The Introduction of an MR-Conditional Prototype for Cardiopulmonary Bypass Support: Technical Aspects and System Requirements

The influence of hemodynamic parameters on the cerebral tissue management during complex cardiac procedures using cardiopulmonary bypass is still under evaluation. Direct assessments of changes in hemodynamic parameters have not been performed thus far due to the lack of available and appropriate measurement systems.

Magnetic resonance imaging (MRI) provides information on the localization and extension of ischemic areas and vascular status as well as the identification and potential recovery of cerebral tissues at risk [1, 2]. These assessments can be made by applying different MR sequences, including perfusion-weighted imaging, diffusion-weighted imaging, MR angiography, and/or T2-weighted gradient-echo [3‒7]. According to the American Society for Testing and Materials (ASTM) Standard F2503-13, the requirements for MR-conditional cardiopulmonary support are safety, proper functioning, and avoidance of artifacts [8]. Previous experimental attempts to construct MR-conditional CPB devices included the introduction of extra-long tubing systems or pump heads used over a long carbon shaft [9‒11].

The purpose of this study was to develop a prototype MR-conditional cardiopulmonary support device that provides high-quality images and is technically feasible for use in pre-clinical studies. For this reason, the technical requirements were investigated first, followed by an assessment of the new device features. The operability of this device was compared with the already-available Stöckert S3 CPB system (Sorin Group, Milano, Italy).

The following is a detailed explanation of the approach used to develop an MR-conditional pump. As a first step, we analyzed the different components of the standard device to define the system requirements. In the second step, we described the features of the device and its implementation. The final step included tests designed to verify its accurate performance.

The first and most important design requirement for this extracorporeal circulatory support system was MR conditionality (ASTM Standard F2503-13) [8]. Additional mandatory and desirable requirements may need to be fulfilled in order to conduct a pre-clinical study (Table 1). The Stöckert S3 roller pump, which is used in existing pediatric heart-lung machines, served as a reference.

A pneumatic circuit was used to account for MR conditionality [8]. All the components of the MR-conditional pump system can be placed inside the MR scanner with the pressurized gas outlets present in the procedure room (Fig. 1a, b).

The prototype device consists of three main components: a pump head (I), blood circuitry (II), and pneumatic elements (III). A control unit (IV) is used to set parameters and for monitoring.

(I) Pump Head. The pump unit consists of a modified roller pump head (SPQ 225; Möller Medical GmbH, Fulda, Germany), a customized air motor (PTM mechatronics, GmbH, and Bibus AG, Fehraltorf, Switzerland), and an encoder (ME 22 LD; PWB encoders GmbH, Eisenach, Germany). Tube squeezing and the use of correct occlusion settings are critical to avoid backflow with increased kinetic energy (i.e., nonocclusive or under-occlusive settings), reductions in the lifespan of the tubing due to spallation (i.e., over-occlusive settings), or hemolysis.

The design of the prototype is based on the aforementioned Stöckert S3 SPQ 225 pump that includes two rollers. All ferromagnetic components were removed and replaced. The roller and rotor wing were replaced by three-dimensional (3D) printed parts (Fig. 1c, d). The roller position was adjusted using a LeeP plastic spring (Lee Springs, Brooklyn, NY, USA). The force needed for full occlusion of the pump tube in static condition (FN) was approximately 65 N. Because the springs exert force on the tubing at an angle of 26°, the absolute force exerted by the springs (FS) must be higher, i.e., FS = FN/cos (α) = 72.3 N. We used the strongest suitable LeeP plastic spring (Art. Nr. LL 050 075 U40 G) and added four springs in sequence, which generated a maximum force of 69.8 N when compressed to solid length. The force on the tube increases when the rotor is in motion and hydrostatic pressure builds up.

The optimal setting of the roller pump is “just-occlusive” which permits no retrograde flow and minimizes spallation [12]. To set the roller pump to the just-occlusive level, the outlet tube is fully clamped, and a pressure of 300 mm Hg is generated by moving the rotor. The pressure decreases by 1–2 mm Hg/s following optimal occlusion settings. Occlusion is achieved after 10 min of constant pumping following the replacement of the pump tube. The occlusion setting procedure for the prototype device is the same as that used for the Stöckert S3.

(II) Blood Circuit. Blood pumped through an oxygenator (D101; Sorin Group, Milan, Italy) connected to a heat exchanger is returned to the patient. The ultrasonic flow probe CO.55 (SONOTEC Ultraschallsensorik Halle GmbH, Halle, Germany) is used to measure the precise rate of blood flow. Pressure is assessed using MX960 pressure transducers (Smiths Medical Inc., St. Paul, MN, USA) and displayed on the CentriMag^® control unit (Thoratec Switzerland GmbH, Zurich, Switzerland [part of Abbott Laboratories]). The entire system is MR-conditional and is placed as close to the patient as possible to reduce the priming volume required by the circuit.

(III) Pneumatic Circuit. An air tube is connected to the gas terminal outlet to provide pressurized air. A piezo-controlled three-way proportional pressure-regulating valve (Hoerbiger Holding AG, Zug, Switzerland) is used to modulate the airflow driving a modified pneumatic motor (PMO 0450). The air that escapes from this system is directed away from the patient and diffused through a silencer. While the motor and silencer can be placed next to the MR scanner, the pressure-regulating valve is placed outside the 20 mT (200 Gauss) line (Fig. 1a, b).

(IV) Control Unit. An Aspire R 11 notebook computer (Acer Inc., New Taipei, Taiwan) was introduced as the user interface. The screen displays the set and actual flow rates of the pump. This interface also permits us to set parameters and control the flow sensors, as well as to transfer and store all data. The hardware control device (myRIO-1900) uses the LabVIEW visual programming language (National Instruments, Austin, TX, USA). The signal from the encoder regulates the air supply that drives the motor by adjusting the position of the proportional valve. The setpoint for the pump speed is regulated with an external rotary knob. The LabVIEW code is designed to control the pump speed and calculate the flow rate (online suppl. Fig. S1, S2; for all online suppl. material, see https://doi.org/10.1159/000531179).

The calculated cardiac output can be used to adjust the flow rate. The cardiac output is calculated by multiplying the patient’s body weight in kg with the perfusion rate in mL/kg/min which will generate the required flow rate for full-body perfusion in each individual. Depending on the temperature, the flow rate can be adjusted based on the patient’s cardiac output.

The controller and display are placed inside a shielded enclosure that blocks the emission of radio frequency (RF) electromagnetic radiation into the MR environment. The aforementioned external rotary knob is placed in a separate shielded box.

A mock circuit was created to evaluate the performance of the peristaltic pump system before its use in a pre-clinical study (Table 2). The performance of the prototype device was determined in several separate tests. The tests evaluated runtime durability (I) and achievable speed and pulsation behavior (II). The tests were designed to compare the characteristics of the prototype pump with the expected values and system requirements. Finally, MR conditionality (III) was assessed.

Water was sufficient for the durability test. A simple clamp was used to generate a counterpressure of approximately 100 mm Hg at all flow rates. The duration of this test was set arbitrarily at 22 h (based on personnel resources and organizational factors). The pump ran at a speed of 158 rpm while under observation for the first 5 h of the testing period. Because the shaft drive heated up to 38°C during this period, the speed was reduced to 79 rpm, and the test was continued for an additional 11.5 h without surveillance.

The air inlet pressure was limited to 7 bars. An infrared camera (PI 450; Optris GmbH, Berlin, Germany) was used to measure heat development. The use of water as the test liquid resulted in a constant offset in the low measurement of greater than approximately 5% as the sensor was calibrated to a water-glycerol mixture.

These tests were performed at room temperature (22°C) using pressurized air at 7 bar. The test fluid was a mixture of water and glycerol that replicated the viscosity (μ) of human blood at 37°C (3–4 mPa·s). The measured dynamic viscosity of the test fluid was μ = 3.01 mPa·s and was therefore a suitable substitute for human blood. An infant oxygenator (D101; Sorin Group) was added to the pump tube outlet. This device is suitable for flow rates between 1 and 2.5 L/min and is typically used in circuits with 5/16” (7.94 mm) pump tube diameters. The oxygenator has blood tube connectors with 1/4” (6.35 mm) diameters. When used in a CPB device, the resistance (i.e., afterload) generated by the patient, the cannulae, and the oxygenator generate a counterpressure that depends directly on the flow rate and viscosity of the fluid [13].

The counterpressure that would be generated by the patient and cannulae was simulated by maintaining a constant pressure of 100 mm Hg at the tube outlet [14]. Pressures were measured before and after the oxygenator to determine the drop in pressure resulting from this component and the pressure at the pump outlet (i.e., pO). The pump inlet pressure (pI) is at the atmospheric level, and the pressure difference Δp = pO−pI is known as the pressure gradient or pressure head [14, 15].

The prototype was tested using rotational pump speeds from 10 to 175 rpm (compared to 32–224 rpm using the Stöckert device). To measure the influence of the afterload on the rotational speed of the pump, the oxygenator was removed and the tube outlet was connected directly to the mock circulation test bench (0–225, in steps of 25 mm Hg) [14].

Items labeled as MR-conditional have been demonstrated as stable and thus are allowed in specified MR environments under specified conditions of use (ASTM International, F2503-05). Most MR rooms have the 20 mT line demarcated by a visible line drawn on the ground. MR-conditional electronic devices, including anesthesia monitors and medical ventilators, must remain outside this line.

According to the MRI Safety Standard (ASTM F2503-13), MR-conditional devices cannot create artifacts during the imaging process. To evaluate its impact on image quality, the prototype was assembled in the MR room that housed a 3T Discovery MR750 scanner (General Electric, Fairfield, CT, USA) with software version DV25. All MRI software packages include a system performance test (SPT) that generates 20 pictures in several different dimensions to determine whether any devices operating in the MR room alter the imaging procedure. Any disturbance will lead to a failed SPT. During the MR-conditionality test, the heat generated on the surface of components was measured with the PI 450 infrared camera while the device pumped the aforementioned water/glycerol solution at a constant speed of 150 rpm through a closed circuit.

Calculations provided by the manufacturer of the pump head indicated that a constant volume of 12.3 mL would be displaced per revolution using an 8-mm tube. The results shown in Figure 2 compare the actual flow which was measured at 12.7 mL per revolution with this calculated value.

The curve shown is nearly linear. Thus, the measured flow (Q) can be approximated by the equation Q = 12.7 mL/revolution *n, where n is the rotational pump speed in rpm.

The pump unit passed the mechanical durability test as it was capable of 22 h of continuous pumping without sustaining any damage. Over the first 3 h, the flow rate dropped from 2,136 mL/min to 2,075 mL/min. This, and despite a difference of 61 mL/min, 97.2% of the original flow rate was maintained after 3 h.

The motor shaft (made of brass) heated up to nearly 38°C. The rotor rollers also heated up to approximately 26.5°C (Fig. 3). Heat generation increased at higher speeds. After running the pump for 2 h at 135 rpm, the hottest spot on the surface of the drive shaft reached nearly 62°C, as shown in Figure 3c. At a pump speed of 150 rpm, the maximum surface temperature was nearly 66°C after 4 h, as shown in Figure 3d.

The piezoelectric valve that controls the air supply to the motor can either be set to a constant value or it can be set to respond to a proportional-integral (PI)-controlled loop feedback mechanism that regulates the air inlet based on the desired rotational pump speed (Fig. 4). The voltage setting that results in a flow rate of approximately 550 mL/min is, on average, slightly higher than that used when operating with PI control. Both flow curves exhibit a low followed by a high peak. While the maximum peak is at a higher value under a constant setting, the minimum peak under PI control is lower (Fig. 4,). The boxplots shown in Figure 4 illustrate these characteristics based on 1,165 values collected over a 22-s sample period. The area within the 25th to 75th percentile is very small when the system is under PI control. This indicates that the flow rate is close to the median value 50% of the time.

In the upper section of Figure 5, the pulsations produced by the prototype device are compared to those from the Stöckert S3 reference pump at flow rates of approximately 650 mL/min (prototype at 638 mL/min and 50 rpm vs. Stöckert S3 at 685 mL/min and 65 rpm). The Stöckert pump produces a small maximum peak and a wide minimum peak. Between the peaks, the flow rate is nearly equal. The prototype device also exhibits an even flow rate between the two extremes, although the maximum peak is higher and the minimum peak is lower than their corresponding counterparts generated by the Stöckert pump. While the two pumps exhibited similar pulsation behavior at flow rates of approximately 1,900 mL/min, the prototype generates higher and lower extrema (prototype at 1,958 mL/min and 150 rpm vs. Stöckert S3, 1,832 mL/min and 192 rpm).

Of note, the difference between the highest and lowest values resulting from this trial using the Stöckert S3 pump is lower than the differences observed for the prototype at all three flow rates. However, compared to the prototype, the areas between the 25th and the 75th percentile determined for the Stöckert S3 were comparatively larger at flow rates of approximately 650 mL/min, equal at flow rates of 1,300 mL/min, and smaller at flow rates of 1,900 mL/min. The boxplots in online supplementary Figure S3 illustrate these characteristics.

The first SPT of the MR scanner was performed with no additional devices inside the MR environment. This served as a baseline and revealed no image disturbances, as shown in Figure 6a.

For the second SPT trial, the RF-shielded enclosure, pump unit, and control encoder were placed inside the MR room at their desired positions (see Fig. 1b). All devices inside the RF-shielded enclosure were supplied with power. The pump unit encoder and the control encoder box were connected to the control unit through a D-sub connector. As shown in Figure 6b, the image that was generated was congruent with the baseline test. Thus, the prototype achieved a passing grade on the second SPT.

For the third SPT trial, the flow sensor was placed in the MR room in addition to the equipment that was already installed as described above. The flow sensor was connected with an unshielded cable. As expected, this resulted in considerable image disturbance (Fig. 6c). Once the unshielded cable was replaced with one that was shielded, image quality was restored, and the device passed the final SPT as shown in Figure 6d. While the image included slightly lighter streaks, this did not exceed the limit of acceptable noise as defined by the test protocol.

The present study aimed to develop an MR-conditional cardiopulmonary support system. Although the system met the initial testing criteria and exhibited encouraging outcomes, some aspects remain to be addressed.

Results from our study highlighted the feasibility of designing a blood pump that could operate in conjunction with an MRI scanner without introducing significant artifacts to image processing. The pump system is practical and suitable for use as a CPB system. Therefore, it may be applicable for use in subsequent pre-clinical studies. This prototype was developed using a peristaltic pump head rotor made completely of non-ferromagnetic materials accompanied by a customized air motor. Our approach differs from previous attempts that featured extension tubes [9, 11] or polycarbonate drive shafts [10].

Given that there is no obvious substitute for an electric motor, we needed to identify a completely alternative technique. The choice of the motor and the characteristics of the drive were crucial considerations for the design of the prototype device. Piezoelectric motors are the only types of electric motors suitable for use in an MR environment. Non-electric motors, for example, fluidic actuators (i.e., pneumatic or hydraulic motors), may also be feasible alternatives. To meet standard requirements, the drive must be able to generate a torque of 2.5 Nm at 250 rpm. Piezoelectric drives (e.g., Shinsei Kogyo Corp., Tokyo, Japan; PiezoMotor AB, Uppsala, Sweden; and Noliac A/S, Kvistgaard, Denmark) are generally 63 to 4,500 times too weak to drive the pump head in this device and thus are not suitable for use in the prototype pump system.

A hydraulic system would be another valid approach. This type of system requires a liquid medium to transport the force produced to the actuator. Water may be a safe and effective medium for this application. Although water-based solutions for hydraulic motors are currently in use in MR environments, they cannot be adapted for use in peristaltic pump drives [16, 17].

Pneumatic motors have high power-to-weight ratios compared to electric motors. They are robust, do not overheat, and are explosion-proof. Furthermore, they are easy to control and can reverse direction rapidly [18]. A significant advantage is the availability of tapping points for compressed air and other medical gases in the MR room. According to the ISO 7396-1 standard for medical gas piping systems, the operating pressure of the gas connections should be at 4 bar. Based on these considerations, we decided to include a pneumatic drive in the design of the prototype CPB device.

A modified roller pump was used to build the prototype device. Our tests demonstrated that the ideal flow rate was 12.7 mL per revolution using an 8-mm tube. The calibration line in Figure 2 shows the linear dependence of the measured flow rate on pump speed. This relationship confirms that it will be possible to maintain precise control of the pump speed.

The dynamic viscosity measured for the water/glycerol mixtures (μ = 3.01 mPa·s) was at the lower end of the possible viscosities of human blood. A dynamic viscosity of at least 4 mPa·s should be used in follow-up tests because the viscosity of blood increases at lower temperatures. For example, at 22°C, the viscosity of blood is around 26% higher than it is at 37°C [19].

The results of the durability testing revealed that the flow rate at constant pump speed decreased only 2.2% over time (to 97.8%). This decrease may be due to spallation and deformation of the pipe and will be verified in follow-up experiments. Similarly, the existing flow measurements should be integrated into the control algorithm so that the temporal drift can be corrected.

As in any motor, an air motor heats up during operation due to air compression and friction. The heat of the drive shaft is transferred to the coupling device which is made of polyoxymethylene. The highest temperature measured in this study was around 57°C. This did not present a problem for the system because the melting temperature of polyoxymethylene is much higher (166°C).

Our measurements revealed strong pulsations from both the reference pump and the prototype. The pulsation behavior of the prototype improved during operation when the airflow of the motor was changed from constant to PI controlled. Compared to the reference pump, the pulsation of the prototype was more pronounced with higher maximum and lower minimum values at equal flow rates.

In a previous study, Mulholland et al. [20] assessed blood flow and potential damage to blood components caused by the roller pumps during CPB procedures. They described the pulsations of the pump but could not correlate them with damage sustained by blood components. Blood flow-related damage resulted from high shear stress that occurs when two rollers occluded the tube simultaneously. This is less likely to occur with our prototype device because its pump head has a larger diameter compared to the reference product, thus facilitating a lower rotational pump speed.

The peristaltic pump system successfully passed the SPTs and thus can be considered MR-conditional. The flow sensor did generate some RF noise and thus needs to be shielded within a small enclosure. Of note, the pump temperature during the testing of MR conditionality was low; this indicates that the eddy currents induced by the magnetic field did not generate heat at the rotating motor shaft.

Currently, the properties of the prototype device are under investigation. The results show that this device can be used safely and effectively for approximately 6 h in a magnetic field at constant power. Thus, we plan to use this device and hope to gain important insights into cerebral blood flow during simulated cardiac surgery in a series of animal model studies in which different cannulation and perfusion strategies will be explored. One important goal would be to determine when and where brain injuries occur. An important advantage of MR imaging is the possibility of using imaging during the surgical procedure. If ischemic areas emerge, perfusion can be adjusted accordingly. The effectiveness of other neuroprotective measures could also be examined.

Our planned animal studies will lead to further optimization of the prototype device. Among the questions to be addressed, we will determine whether the degree of hemolysis is within acceptable limits and whether the device can be used reliably on a continuous basis. In the medium term, specific improvements (e.g., usability and mobility of the prototype) are planned. Human clinical use may be conceivable in the long term. A device such as this would be of particular benefit to pediatric patients in need of circulatory support. One decisive advantage of this technology would be the use of MRI rather than computed tomography (and thus significant doses of radiation) for one or more of these indications.

The development of an MR-conditional pump system was successful: the constructed system caused no relevant image artefacts performant in an MR environment. Feature testing of the pump in the laboratory and MR room showed that requirements like operability, controllability, and flow rate ranges are fulfilled. Therefore, this prototype is suitable to perform an open-heart surgery in an animal model to assess brain perfusion.

We would like to thank Bibus AG for the custom-made production of an air motor.

Ethical approval is not required for this study in accordance with local or national guidelines; no research on humans, identifiable human material, or identifiable data was planned as part of the study. Furthermore, no research on animals has been conducted.

The authors report no conflicts of interest.

This work was supported by funding from the Swiss Heart Foundation (17029).

Michael Hofmann: writing of this manuscript and discussion of the experiments’ findings; Martin Schmiady: writing of this manuscript; Dominik Schulte: discussion of the experiments’ findings and revision of the manuscript; Samuel Sollberger: development of the MR-conditional pump, execution of experiments, data collection, analysis of the findings, and reporting; Thierry Carrel: contribution to the writing of this manuscript; Peter Hasenclever: study preparation, results’ discussion, and adjustments of extracorporeal circulation protocols; Beat Werner: evaluation and monitoring of the MR conditional tests; Mirko Meboldt: project supervision; Michael Hübler: conception of idea and supervision; Marianne Schmid Daners: conceptional study and hardware development, supervision, discussion of results and analysis of the experiments, revision of the manuscript.

All data generated or analyzed during this study are included in this article and its online supplementary material. Further inquiries can be directed to the corresponding author on reasonable request.

Michael Hofmann and Martin Schmiady shared first authorship.