P. Zauberksten. ct Heise Zeitschriften Verlag 2012;11:92101.
[10] Valentan B, Pogacar D, Brajlih T, Hartner TZ, Pilipovic A, Drstvensek I.
Development of a 3D printer for thermoplastic modelling. Mater Technol
2012;46:58994.
[11] Pearce JM. Building research equipment with free, open-source hardware. Science 2012;337:13034.
[12] Zhang C, Anzalone NC, Faria RP, Pearce JM. Open-source 3D-printable optics
equipment. PLoS ONE 2013;8:e59840.
[13] Wang T, Xi J, Jin Y. A model research for prototype warp deformation in the
FDM process. Int J Adv Manuf Technol 2007;33:108796.
[14] Pei E, Campbell RI, de Beer D. Entry-level RP machines: how well can they cope
with geometric complexity? Assem Autom 2011;31:15360.
[15] Kantaros A, Karalekas D. Fiber Bragg grating based investigation of residual
strains in ABS parts fabricated by fused deposition modeling process. Mater
Des 2013;50:4450.
[16] Shellock FG, Crues JV, Karacozoff AM, editors. MRI bioeffects, safety, and patient
management. Playa Del Rey, CA: Biomedical Research Publishing Group; 2014.
[17] ASTM Standard F2503. Standard practice for marking medical devices and
other items for safety in the magnetic resonance environment. West Conshohocken, PA: ASTM International; 2013, http://dx.doi.org/10.1520/F2503.
www.astm.org
[18] Shinyama K, Fujita S. Study on the electrical properties of a biodegradable plastic. In: In Proceedings of the 7th Intern. Conf. on Properties and Applications of
Dielectric Materials. 2003. p. 70710.
[19] Doty FD, Entzminger G, Kulkarni J, Pamarthy K, Staab JP. Radio frequency coil
technology for small-animal MRI. NMR Biomed 2007;20:30425.
[20] Herrmann KH, Schmidt S, Kretz A, Haenold R, Krumbein I, Metzler M, et al.
Possibilities and limitations for high resolution small animal MRI on a clinical
whole-body 3T scanner. MAGMA 2012;25:23344.
[21] Herrmann KH, Pfeiffer N, Krumbein I, Herrmann L, Reichenbach JR. MRI
compatible small animal monitoring and trigger system for whole
body scanners. Z Med Phys 2013, doi:pii: S0939-3889(13)00092-5.
10.1016/j.zemedi.2013.07.005.
[22] Lichy MP, Wietek BM, Mugler JP, Horger W, Menzel MI, Anastasiadis A,
et al. Magnetic resonance imaging of the body trunk using a single-slab,

1380

K.-H. Herrmann et al. / Medical Engineering & Physics 36 (2014) 13731380

3-dimensional, T2-weighted turbo-spin-echo sequence with high sampling

efciency (SPACE) for high spatial resolution imaging: initial clinical experiences. Invest Radiol 2005;40:75460.
[23] Busse RF, Brau AC, Vu A, Michelich CR, Bayram E, Kijowski R, et al. Effects of
refocusing ip angle modulation and view ordering in 3D fast spin echo. Magn
Reson Med 2008;60:6409.
[24] Gaser C, Schmidt S, Metzler M, Herrmann KH, Krumbein I, Reichenbach JR, et al.
Deformation-based brain morphometry in rats. Neuroimage 2012;63:4753.
[25] Yoshimaru E, Totenhagen J, Alexander GE, Trouard TP. Design, manufacture, and
analysis of customized phantoms for enhanced quality control in small animal
MRI systems. Magn Reson Med 2013, http://dx.doi.org/10.1002/mrm.24678.

[26] Serra T, Mateos MA, Planell JA, Navarro M. 3D printed PLA-based scaffolds: a versatile tool in regenerative medicine. Organogenesis 2013;9:
23944.
[27] Anitha R, Arunachalam S, Radhakrishnan P. Critical parameters inuencing the
quality of prototypes in fused deposition modelling. J Mater Process Technol
2001;118:3858.
[28] Guo C, Cai N, Dong Y. Duplex surface modication of porous poly(lactic acid)
scaffold. Mater Lett 2012;94:114.
[29] Mkitie AA, Korpela J, Elomaa L, Reivonen M, Kokkari A, Malin M, et al. Novel
additive manufactured scaffolds for tissue engineered trachea research. Acta
Otolaryngol 2013;133:4127.