Background

Biomedical microrobots (here termed simply “robots”) will potentially revolutionize medicine. Many research groups have studied the biomedical applications of robots, including targeted drug delivery, biopsy-taking, hyperthermia control, radioactive therapy, scaffolding applications, in vivo ablation, stenting, sensing, and marking [1,2,3,4,5,6,7]. Most robots are controlled and operated in low Reynolds number fluids [8,9,10]. Therefore, the robots must be appropriately designed and fabricated; size, geometry, and material properties must be considered if they are to operate within the body [8, 11, 12]. More recently, tethered robots have been developed for diverse biomedical applications.

Especially, robots may play roles in minimally invasive vascular surgery. One treatment for chronic total occlusion (CTO), i.e., complete blockage of a coronary artery, is percutaneous coronary intervention (PCI) [13,14,15,16,17] employing guidewires and catheters. Some guidewires penetrate blood clots while others guide catheters. The guidewires are manually controlled by pushing them back-and-forth or rotating them about their axes. However, this is time-consuming and highly dependent on operator skill. Thus, to reduce operative time and increase success rates, many groups are seeking robotic solutions.

Several groups have performed distal shaping of catheters and guidewires to improve steerability and controllability [18,19,20,21,22,23,24,25,26]. Krings et al. and Lalande et al. developed magnetic microguidewires fitted with permanent magnets at the tips; an external magnetic field was used for steering [20, 21, 24]. However, a high-intensity magnetic field was required because the microguidewires were not very flexible and the steering system was composed of permanent magnets, which limits the real-time steering for a catheter.

Clogenson et al. developed a steerable guidewire compatible with magnetic resonance imaging (MRI) [27]. Settecase et al. designed a steering catheter for use with MRI systems. A mathematical model was derived by exploiting the equilibrium between the magnetic and mechanical restorative torque; the equation was expressed in a linear form to simplify modeling. Thus, the model has certain limitations when used to estimate large nonlinear deformations [25]. Thus, we here develop a novel magnetic robot attached to a guidewire that exhibited high-level steerability through multi-angled branches of a blood vessel phantom under the influence of a low-level magnetic field, improving real-time delivery performance. In this study, an electromagnetic steering system was used to improve the steering performance. Moreover, the proposed guidewire-based microrobot with flexible material is easier to control using external magnetic field compared with the previously reported magnetic catheter in Ref. [21].

Our robot was fabricated from polydimethylsiloxane (PDMS), which has a low elastic modulus and a high Poisson ratio. Because it is highly deformable, the robot can be readily steered through various vessel branches using a low-level magnetic field. A permanent magnet placed at the end of the PDMS beam is used for steering. To verify robot deformation and effective steering, we used a mathematical model based on the Euler–Bernoulli beam and a finite element model (FEM) to predict the deformation angles of the microrobot tip. We compared the results of the mathematical model, the FEM simulation, and the experimental values. To demonstrate its practical utility, we performed steering and tracking using a complex two-dimensional phantom. To demonstrate the potential of the robot, we performed biocompatibility testing using human colorectal cancer (HCT116) cells.

Methods

Robot design

The tip of the robot is attached to a guidewire, and the robot is steered or deformed by magnetic torque generated by a uniform external field as shown in Fig. 1. The deformation angle is controlled by either magnetic field magnitude or direction. Therefore, within the phantom, the guidewire is steered in the desired direction by controlling robot deformation; we modeled this process using the FEM simulation of COMSOL software (COMSOL Multiphysics 5.25a; COMSOL Inc., USA).

Fig. 1
figure 1

Schematic of the magnetically actuated flexible microrobot used to improve guidewire steerability

Full size image

The mathematical model and the FEM

Robot deformation angles were calculated using the Euler–Bernoulli beam model; the equilibrium between the magnetic and mechanical restorative torque is given by:

$${\text{T}}_{\text{magnetic}} = {\text{T}}_{\text{mechanical}}$$
(1)
$$\theta = \frac{mBL\sin (\gamma - \theta )}{EI}$$
(2)

where m is the extent of magnetization of the permanent magnet, B is the magnitude of the external magnetic field, γ is the field direction, θ is the deformation angle of the robot tip, L is the robot length, E is the Young’s modulus of PDMS, and I is the second moment of inertia of the beam as shown in Fig. 2. Although this linear model can be used to create a deformation curve in the presence of an external magnetic field, the large nonlinear deformations of the robot compromise mathematical accuracy. The assumption that the microrobot length (L) is constant reduces the accuracy of the mathematical model. At the higher strain angles, therefore, the difference between the mathematical model and the experimental results diverges. However, the COMSOL model takes into account the structural deformation, leading to a better match between the experimental and simulated results.

Fig. 2
figure 2

Steering and actuation of a flexible microrobot controlled by an external magnetic field

Full size image

Consequently, we used an FEM to model deformation under an external magnetic field, in COMSOL software. Figure 3a shows a schematic of the two-dimensional model. In the FEM simulation, only the robot was considered. The model features a permanent magnet, PDMS structure, surrounding region of interest (air), and infinite element domain. Tables 1 and 2 list the material properties and robot geometry, respectively. Figure 3b–e illustrate the FEM (the elements are in blue and the red arrow is the field direction). This direction γ (Fig. 2) varied from 0 to 170º. The torque created deformed the robot as shown in Fig. 3b–e. Maximum deformation was observed at a field direction of 80°.

Fig. 3
figure 3

Simulation of flexible microrobot deformation with a magnetic field of intensity 15 mT. a Microrobot geometry. b Magnetic field directions of 50°, c 90°, d 130°, and e 170°

Full size image
Table 1 Material properties of flexible microrobots used in mathematical modeling and establishment of the FEM
Full size table
Table 2 Geometric properties of flexible microrobots with guidewires PDMS, polydimethylsiloxane
Full size table

Robot fabrication

We considered several factors when designing and fabricating the robot. First, the robot diameter must be less than that of the coronary arteries. Second, the robot must be actuated by a low-intensity magnetic field. Third, the robot must be connected to a conventional guidewire. Lastly, the robot must be biocompatible. Therefore, the robot was built of flexible biocompatible PDMS with a low Young’s modulus; the robot bore a permanent magnet affording high magnetization and a microspring was used to connect the robot to a conventional, commercially available guidewire. The geometrical parameters are listed in Table 2. When casting flexible robots, a mold is needed. A PDMS mold was fabricated using the replica-molding method and a metal master. This transparent mold (Fig. 4a) was used for fabrication. Thereafter, the robot must be detached from the mold. As the mold and the robot are made of the same material, they tend to stick together. We coated the mold with trichloro-(1H,1H,2H,2H-perfluorooctyl) silane (PFOTS) (Sigma-Aldrich, USA) to decrease the surface energy and render robot detachment easier. A PFOTS thin film was deposited on the PDMS mold in a vacuum chamber for 2 h [28]. As shown in Fig. 4a, the robot components were first aligned on the mold. The distance between the microspring and the permanent magnet was 3 mm. A brass pipe was placed in front of the magnet to keep it steady during molding. As shown in Fig. 4b, the mold was filled with Sylgard 184 silicone elastomer mixture (PDMS; Dow Corning Corp., USA) at a PDMS:curing agent weight ratio of 10:1; trypan blue was used to visualize the mixture within the mold. Oven-curing for 24 h followed and the robot was detached from the mold (Fig. 4c). The brass pipe was removed and the final structure connected to a conventional guidewire (Fig. 4d).

Fig. 4
figure 4

Fabrication of the guidewire-based flexible microrobot

Full size image

Experimental setup

We used a uniform external magnetic field to steer the robot (Fig. 5); the magnetic steering system (MiniMag; Aeon Scientific GmbH, Switzerland) generated a uniform magnetic field with five degrees of freedom (DOF; three positional and two vectorial) in the region of interest [29,30,31,32,33]. MiniMag provides a maximum magnetic field strength of 20 mT, which can be controlled by the MiniMag’s operating software [30]. Robot deformation was measured using a VZM 600i Zoom Imaging Lens (Edmund Optics Inc., USA) linked to a charge-coupled device (CCD) camera (Grasshopper; Point Gray Research, Inc., Canada).

Fig. 5
figure 5

Experimental setup used to test steering and tracking of the flexible microrobot using a magnetic coil system (MiniMag)

Full size image

Results and discussion

Robot steering

As described above, we mathematically modeled robot deformation angles and established an FEM using COMSOL software. Here, we compare the mathematical data, and the FEM and experimental results, when the robot was deformed by an external magnetic field. We used the parameters listed in Tables 1, 2 and 3 both for the experiments and for modeling. Table 3 lists the experimental conditions; the direction of the magnetic field varied from 0 to 170° and the field intensity was either 5, 10, or 15 mT. Figure 6 shows the mathematical data, and the FEM and experimental results. The robot deformation angle increased as the magnetic field increased. The mathematical model did not correctly predict the deformation angles; the model lacks the nonlinear terms required to model large deformations. However, the FEM correctly predicted the experimental results (Figs. 6, 7). The deformation angles were controlled from 0 to 80° in a magnetic field of 15 mT. These angles increased as magnetic field intensity increased. For example, when the magnetic field direction was 90°, the deformation angles were 20, 36, and 44° for magnetic field intensities of 5, 10, and 15 mT, respectively. In addition, the deviation between the simulated and the experimental results was about 6.23°, which is acceptable.

Table 3 Experimental and simulated conditions used to evaluate steering of the flexible microrobot
Full size table
Fig. 6
figure 6

Comparison between the results of the mathematical model, the finite element model (FEM), and experiments when the deformation angle of the flexible microrobot was affected by the magnetic field direction (which varied from 0 to 170°) at magnetic field intensities of a and d 5 mT, b and e 10 mT, c and f 15 mT. d, e and f show errors between simulated and experimental results

Full size image
Fig. 7
figure 7

Comparison between the experimental results and those of the FEM in terms of the deformation angles of the flexible microrobot actuated by a magnetic field of intensity 5 mT in the 90° direction. In the simulation panel, the x- and y-axis units are millimeters and the right colored bar shows the magnetic flux density

Full size image

Tracking experiments

We explored tracking in a multi-branched complex filled with deionized water (Fig. 8). This was a two-dimensional phantom fabricated by a three-dimensional printer (ProJet MJP 3500; 3D Systems, USA) with branching at angles of 15, 30, 45, 60, 75, and 90° (Fig. 8). The phantom was placed in the region of influence (ROI) of MiniMag to magnetically actuate the robot. The guidewire-based robot was then steered to the desired positions. The magnetic field was used for rotational steering and the guidewire for manual pushing (rectilinear motion). Initially, the robot was positioned in the top branch of the phantom (Fig. 9a–d), was then steered 45° counterclockwise (Fig. 9a), guided manually into the branch, magnetically steered 45° clockwise, and finally pushed manually to the desired position. The same process was used to guide the robot to a branch in the middle of the phantom (Fig. 9e–h). The steering was dexterous (Additional file 1). Figure 9 confirms that the robot can be steered within the complex environments of the phantom. In all experimental studies using the guidewire with microrobot, the guidewire was successfully tracked and reached the desired position.

Fig. 8
figure 8

Tracking of the guidewire-based flexible microrobot in a two-dimensional blood vessel phantom with branches at various angles

Full size image
Fig. 9
figure 9

Tracking of the flexible microrobot in a two-dimensional phantom with various branches. a, c magnetic steering at 45°; b, d manual rectilinear motion imparted by pushing the guidewire; e, g magnetic steering at 80°; and f, g manual rectilinear motion imparted by pushing the guidewire. Additional file 1 shows the tracking

Full size image

Biocompatibility

To explore robot biocompatibility, we evaluated the effect of the robot on the viability of human colorectal cancer (HCT116) cells seeded into 24-well plates, with or without robots, and cultured for 3 days at 37 °C under 5% (v/v) CO2 in a humid (95% relative humidity) incubator. The cell culture processes were carried out as follows [34]:

  1. 1.

    The microrobot without guidewire was sterilized twice with 70% ethanol and phosphate buffered saline (PBS, Welgene, Gyeongsan, South Korea) for sterilization and then dried for 20 min.

  2. 2.

    The microrobot was placed in each well of a 24-well plate except for the control groups.

  3. 3.

    One mL cell culture medium with HCT 116 cells (5 × 105 cells/mL) was dropped into each well plate.

  4. 4.

    Well plates were incubated for 3 days in an incubator (Fig. 10c) and then the cell viability test was performed according to the manufacturer’s instructions.

    Fig. 10
    figure 10

    The biocompatibility test results. Bright-field and fluorescent images of HCT116 cells in wells a without flexible microrobots (control group) and b with flexible microrobots. c HCT116 cells around flexible microrobots. d Cell viabilities

    Full size image

Cell viability was assessed using a LIVE/DEAD Cell Imaging Kit (excitation wavelength 488 nm, emission wavelength 570 nm; Molecular Probes, Life Technologies Corp., USA) to stain live (green fluorescent signal) and dead (red fluorescent signal) cells. Cells cultured without robots (control cells) are shown in Fig. 10a, and cells cultured with microrobots in Fig. 10b, c. Figure 10d shows the cell viabilities, which were similar between the two groups. Thus, the robot is non-toxic.

Conclusion

We designed and fabricated, via replica-molding, a guidewire-based robot for use during PCI. A low-level magnetic field affords actuation; the robot is highly flexible, smaller than a coronary artery, and connected to a conventional guidewire. The robot is steered and controlled using a low-intensity, external magnetic fields (5–15 mT). Robot deformation is controlled using geometric parameters (length and diameter), magnetic steering (extent and direction), and the material properties. We used Euler–Bernoulli theory and an FEM to predict robot deformation. The experimental results were in good agreement with the simulated values, but not the mathematical data. The robot moved within a two-dimensional blood phantom featuring many branches. The robot was guided to desired locations and steered into branches at angles of 45 and 80º. Furthermore, the robot was not toxic to human cells, and can thus be used in vivo.