|Home | About | Journals | Submit | Contact Us | Français|
Specific ventilation imaging (SVI) is a noninvasive magnetic resonance imaging (MRI)-based method for determining the regional distribution of inspired air in the lungs, useful for the assessment of pulmonary function in medical research. This technique works by monitoring the rate of magnetic resonance signal change in response to a series of imposed step changes in inspired oxygen concentration. The current SVI technique requires a complex system of tubes, valves, and electronics that are used to supply and rapidly switch inspired gases while subjects are imaged, which makes the technique difficult to translate into the clinical setting. This report discusses the design and implementation of custom three-dimensional (3D) printed hardware that greatly simplifies SVI measurement of lung function. Several hardware prototypes were modeled using computer-aided design software and printed for evaluation. After finalization of the design, the new delivery system was evaluated based on O2 and N2 concentration step responses and validated against the current SVI protocol. The design performed rapid switching of supplied gas within 250ms and consistently supplied the desired concentration of O2 during operation. It features a reduction in the number of commercial hardware components, from five to one, and a reduction in the number of gas lines between the operator's room and the scanner room, from four to one, as well as a substantially reduced preparation time from 25 to 5min. 3D printing is well suited to the design of inexpensive custom MRI compatible hardware, making it particularly useful in imaging-based research.
Specific ventilation (SV) is a dimensionless quantity that measures how efficiently a lung region is ventilated.1 The Pulmonary Imaging Laboratory at UCSD pioneered specific ventilation imaging (SVI), a technique that measures this metric spatially in the human lung using proton magnetic resonance imaging (MRI).2 This method quantifies SV by measuring the rate of change in the MRI signal of lung regions after the subject has undergone a step change in the fraction of inspired oxygen (FIO2). During the imaging sequence, research subjects are supplied air (21% O2) and 100% O2 in alternating cycles of 20 breaths while images of their lung are acquired after each expiration. The original method for supplying the inspired gas involved a complicated plumbing system consisting of a facemask, gas reservoir bag, nonrebreathing T-shaped three-way valve, large-bore tubes, and a pneumatically operated remote-controlled valve to achieve rapid switching between gases.
Numerous safety and space restrictions governed the design and operation of this system. The hardware inside the MRI room is subject to multiple safety precautions to prevent injury to the subject. The MRI uses a powerful magnetic field, which is active at all times and attracts ferrous materials toward the bore of the scanner with considerable force. Additionally, there are space restrictions for hardware that is worn or in contact with the subject, such as facemasks and nonrebreathing valves, due to the narrow space available inside the scanner bore. To ensure safety in the MRI environment, hardware placed in the scanner room must be constructed of plastics and nonferrous metals. Components worn by the subject must be able to fit inside the scanner bore without interference.
In this report, we discuss the design of an improved method of delivering and rapidly switching inspired gases in the magnetic resonance (MR) environment. Using three-dimensional (3D) printing-assisted rapid prototyping, we developed custom hardware that performs gas delivery and rapid switching without the use of expensive commercial hardware or electronics. 3D printing enabled inexpensive and speedy design iteration compared with traditional machining or purchase and assembly of commercially available products and provided the advantage of fabricating custom-designed hardware with MR safe materials. This resulted in a final design that is a simpler and more robust gas delivery method than the original plumbing system.
A diagram of the components of the original plumbing and improved systems for supplying inspired gas is displayed in Figure 1. A number of components are placed inside the operator's room due to MR compatibility restrictions. These include a compressed oxygen tank, compressed air tank, a flow meter (model 4830; Hans Rudolph, Inc., Shawnee, KS), and a pneumatic valve controller (model 4285; Hans Rudolph, Inc.). The components inside the MR scanner room include a gas reservoir bag (model 6170; Hans Rudolph, Inc.), a pneumatic switching valve (model 8500; Hans Rudolph, Inc.) placed outside the scanner bore, and finally a nonrebreathing T-shaped three-way valve (model 2600; Hans Rudolph, Inc.) connected to the facemask worn by the subject. The pneumatic switching valve has two inlets, one connecting the valve to the gas reservoir bag and the other open to room air. The outlet of this valve connects to the inlet of the facemask valve through a large-bore tube. The connection lines between the operator's room and scanner room components include a 1/4 inch tube connecting the oxygen tank to the gas reservoir bag, two 1/4 inch tubes connecting the pneumatic switching valve to its operating module, and a large-bore tube from the expiratory side of the facemask valve to the flow monitor, making the gas lines a closed circuit. These connections are fed through a small porthole in the wall of the scanner room called a pass through.
During operation, the subject inspires the supplied gas at room pressure, either through a gas reservoir bag prefilled with oxygen or through a low-resistance large-bore tube open to the air in the scanner room. The pneumatic valve controller used supplied tank air to actuate the switching valve between the inlets to these two reservoirs. A flow monitor is placed at the outlet of the system to record the subject's tidal volume, or volume of expired gas, used later during the computation of SV to correct for an inherent plumbing delay imposed by the scanner's physical constraints. The limited space in the scanner bore is also the reason why the flow monitor is placed in the expiratory line. The new replacement method of gas delivery was designed to reduce complexity while performing similar function with the original plumbing system.
The new design utilizes bypass flow to supply a stream of gas for subjects to breathe from during operation. To our knowledge, there are no commercial components readily available that perform this specific function and are MR compatible. Figure 2 displays a schematic drawing of the featured bypass flow attachment, which inserts into a standard facemask. The outlet of the bypass flow attachment remains open to the atmosphere at all times regardless of whether it is connected to a source or supplied by flowing gas. This new fail-safe is an important advantage over the previous plumbing system that ensures that the subjects are not in danger of having their breathing occluded in the event of a failure in the gas supply.
The bypass flow attachment was modeled on SolidWorks (Dassault Systèmes S.A., Velizy, France) computer-aided design software and printed on a Makerbot Replicator 2 (Makerbot Industries, Brooklyn, NY) 3D printer in polylactic acid (PLA). The model was sliced and prepared in G-code through the MakerWare software available with the printer. The settings used for printing were 0.2mm layer height, 15% internal fill, and a wall thickness of two shells (0.8mm). The total material used is 45g PLA and printing time is 3h and 11min.
The design has two modes of operation using one or two supplied gases. In the one-gas configuration, only 100% O2 is supplied during operation, with the subject breathing room air when O2 is not flowing. In the two-gas configuration, both O2 and another gas mixture are connected to the switching valve such that the subject is supplied either inspired gas at all times during scanning. This also makes it possible to use in studies involving other inspired gas mixtures, such as hypoxic and hyperoxic gas (e.g., 12.5% and 30% O2, respectively).
Polycarbonate 1/4 inch diameter tubing connects each tank to the inlets of the switching valve, which is clamped onto the edge of the operator's desk. The tubing connecting the switching valve outlet to the prototype was assembled from 30ft of 1/4 inch, 6ft of 3/8 inch, and 3ft of 1/2 inch diameter tubing connected with brass tube fittings. This provided adequate length to reach the subject from the console room and reduced noise due to gas flow. A converging nozzle is attached to the exhaust outlet of the bypass flow attachment to reduce noise and increase outlet flow to prevent inspiration of outside gases. The combination of expanding tube diameter and nozzle reduced noise due to gas flow from 123 to 98 dB, a level at which the subjects could still hear verbal instructions from the operator during scanning.
Figure 3 displays the featured hardware of the improved gas delivery system. A length of tubing connects the bypass flow attachment to the outlet of a Swagelok (Swagelok, Solon, OH) three-way switching valve in the console room. The inlets of this switching valve are then connected to up to two compressed gas tanks.
After each use, the 3D printed hardware was submerged in Cidex (Advanced Sterilization Products, Irvine, CA) cleaning solution for 20min for sterilization, rinsed with soap and water, then soaked in clean water overnight. While there is no direct contact between the part and the subject during operation, parts such as this are routinely sterilized either by high-level liquid disinfectant or gas sterilization. It was previously shown that PLA components can be cleaned with common disinfectants, although not by autoclaving as the PLA will become soft at temperatures of 60°C or higher.3 While no damage to the component surfaces or leaking was observed, the wall thickness and internal fill may be adjusted in subsequent prints to increase strength.
The functional requirement of the design can be described as providing a step function with fast rise time and low steady-state error. For the experiment to be performed properly, the switch between inspired O2 levels must occur rapidly such that the FIO2 presents a step change to the subject. The rise time is considered to be the time taken for the oxygen concentration to change from 0% to 90% of its final value (from 21% to 92% O2, respectively).
The previous plumbing system switched inspired gases with a significant delay due to the large volume (600mL) of the inspiratory line between the switching valve and the facemask. After switching, subsequent inspirations were of a mixture of gases until the line was cleared of the previous gas and filled with the new supplied gas. This hardware arrangement created a delay in the switching of FIO2 in the subject that had to be corrected for computationally after scanning. By supplying a bypass flow through small-bore tubing across the subject's airway, the inspired gas can be switched rapidly between breaths and the subject will experience a step change in FIO2 with fast rise time at the next inspiration.
The desired behavior of the design is to supply the subject with a steady state of the desired O2 concentration during operation. If the subject's inspiratory flow exceeds the bypass flow, outside gases will be inspired and create a steady-state error in O2 concentration. This leads to unreliable signal acquisition in the resulting images. A bypass flow was chosen to exceed the peak inspiratory flow of the subject to prevent inspiration of outside gases.
Instrumental dead space refers to a volume enclosed in hardware where expired gases may be reinspired. A high volume of dead space can lead to undesired gas mixing that contributes to signal delay and steady-state error. The internal volume of the bypass flow attachment was kept small enough to prevent reinspiration, but still allow subjects to breathe comfortably. During operation, the bypass flow partially fills the internal volume of the attachment with the supplied gas, further reducing instrumental dead space.
The multiple breath inert gas washout is used to analyze pulmonary function. During the nitrogen multiple breath washout, the subject breathes air initially, then the inspired gas switches to 100% oxygen (0% nitrogen).1 Flow and the expired gas concentrations of N2 and O2 are recorded for a certain number of breaths, typically until the concentration of N2 falls below 2%. The expired N2 concentration decreases, while the expired O2 concentration rises as the residual N2 in the lungs is diluted in each successive breath.
To evaluate the functionality of the design, multiple breath washouts were performed according to the procedure described in Singer et al.4 using the improved gas delivery system to observe the step response of the inspired gas switching. A successful washout of N2 indicates stability of the supplied O2 concentration, and rise time was observed visually after switching the inspired gas between breaths.
Subjects performed the washout while seated with the flow meter positioned between the facemask and bypass flow attachment. A sampling line for a medical gas analyzer (Perkin-Elmer, Waltham, MA) was placed at the mouth of the facemask. The washout was completed when the expiratory N2 concentration was at 2% or lower for multiple breaths, at which time the inspired gas switched back to air. The gas concentration signals were recorded using data acquisition tools in LabVIEW (National Instruments, Austin, TX) at a sampling rate of 500Hz. The data were then processed in Matlab (Mathworks, Natick, MA).
Functionally, the SVI procedure is comparable with performing a series of multiple breath washouts done in succession where N2 is alternately washed out and washed in, depending on the supplied gas. After achieving the desired performance, the improved gas delivery system was integrated into the SVI protocol and tested. Images were acquired for each breath and analyzed in MATLAB (Mathworks) using the methods described in Sá et al.2 The resulting map of SV values was then compared with an SV map for the same subject using the original plumbing system. Statistical analyses were performed using Prism (GraphPad, San Diego, CA).
A bypass flow of 2.0L/s was used to exceed subjects' peak inspiratory flow during the SVI procedure. At this rate, the supplied gas will physically clear the inspiratory line of the previous gas in 100ms after switching, a significant improvement over the original plumbing system.
The bypass flow attachment has an instrumental dead space of 41mL in addition to the standard facemask (73–113mL depending on size). At this low volume, the effects of reinspiration of gases were negligible.
Figure 4 displays the flow and gas concentrations recorded during the nitrogen multiple breath washout. The rise time, the time taken for the inspired O2 concentration to reach at least 92%, was 150ms (Fig. 4A). The figure also shows stability in the supplied gas concentration as the inspired O2 remains at ~100% during operation. The curve of expired O2 and N2 concentrations follows the expected quasiexponential behavior for multiple breath washouts as the O2 concentration rises, while N2 is gradually diluted out of the lung (Fig. 4B).
Figure 5 displays SV maps for the same subject obtained using the original plumbing system (Fig. 5A) and the improved gas delivery system (Fig. 5B). Figure 5C presents a histogram of the SV distribution of the original map and two separate maps using the improved system obtained 7 days apart. The average values for the SVI distribution, assuming a log (Gaussian), were as follows: original map, amplitude 0.10, center 0.24, and width 0.48; the average±standard deviation values of the two improved gas delivery maps were amplitude 0.12±0.01, center 0.23±0.01, and width 0.41±0.06. In a previous SVI validation study, the 95% confidence interval difference between repeated SVI measurements of heterogeneity (width) using the original plumbing system was found to be −0.127 and 0.141 (repeated measures in 13 subjects, confidence interval estimated using Student's t-distribution, df=12).5 When comparing the original plumbing system with the new gas delivery system, the individual differences in width were −0.113 and 0.034, falling within this range.
The design was shown to deliver and rapidly switch inspired gases with fast rise time and reliable supplied O2 concentration stability. It was successfully used in multiple breath washout and SV procedures. The average preparation time for an experiment was shortened from 25 to 5min using the improved gas delivery system.
In the original plumbing system, due to the limited space within the MR scanner bore, the remotely operated valve was placed outside the bore away from the subject's mask and T-shaped three-way valve and connected through a large-bore tube. This introduced a delay in the abrupt switch of inspired gas concentrations as the volume of the inspiratory line had to be cleared of the previous supplied gas through subsequent breaths. This delay had to be corrected for by measuring the tidal volume, or volume of expired gas, of each of the subject's breaths to calculate when the supplied gases had completely switched. To measure tidal volume, large-bore tubing attached the outlet of the T-shaped valve to a flow meter placed in the console room, which also made the plumbing system a closed circuit and could possibly occlude breathing in the event of operational error. The signal delay had to be calculated for each individual subject and added complexity, time, and effort to the computation of SV.
The improved method eliminated this delay by placing the manually operated switching valve in the console room and connecting the small-bore inspiratory line directly to the bypass flow attachment on the subject's facemask, while leaving the outlet open to the environment. The constant bypass flow clears the inspiratory line of the previous gas such that the subject experiences a true step change in O2 concentrations with a combined plumbing delay and rise time of 250ms. This rapid switching reduced the inspired gas signal delay to negligible values and eliminated the need for tidal volume measurement and exhaust tubing, which allowed for an open circuit operation that ensures that the subject's breathing is in no danger of being occluded due to hardware failure or human error. These improvements not only reduced the hardware components and computational complexity for SV but also reduced the potential for experimental error.
The center and amplitude of the SV distributions obtained with the novel and the original plumbing (shown in Fig. 5) were of negligible average difference. The widths of distributions for the original and improved systems were shown to have individual difference values within the 95% confidence interval for repeated measures obtained using the original plumbing system.5 Therefore, we can conclude that the results obtained using the new gas delivery system are not significantly different from results obtained using the previous method. The improved system was shown to be similarly reliable in obtaining data for SV distribution maps.
The compatibility of PLA components to the MRI environment, as well as the advantages of 3D printing components over traditional fabrication methods, was previously discussed in Herrmann et al.3 By comparison, to have the bypass flow attachment machined out of polyetheretherketone (PEEK), a medical grade plastic with high dimensional stability and chemical resistance, would cost an estimated $890 and take up to 2 weeks to deliver. With a locally available 3D printer, the design can be delivered overnight at a material cost of $2 based on a price of $45/kg of PLA filament.
The supplied flow used during operation is dependent on the subject's inspiratory flow. For example, while 2.0L/s is used for adult subjects, pediatric subjects require less flow to supply 100% inspired gas. Pediatric subjects would also benefit from reduced dead space. For this case, the design may be printed in smaller, patient-tailored versions to accommodate the size or age of the subject. This adds to the versatility of the design when fabricated through 3D printing.
The inspired gas delivery system can be implemented in any research study involving rapid switching of inspired gasses while in the MRI. It is MR safe, being constructed from PLA and nonferrous metals, and is manually operated without pneumatic remote control valves or other electronic equipment. The new gas delivery system improves safety by allowing normal breathing with and without supplied gas flow, and also adds the ability to perform SVI and multiple breath washout procedures simultaneously in the MR scanner.
This report presented a replacement method of delivering inspired gas to research subjects in the MRI environment. The technology of 3D printing introduces many advantages for designing custom MR-compatible hardware cheaply and quickly, allowing the improved gas delivery system to overcome safety and space restrictions that hampered the previous method. The commercial availability of 3D printers makes this method readily available to be cheaply fabricated for any laboratory or clinic attempting to perform SVI or other MRI-based physiologic studies. The design can also be easily modified, using computer-aided design software, to fit different equipment, making the technique of 3D printing an advantageous alternative to fabrication by machining or purchase of commercial hardware.
This work benefited from the technical contributions of Janelle Fine for MBW hardware. This work was supported by the National Institutes of Health through National Heart, Lung, and Blood Institute Grants R01-HL104118, R01-HL119263, and F30-HL-110755.
No competing financial interests exist.