|Home | About | Journals | Submit | Contact Us | Français|
To develop a process for rapidly and inexpensively producing customized animal handling devices for small animal imaging.
To meet the specific needs of a particular imaging experiment, measurements are taken from imaging data, and the animal handling devices are designed using 3D computer-aided design (CAD) software. Parts are produced in a few days using solid freeform fabrication (SFF, aka rapid prototyping).
This process is illustrated with the production of an animal handling system for stereotaxically prescribed therapeutic ultrasound and MRI of the mouse brain. The device provides integrated head-fixation, anesthesia delivery, and physiological monitoring in a modular system. Design and production took approximately one week and the cost was a small fraction of a traditional machine shop.
Commercial animal handling products typically have limited functionality and are not integrated with other laboratory infrastructure. However, using CAD and SFF, sophisticated animal handling devices can be produced to meet the specific experimental needs. This process is typically faster and less expensive than using a traditional machine shop, and the products are more robust than typical homemade devices. Using high-quality purpose-built devices permits experiments to be executed with greater consistency and higher throughput.
In the development of a small animal MR imaging technique, the technical aspects of acquiring data (e.g., pulse sequences) necessarily receive the greatest attention, while animal handling is only a minor concern. However, once a technique is established and is being used to study animal models of human disease, the experiment depends critically on the quality of animal handling. Subtle, biologically relevant experiments require 1) large numbers of animals to be processed efficiently (1); 2) reliable and repeatable positioning of the animal; and 3) precise and consistent anesthesia and monitoring (2, 3).
Animal handling needs are specific to each imaging experiment. For example, blood oxygen level-dependent (BOLD) fMRI in the rat (4, 5) requires very different positioning, anesthesia, and monitoring than combined MRI/PET of flank tumors (6). Because the applications are specialized, no general solution exists for animal handling. In our experience, commercially available systems rarely meet the particular needs of an experiment and do not integrate well with other laboratory components (e.g., animal monitoring transducers). Consequently, when developing a new imaging protocol, researchers often modify existing devices or build their own. While some laboratories have the facilities to produce high-quality devices, home-built devices typically lack sufficient robustness for long-term use. When the imaging technique is extended to large numbers of animals in a subtle biological study, these awkward systems are inconsistent, unreliable, and slow.
High-quality, purpose-built animal handling devices can be produced by working with a machine shop (7-10). However, for sophisticated devices that closely fit the animal anatomy and incorporate anesthesia and monitoring, the cost and time for custom-machined parts is extremely high. The small animal imaging community needs a fast and inexpensive process for producing sophisticated animal handling devices tailored to the needs of a particular experiment. Here, we describe such a process in which measurements are taken from anatomical images of the animal, devices are designed in 3D computer-aided design (CAD) software, and parts are produced by a solid freeform fabrication (SFF, also known as rapid prototyping) service. While this process is commonly used in the product design community, it has yet to be fully harnessed by the research community. By adopting this process, we are able to have high-quality, affordable devices that specifically meet the needs of the experiment and result in higher throughput, more consistent imaging, and more reliable results.
Our process for producing custom devices includes the following steps:
We have used this process to develop over a dozen highly specialized animal handling devices ranging from tiny racks for ex vivo MR microscopy of mouse kidneys to large cradles for in vivo MRI of the rat. Here, the steps will be illustrated concretely with a specific example—the development of an animal handling device for therapeutic ultrasound and subsequent in vivo MRI of the mouse brain. The goal was to ultrasonically open the blood-brain barrier in specific focal locations, thus enabling the focal administration of MRI contrast agent to the brain parenchyma. For this experiment, the device needed to fix the brain in the alignment with the stereotaxic brain atlas (11) for both ultrasound and subsequent MRI.
The first step in the process is to enumerate all the constraints and requirements of the device. While this seems obvious, it merits special consideration: the computer-aided design and manufacture permit a level of nuance and flexibility not available in traditional small-volume manufacturing. Furthermore, while complicated extra features typically add great cost at a machine shop, they have little effect on the price in SFF. Thus, to fully exploit the power of these technologies, the researcher should include every feature that will facilitate the experiment. For example, to maximize throughput, we wanted to build all of the monitoring sensors into the animal cradle. To insert the temperature probe into the cradle required a long hole with a bend—an impossible feature for a typical machine shop, but a trivial feature for SFF.
For the example device, we began by listing the requirements (shown below) with special attention to animal’s positioning, anesthesia, and monitoring. Equally important were the physical constraints imposed by the imaging system, such as magnet compatibility and coil size.
High-quality animal handling devices need to fit the anatomy with great precision. For example, a head holder that precisely fixes the boney structures of the skull saves setup time, improves reproducibility, and prevents transmission of respiratory motion to the brain. In contrast, a poorly fitting head holder simply causes frustrating delays and complicates the experiment. Fortunately, most animal imaging laboratories have a wealth of anatomical image data that can be used to guide the design of animal handing devices.
In this example, the boney structures were studied using planar x-ray images (Kodak In-Vivo FX, Eastman Kodak, Rochester, NY). From these images (Fig. 1), it was determined that the head could be reliably and firmly held in place, while still allowing open access to the top of the brain, by clamping the skull between the nasal bones and the premaxilla. Measurements were taken from 7 mice ranging in size from 24-29 g. The orientation of the dorsal cranium with respect to the nasal bones and premaxilla was measured to be 22° (SD ± 2°) and 0° (SD ± 1°), respectively. In a similar fashion, whole-body localizer MR images (not shown) were acquired and used to consider the relative positioning of the body and head to avoid kinking or obstructing the animal’s airway.
The device is designed using 3D CAD software, which allows the creation 3D parts with geometries of arbitrary complexity. The geometries can be then used to produce traditional engineering drawings, or the geometries can be exported digitally for computer-controlled manufacturing. Typical file sizes are 0.5 to 4 MB, depending on the file format and geometry complexity.
The device for our example was designed using the program, Inventor (Autodesk, San Rafael, CA). The devices was designed as three separate parts (shown in the CAD model in Fig. 2): 1) a drawbar (green) that fits in the mouth and delivers anesthesia to the nose; 2) a nose cone (red) that fits over the drawbar and clamps the nasal bones; and 3) a cradle (white) that holds the body and the physiological monitoring transducers. The drawbar ends in a small platform that fits against the premaxilla inside the mouth. A small hole in the platform catches the incisors and keeps the head pulled caudally into the nose cone. The drawbar and a small projection on the roof of the nose cone together act as a vise to clamp the caudal-most portion of the skull. The roof of the nose cone has a projection that precisely meets the nasal bones. A fulcrum on the inside of the nose cone allows this projection to be clamped against the nose bone by tightening a screw. (A corresponding fulcrum and hole on the bottom allows the same nose cone and drawbar to be used when the mouse is in the prone position.) Importantly, this mechanism does not depend on ear bars, which do not provide reliable fixation in the mouse and cause deformation of the skull.
The distal end of the drawbar fits into a bracket on the animal cradle, which slides into a 35mm volume radiofrequency coil and holds the animal in the center of the coil. Slots in the cradle accommodate transducers for respiratory, ECG, and temperature monitoring (Fig. 3 a). A circular pneumatic pillow (Graseby from SA Instruments, Stony Brook, NY) lies under the abdomen (left edge of Fig. 4). The ECG pads (Blue Sensor BRS, Ambu, Glen Burnie, MD) lie on vertical panels on either side of the head (Fig. 3 b). The hands of the mouse are affixed to the pads using the pad’s gel, a small piece of tape, or a reusable clip. The thermistor (Cole-Parmer, Vernon Hills, IL) runs through a channel in the cradle and can be placed either under the thorax (for abdominal imaging) or in the rectum (for brain and cardiac imaging).
Because the experimental protocol also required pre-imaging procedures, such as shaving of the head and placement of intravenous and intraperitoneal catheters, a procedure bed (not shown) was designed, which also had a bracket for the drawbar. The anesthetized mouse could be positioned on the bed for the pre-imaging procedures and subsequently transferred to the imaging cradle without interrupting anesthesia.
Prior to manufacturing, the design can be validated using visualization software to ensure that the parts will fit the specimen as expected. In this example, the drawbar and nose cone parts were imported in into the AMIRA visualization environment (Visage Imaging, Carlsbad, CA) with a publicly available digital mouse phantom (12) (Fig. 2c) to make sure the parts would fit the skull.
The final step in this process is production of the parts by solid freeform fabrication (SFF), also known as rapid-prototyping. There are many different SFF technologies, all of which share the same fundamental principle: parts are formed by computer-controlled incremental addition of material. This is in contrast to typical machining, where material is removed from a solid piece of raw material. Because the part is built in very small increments (as small as 0.015 mm), SFF can create extremely intricate parts that could not physically be machined with traditional techniques. Furthermore, in traditional machining, complicated parts are more expensive because they require more of the machinist’s time. However, because SFF is computer-controlled, complexity has little or no impact on cost.
When using SFF, the user must choose 1) the material used to fabricate the part and 2) the process by which the part will be fabricated. Most SFF services can assist in selecting the most appropriate combination for a particular application. For SFF materials, the manufacturer reports a variety of mechanical properties (e.g., tensile and flexural strength); however in our experience, these numbers are very poor predictors of a part’s durability. In our work, when durability is important but a fine finish is not important, we prefer the Selective Laser Sintering (SLS) process using a glass-filled nylon material (e.g., Duraform GF, 3D Systems, Rock Hill, SC). For most parts, we find that the Stereolithography (SLA) process using a durable polypropylene-like material (e.g., Somos 9420, DSM Somos, Elgin, IL; or Accura 25, 3D Systems) provides a good compromise between fine finish and durability.
A number of companies offer SFF services, including several web-based services that allow parts to be ordered quickly online. The user uploads the designs to the website and chooses the process and material. Then, the website immediately displays a price quote. For relatively small and complicated parts (such as those in this example) the cost is dramatically lower than a traditional machine shop. Furthermore, because several parts can be made on a given run of the SFF process, ordering multiples of a part or several parts together may add only marginally to the cost. Typically, the turn-around time is about a week. In our experience, this compares favorably to traditional machine shops where turn-around for even a simple part can take weeks. For this example, the parts were manufactured by Quickparts.com (Atlanta, GA) using stereolithography with a durable polypropylene-like material.
After the parts are received, a few minor modifications may be necessary, as determined by the design. In this case, the thumbscrew holes in the drawbar brackets had to be tapped and the sensors attached to the cradle (Fig. 3a). During the experiment, animal preparation time was short—from the isoflurane induction chamber to the radiofrequency coil took less than a minute. Consistent positioning and monitoring of the animal was achieved with little or no adhesive tape (Fig. 3b). Planar X-rays showed that the drawbar and nose cone clamped the nasal bones and premaxilla, thus holding the mouse skull precisely and firmly in the “skull-flat” orientation (Fig. 4). Most importantly, this new device enabled stereotaxically registered ultrasound and subsequent MRI (Fig. 5). To do this, the device was clamped into a three-axis positioning system fitted with the ultrasound transducer. Once calibrated, the positioning system could be used to precisely target different regions of the brain for ultrasonic blood-brain barrier disruption, as described in reference (13). This prescribed ultrasonic blood-brain barrier disruption enabled the delivery of agents (in this case a gadolinium chelate) to the brain parenchyma.
Because of their small size and accelerated physiology, imaging of small animals is difficult. Nevertheless, the small animal imaging community has overcome the technical challenges to develop a broad arsenal of small animal imaging techniques. In most cases, the techniques are intended for studying animal models of human disease. However, biomedically relevant studies look for subtle effects; therefore, these studies require large numbers of animals to be imaged in a highly consistent fashion. In executing these studies, often the limitation is not the imaging technology, but rather the associated animal support. For example, in a study of cardiac performance in a transgenic mouse, inconsistencies in animal positioning, anesthesia level, or body temperature could make the interpretation of imaging data difficult or impossible. By using high-quality animal handling devices designed specifically for a particular experiment, animal setup is efficient and imaging is consistent. Well-made animal handling devices enable sophisticated imaging techniques to answer more subtle biomedical questions.
In conclusion, because there is no general solution for animal handling—every experiment is different—commercial products rarely meet the needs of the animal imaging community. While some laboratories have the facilities to build high-quality devices, typical home-built devices lack the reliability and robustness for consistent, high-throughput imaging. Traditional machine shops produce high-quality devices, but are slow and very expensive. The alternative approach we have presented here—using CAD and SFF—offers a fast, convenient, and inexpensive way for the imaging community to obtain well-made, purpose-built animal handling systems. While the use of CAD and SFF is well established in the product design field, it has yet to be adopted by imaging researchers. We hope that this communication will increase awareness and adoption of such tools and serve as a useful example. To this end, our models and geometries are available via our web site <http://www.civm.duhs.duke.edu/JMRI200901/>.
All work was performed at the Duke Center for In Vivo Microscopy. The authors would like to acknowledge Kristin Frinkley Bing, with whom the ultrasound study was conducted; Laurence Hedlund, who developed previous animal handling systems at our Center; and Sally Zimney, who assisted with manuscript preparation.
Grant Support: NIH/NCRR National Biomedical Technology Research Resource (P41 RR005959) NCI Small Animal Imaging Resource Program (U24 092656)