Design and performance evaluation of collision protection-based safetyoperation for a haptic robot-assisted catheter operating system

Linshuai Zhang1,2& Shuxiang Guo1,3

& Huadong Yu2& Yu Song1

& Takashi Tamiya4 & Hideyuki Hirata1 &

Hidenori Ishihara1

# Springer Science+Business Media, LLC, part of Springer Nature 2018

AbstractThe robot-assisted catheter system can increase operating distance thus preventing the exposure radiation of the surgeon to X-ray forendovascular catheterization. However, few designs have considered the collision protection between the catheter tip and the vesselwall. This paper presents a novel catheter operating system based on tissue protection to prevent vessel puncture caused by collision.The integrated haptic interface not only allows the operator to feel the real force feedback, but also combines with the newly proposedcollision protectionmechanism (CPM) tomitigate the collision trauma. TheCPMcan release the catheter quicklywhen themeasuredforce exceeds a certain threshold, so as to avoid the vessel puncture. A significant advantage is that the proposed mechanism canadjust the protection threshold in real time by the current according to the actual characteristics of the blood vessel. To verify theeffectiveness of the tissue protection by the system, the evaluation experiments in vitro were carried out. The results show that thefurther collision damage can be effectively prevented by the CPM, which implies the realization of relative safe catheterization. Thisresearch provides some insights into the functional improvements of safe and reliable robot-assisted catheter systems.

Keywords Robot-assisted catheter system . Tissue protection . Endovascular catheterization . Safety operation . Vascularinterventional surgery (VIS)

1 Introduction

A report from the AmericanHeart Association (AHA) showedthat: cardiovascular and cerebrovascular diseases have be-come one of the three major causes (heart disease, strokeand vascular diseases) of death in human beings, which is aserious threat to human health (Lloyd-Jones et al. 2010). Evenin the developed countries, cardiovascular disease remains themajor cause of mortality, accounting for 34% of deaths eachyear (Lloyd-Jones et al. 2010). Along with the rapid develop-ment of the medical technologies, vascular interventional sur-gery (VIS) as a revolutionary surgical technique and an effec-tive treatment method is applied to treat these vascular dis-eases (Guo et al. 2015b). In the traditional VIS, a rigid hose,called a flexible catheter, is manipulated to access a lesiontarget followed the human vascular vessels from the smallincisions in the neck, arm or groin area. Then the catheter issteered under the guidance of a digital reduction shadow an-giography (DSA) system to be positioned properly on thetarget location (Khoshnam and Patel 2013). VIS has a greatapplication in the world, because it has many advantages, suchas smaller incisions, quicker recovery and fewer

* Shuxiang Guoguo@eng.kagawa–u.ac.jp

* Huadong Yuyuhuadong@cust.edu.cn

Linshuai Zhangs16d502@stu.kagawa-u.ac.jp

1 Faculty of Engineering, Kagawa University, 2217-20 Hayashicho,Takamatsu, Kagawa, Japan

2 School of Mechatronical Engineering, Changchun University ofScience and Technology, Changchun, Jilin, China

3 Key Laboratory of Convergence Medical Engineering System andHealthcare Technology, the Ministry of Industry InformationTechnology, School of Life Science, Beijing Institute Technology,No. 5, Zhongguancun South Street, Haidian District, Beijing 100081,China

4 Department of Neurological Surgery, Faculty of Medicine, KagawaUniversity, Takamatsu, Kagawa, Japan

Biomedical Microdevices (2018) 20:22 https://doi.org/10.1007/s10544-018-0266-8

complications (Guo et al. 2016a). Nonetheless, some potentialchallenges also have been introduced in. The surgeon’s fatigueand physiological tremors will affect the success of the sur-gery, and long radiation exposure has the risk to the surgeon’shealth (Mohapatra et al. 2013).What’s more, the surgeonmustbe highly skilled and specialized due to the high risksinvolved.

To solve these disadvantages mentioned above, the researchof surgical robot-assisted systems has become a hot study topicin recent years. The robot-assisted technology has played animportant role in multiple medical fields. Okamura’s researchteam has done a lot of work in the robot-assisted needle inser-tion (Okamura et al. 2004; Bettini et al. 2004; Webster et al.2006). These studies play an important role in promoting thedevelopment and high precision control of medical robots.Deshpande et al. (2015) presented a new motorized microma-nipulator based on a spherical orienting device for the robot-assisted laser phonomicrosurgery. It can provide greater accu-racy and effectiveness, thus enhancing the safety of the opera-tion. Also, the electromagnetic (EM) guiding and tracking tech-nology is widely proposed in the context of computer-assistedendovascular procedure (Condino et al. 2016; Piazza et al.2017). Especially in the study of the catheter operation, somerobot-assisted catheter systems have been developed by thecommercial companies. The Amigo ™ (Catheter Robotics,Inc., Mount Olive, NJ) remote catheter system (RCS) can po-sition the catheter into the right side of the heart safely andeffectively (M. Khan et al. 2013). The Sensei X robotic cathetersystem (Hansen Medical, Inc., Mountain View, CA, USA) is anew generation of flexible robot platform, which combinesadvanced 3D catheter control technology and 3D visualizationtechnology. It is a combination of collaborative technologies toprovide physicians with better accuracy and stability (Willemset al. 2010; Hlivák et al. 2011). Moreover, many researchgroups in universities are committed to the development ofrobot-assisted catheter systems. Yogesh et al. (2009) proposeda novel remote catheter navigation system, which can reducethe physical stress and irradiation to operators under the X-ray2D fluoroscopy for image guidance. And it had the ability tosense and replicate motion within 1 mm and 1° in the axial andradial directions, respectively. A new compact tele-roboticcatheter navigation system with three degrees of freedom wasdeveloped, which allowed the interventionalist to remotelysteer a conventional catheter from a safety place (Tavallaeiet al. 2016). To reflect the force information of the blood vesselmore intuitively during the operation, some researchers embedthe force measuring device to robot-assisted catheter systems.Fu et al. (2011) constructed a master-slave catheterization sys-tem to position a steerable catheter under the 3D guiding image,and its insertion mechanism with force feedback assistedsurgeons to operate the catheter. J. Payne et al. (2013) designeda novel master-slave force feedback system for theendovascular catheterization, and it reduced the magnitude

and duration of contact forces exerted on the vessel wallsaccording to the force feedback during a simulatedendovascular procedure. Guo et al. (2012) proposed a novelmaster-slave robotic catheter operating system to train the un-skilled surgeons to perform the vascular interventional surgerywith force feedback and visual feedback. A load cell and torquesensor were used to detect the force signals in the axial direc-tion and the torque signals in the radial direction, respectively.The evaluation of dynamic and static performance and the syn-chronization for the robot-assisted catheter system were con-ducted by (Ma et al. 2013). To get the force information directlywithout any mechanical transmission, a compact, cost-effectiveforce-sensing device based on strain gauges was placed at thefront end of the slave manipulator, thus increasing the accuracyof force measurement (Zhang et al. 2017a). Currently, the vir-tual reality (VR)-based technology as a potential method wasequipped with robot-assisted catheter systems (Zhou et al.2015; Wang et al. 2017). The vasculature (Johnson et al.2011; Zhang et al. 2010) and catheter (Lenoir et al. 2006;Tang et al. 2012) models can be built by the VR simulatorand the visual images of surgical region can be displayed inthe meantime. For the sake of improving the skill of cathetermanipulation for novices, Wang et al. (2016) developed a train-ing system integrated cooperation of VR simulator and hapticdevice to train the surgeon, and improved the coordination ofsurgeon’s eye-hand. However, the collision trauma during thecatheterization remains unresolved, despite of the visual assis-tance. Surgeons often rely on visualization to avoid major vas-cular collision in the catheterization study. It is difficult to de-termine whether a vascular collision occurs, due to the absenceof depth perception and low definition of the fluoroscopic im-ages (J. Payne et al. 2013). Thus, the surgeon has to observe thedeflection of the catheter to determine whether the collisionoccurred. With such consideration, a novel master manipulator,damper-based magnetorheological (MR) fluid, was developedto realize the true force feedback (Guo et al. 2015a, b). Basedupon the same principle, Yin et al. (2014, 2015) designed ahuman operator-centered haptic interface to detect the cathetertip collision with the blood vessel during the surgery practice.The application of magnetorheological (MR) fluid in the robot-assisted catheter system allows the surgeon to feel as if thecatheter is inserted directly into the patient’s blood vessel, whilerealizing the safety operation consciousness. On this basis,Wang et al. (2017) introduced a speed adjustable mechanism(SAM) for tissue protection, which was installed before thehaptic interface device. Nevertheless, up to now, few designshave taken into account the collision protection function of acatheter manipulator in the catheterization. The fragile vessels(especially cerebral vessels) have the risk to be penetrated bythe catheter tip when the collision occurs, because the noviceshave little experience to adjust the catheter manipulation rapid-ly. Motivated by such consideration, it is extremely urgent fortele-operated robotic catheter systems to provide patients with

22 Page 2 of 13 Biomed Microdevices (2018) 20:22

collision protection during endovascular catheterization proce-dures because excessive forces could rupture the blood vesselwalls and result in bleeding.

In this paper, the novel developed robot-assisted cathetersystem devotes to alleviate the collision trauma to blood ves-sels, so as to improve the safety operation of catheterization.In this system, the clamping structure based on electromag-netic brake is introduced to realize the clamping and relaxationof the catheter. It has big contact area and reliable clampingwhich can ensure the operation accuracy of the catheter. Sincethe clamping force can be controlled by adjusting the inputcurrent (Zhang et al. 2016), the designed clamping structurewill automatically release the catheter when the measuredforce is greater than a setting threshold value, thus effectivelyavoiding vessel puncture caused by system malfunction orhuman error. To sum up, the novel developed robot-assistedcatheter system contributes to the tissue protection in case ofcollidingwith the vessel wall. To verify the performance of theCPM, the insertion tests in vitro were carried out.

2 Structure of the whole system

2.1 Overview of the robot-assisted catheter system

The conceptual diagram of the robot-assisted catheter system,shown in Fig. 1, describes the flow chart of the operationalprocess (Zhang et al. 2017b). The robot-assisted cathetersystem comprises five parts: master manipulator, slavemanipulator, local control subsystem on the master side andthe slave side, and communication subsystem. When theoperation in the master manipulator is given by the surgeon,the motion signals will be obtained by the control subsystemon master side and transmitted to the control subsystem onslave side via the TCP/TP communication protocol or localcontrol (Guo et al. 2016b; Zhang et al. 2017a). The slavemanipulator clamping the catheter will do the same motionin the blood vessel of the patient, according to the controlsignals from the master side. Our developed robot-assistedcatheter system is devoted to reducing collision trauma.Since the friction coefficient between the vessel and thecatheter wall are small as well as the deformation of theblood vessel (Takashima et al. 2007), the collision traumamainly occurs at the vascular bending area of the forwardmotion between the catheter tip and the vascular wall.Therefore, the collision protection mechanism is proposed inthe slave side of the system. Once the collision occurs betweenthe catheter tip and the vascular wall, the catheter will bereleased automatically by the clamping structure. The cathetercan not continue to insert with the slave manipulator, thusavoiding the vessel puncture. In the meantime, the operatorneeds to retract the catheter and adjust the orientation of thecatheter tip for the next insertion.

2.2 Design of the master manipulator

In recent years, some haptic devices based on intelligent ma-terials have been introduced because of its development. Themagnetorheological (MR) fluid with high permeability andlow hysteresis as the most typical representative is applied.Rizzo et al. proposed the Haptic Black Box I and II (HBB Iand HBB II) with the concept of freehand to acquire tactilesensation (Sgambelluri et al. 2006; Rizzo et al. 2007). Tsujitaet al. (2013) designed a novel encountered-type haptic inter-face with MR fluids to increase a sense of reality in surgicalsimulators. Blake and Gurocak (2009) developed a hapticglove with MR brakes for virtual reality.

Thus, in combination with the magnetic field characteristicsof the magnetorheological fluid and the requirement of therobot-assisted catheter system, our lab proposed amaster hapticdevice based onMR fluid shown in Fig. 2, which can achieve arealistic sensation (Yin et al. 2014). In the magnetic field, themagnetorheological particles become the chain structures dueto its characteristics. For this reason, the operator will feel sub-tle resistance forces caused by the viscosity of the MR fluid.When a catheter is inserted into theMR fluid (applied magneticfield), the shearing force between the MR fluid and the catheterwill be generated. And the shearing force can be adjusted by themagnetic field intensity. The intensity of magnetic field can becontrolled by the current which is the display form of the feed-back force from the slave side. In other words, the shearingforce varies with the feedback force from the slave side. Andthen the shearing force will be transmitted to the operator in theform of a haptic force. It is like operating a catheter inside theblood vessel of a patient in the vascular surgery.

The prototype of the haptic master manipulator is shown inFig. 3. This haptic interface system consists of two parts: readpart and haptic display part. The read part is used for measuringthe motion of the operator and the haptic display part is used fordisplaying the force measured in the slave side. The detailedstructure design was introduced in (Yin et al. 2015). Althoughthe haptic sensation can be recreated by the current according tothe measured force from the slave side, this is only a kind offeeling. In order to ensure the authenticity of the haptic sensa-tion, it is necessary to calibrate the kinesthetic sensation. Aforcemeasurement system called ‘haptic calibration subsystem’was proposed. The detail was introduced in (Yin et al. 2014).

2.3 Design of the slave manipulator

The slave manipulator is used to operate the catheter instead ofthe surgeon’s hand and it requires a stable clamping and forcemeasurement to realize the motion of the catheter during thecatheterization. In addition to these basic functions, the colli-sion protection is very necessary for the robot-assisted cathetersystem. With such consideration, we introduced a novelclamping mechanism based on electromagnetic braking to

Biomed Microdevices (2018) 20:22 Page 3 of 13 22

clamp the catheter for the forward motion, backward motionand rotation.

As shown in Fig. 4, when the coil is not energized, thecollet will be pressed into the taper hole of the clamping ringby the compression spring, so that the collet will be tightenedup to clamp the catheter. Then the catheter can do the forwardmotion or backward motion steadily with the slave manipula-tor. When the coil is energized, the electromagnetic force gen-erated by the coil will absorb the iron corn, so that the colletcanmovewith the iron corn to release the catheter. In addition,the adopted clamping ring is removable, which can be con-trolled by two screws to adjust the range of the initial clampingforce, and then the electromagnetic force is used to fine-tuningin the range (Zhang et al. 2016).

The structure diagram of the slave manipulator is shown inFig. 5. The load cell is fixed on the support plate. A force ringfixed on the load cell is linked to the force plate installed onthe clamping structure. The clamping structure supported bytwo omnidirectional bearings can pull or push the load cell todetect the counter force. The force plate can rotate freely whenthe catheter is moving in the axial direction. The torque signalscan be measured by the torque sensor which is installed in therotation driving mechanism. In order to prevent the radial

motion from interference with the line of power supply, thestructure of electric brush is set on the right side of the elec-tromagnetic chuck to supply the power.

2.4 Principle of the collision protection mechanism(CPM)

The force analysis of the catheter is shown in Fig. 6. The forcewill be balanced when the catheter is clamped. In addition, thecontact method of point-surface is adopted to reduce the in-fluence of the friction force on the clamping effect. Accordingto the mechanical equilibrium, we get:

Fn⋅tan θ ¼ Fs−Fe ð1ÞFth ¼ f ⋅Fn ð2Þ

That is,

Fth ¼ − f ⋅Fe⋅cot θþ f ⋅Fs⋅cot θ ð3Þwhere, Fth is the maximum force that can be held when thecatheter is clamped. Fe is the electromagnetic force generated

Fig. 1 The overview of the master-slave robot-assisted catheter system

Fig. 3 Prototype of the haptic master manipulatorFig. 2 Schematic of the haptic master device

22 Page 4 of 13 Biomed Microdevices (2018) 20:22

by the electromagnetic chuck. Fs is the elastic force generatedby the compression spring. f is the coefficient of static frictionbetween the catheter and the collet. And θ is the oblique angleof the clamping ring’s inside.

The electromagnetic force can be defined by (Zhanget al. 2016),

Fe ¼ 0:31I2N 2 d2 þ 2:5δ2

δ2� 10−6 ð4Þ

where, I is the current through the electromagnetic chuck. N isthe number of turns for the electromagnetic coil. d is the di-ameter of the iron core. And δ is the removable distance of theiron core. When the device has been manufactured, N, d and δare all fixed. To simplify the equation, we set a constant k,

k ¼ 0:31N 2 d2 þ 2:5δ2

δ2� 10−6 ð5Þ

The threshold value of collision protection is obtained asindicated in Eq. (6),

Fth ¼ − f ⋅k⋅cot θ⋅I2 þ f ⋅Fs⋅cot θ ð6Þ

From the Eq. (6) we can see that the threshold value of col-lision protection can be controlled by the current and the springforce. Once the contact force between the catheter and the vesselis bigger than the protection threshold, the collet will move to theside of electromagnetic chuck to release the catheter, so as toavoid the vessel puncture. In this process, the small increase ofthe spring force caused by spring compression has no effect onthe loosening of the catheter, because the electromagnetic force isalso increasing. Thus, the Fs in Eq. (6) is seen as a fixed value,and the current is the one and only factor to adjust the collisionprotection threshold for protecting different collision areas.

3 Performance of the whole integratedsystem

The bilateral teleoperation is defined as an operator using arobotic system to complete some tasks at a distance, whilingreceiving haptic feedback and visual feedback from the target

environment (Niemeyer et al. 2008). These two kinds of feed-back information can improve the performance of the robot-assisted system (Gwilliam et al. 2009). In this paper, the sta-bility issues as important themes in bilateral teleoperation willbe reported in this section.

3.1 Evaluation of the bilateral control performance

The experimental setup on master side is shown in Fig. 7. Thedriver of the stepper motor (ASM46AA, Oriental Motor CO.LTD) is connected with the conversion terminal (CCB-SMC2,CONTEC), thus being easily regulated by the motion controlboard (SMC-4DF-PCI, CONTEC). The motion control boardwas embedded in PC. The axial motion information of thecatheter is acquired by the reading part and this performancewas reported by (Yin et al. 2015). Two rotary encoders (MTL,

Fig. 6 Schematic diagram of the force analysis for the catheter

Fig. 5 Structure diagram of the slave manipulator

Fig. 4 The schematic diagram of clamping mechanism

Biomed Microdevices (2018) 20:22 Page 5 of 13 22

MES020-2000P, Japan) are adopted to transmit the motioninformation. One is for axial motion information and the otheris for radial motion information. They are transmitted to thesalve side by a controller.

Figure 8 describes the detection device on the slave side.The laser displacement sensor (KEYENCE, LK-500, Japan)and the hollow rotary encoder (MUTOH, UN-2000, Japan)were adopted to measure the displacement and the rotationangle of the catheter on the slave side, respectively. The hol-low rotary encoder was placed next to the torque sensor. Thedisplacement and rotation angle were collected into the PC.

Figures 9 and 10 display the axial motion tracking anderror, respectively. From the results, the maximum value oftracking error in axial motion is less than 2 mm which canmeet the requirement of the surgery. In traditional VIS, even ifa well skilled surgeon operates a catheter in the axial move-ment, it produces a motion error greater than 2 mm.

The performance of radial motion tracking and error areshown in Figs. 11 and 12, respectively. The results show thatthe radial tracking error is less than 4 °, and the fluctuation ofamplitude is also larger. However, the rotation of the catheter isjust to adjust the direction of the catheter tip, which has a smallrisk of damaging the blood vessel during the catheterization.

3.2 Evaluation of the haptic system performance

For investigating the performance, the haptic system is used totrack a vascular model shown in Fig. 13.The experimentalsetup is shown in Fig. 14. A 5Fr catheter was operated bythe slave manipulator to do the insertion motion. The forcesat the slave side generated when the catheter comes in contact

with the vascular model are transmitted by a force transducerto the data acquisition unit. Concurrently the signals togetherwith the model of MR haptic interface in the master sidecombine to deduce the current in the coils that will generatesimilar forces to the operator as measured in the slave side.The detailed modeling and control system of the MR hapticinterface were reported by (Yin et al. 2014; Yin et al. 2015).

The result of force tracking is displayed in Fig. 15. The rootmean squared error for the generated force in the master side is0.0347 N. When the catheter tip passes through a curved re-gion, the force generated in the master side and the force mea-sured at the slave side have an error less than 0.1 N which isacceptable in the actual surgery (Okumura et al. 2008). Perhapsthe reason for this error is due to the dynamics of activationcoupled with the regulator response can cause a slight delay ofabout 10 ms between the catheter tip contacting the curvedregion of the vascular model wall and the master haptic systemresponding to the contact (Ahmadkhanlou et al. 2009).However, even with the technical difficulty, the usability ofthe haptic system increased by having force feedback available.

4 Performance of the collision protectionmechanism

4.1 Relationship between the collision protectionthreshold and the current

In order to adjust the collision threshold value quickly andaccurately during the catheterization, the relationship betweenthe collision protection threshold and the current should be

Fig. 8 The detection setup ofmotion information for theslave side

Fig. 7 The input setup of motioninformation for the master side

22 Page 6 of 13 Biomed Microdevices (2018) 20:22

determined. The experimental setup is shown in Fig. 16. A rodwith the silica gel film on the surface was clamped by the slavemanipulator, and the other side was fixed with the sliding table.The current controlling the electromagnetic force was suppliedby the power source. The force information was collected bythe data collector and displayed on the PC screen. In the cali-bration experiment, when the current was smaller than 0.44A,the iron core could not be attracted because of the smallerelectromagnetic force. When a greater force was exerted onthe rod, it would slide relative to the collet, resulting in failureof the collision protection function. When the current was big-ger than 0.65A, the rod would be released since the greaterelectromagnetic force attracted the iron core directly.Therefore, the test current was controlled from 0.44A to0.65A with 0.01A increase. After setting the current value,the force exerted on the rod was increased with the micro feedof the sliding table. The rod was released when the exertedforce reached the threshold value. The peak of the force curvedisplayed on the PC screen was the collision protection thresh-old for the current. And this test was performed ten times for

each current. The average threshold for each current and thefitting curve were shown in Fig. 17. From the results, the fittingequation between the threshold value and current is established,

Fth ¼ −7:02949I2 þ 2:95312 0:44A≤ I ≤0:65Að Þ ð7ÞIt is formally the same as Eq. (6). And it can be transformed

into the control algorithm applied in the slave controller.According to the actual condition of the blood vessel, thecollision protection threshold can be adjusted in real time bythe input current to realize the tissue protection.

4.2 Evaluation of the collision protection mechanism(CPM) in vitro

4.2.1 Setting of the collision protection threshold

The blood vessels of a human are resilient. When the tip of thecatheter collides with the vessel wall and the vessel wall is notpenetrated, the maximum elastic force generated by the elasticdeformation of the vessel wall is the safety operation interval of

Fig. 9 Results for the axial motion tracking

Fig. 12 Errors of the radial motion tracking

Fig. 11 Results for the radial motion tracking

Fig. 10 Errors of the axial motion tracking

Biomed Microdevices (2018) 20:22 Page 7 of 13 22

the catheter. Collision protection is safe and effective onlywhen the triggering force occurs within this interval. In otherwords, the setting of safety operation threshold is very neces-sary for the collision protection. The vascular model, shown inFig. 13, was used to simulate the blood vessel of a human.During the vascular interventional surgery, the most vulnerablepositions of the puncture are the regions with big curvature. Asshown in Fig. 13, region A, B and C are the vulnerablepositions for vessel puncture when the catheter tip collideswith them. Che Zakaria et al. (2013) pointed out that the contactforce of the catheter and blood vessel was more than 0.12 N,the vascular wall had the risk to be penetrated. In addition, asystematic analysis of contact forces between catheter tip andreal tissue was made. When the contact force of the catheter tipreached 0.12 N, no perforation occurs in tissue (Okumura et al.2008). Therefore, during the actual operation of the catheter,the safety operation threshold is governed as follows:

Fsa;th ¼ 0:12þ Fse;th ð8Þwhere, the ‘0.12’ is the alarm value for the safety operation.Fse,th is the setting protection threshold for collision protection.

It includes the viscous drag force of the blood, the friction forcebetween the catheter surface and the vascular wall, and thefriction force between the catheter and the catheter sheath.This suggests that Fse,th is the variable factor, which can beset according to the actual situation and requirement.

In this experiment, the ‘alarm value’ may be less than‘0.12’, because the adopted vascular model is not a real vas-cular tissue. Therefore, a 5F catheter was used to insert into thevascular model for detecting the force information to confirmthe setting protection threshold and safety operation threshold.The experimental setup is shown in Fig. 14.When the catheterpassed through the region A, B and C without tip collision for20 times, the maximum forces for the three regions were0.38 N, 0.61 N and 0.75 N, respectively. Therefore, the settingprotection thresholds for region A, B and C were 0.38 N,0.61 N and 0.75 N, respectively. When the catheter passedthrough the region A, B and C with tip collision for 20 times,the minimum forces for the three regions were 0.55 N, 0.76 N

Fig. 14 The experimental setupfor evaluating the haptic system

Fig. 15 Force transparency results of the haptic feedback system

Fig. 13 A vascular model for in vitro experiments.

22 Page 8 of 13 Biomed Microdevices (2018) 20:22

and 0.89 N, respectively. Based on the above measurements,the minimum ‘alarm value’ is 0.14 N, which is biggerthan 0.12 N. Thus, the ‘0.12 N’ was adopted as the ‘alarmvalue’ in this paper. In this case, the safety operationthresholds for region A, B and C are 0.5 N, 0.73 N and 0.87N,respectively.

4.2.2 Evaluation of the collision protection mechanism (CPM)for the single threshold

In order to evaluate the performance of the CPM described onthe previous sections, the experiments in vitro were per-formed. The experimental setup is shown in Fig. 14. In addi-tion, because the collision force can be greatly influenced bythe moving speed of the catheter (Wang et al. 2017), we set themoving speed of the catheter is 10 mm/s, which can meet theclinical demands of VIS (Wang et al. 2010). After setting theprotection threshold, we operate a catheter in the master hapticinterface to control the slave manipulator, which will operatethe required catheter to pass through the region A, B and Cwith the tip collision, respectively. Concurrently the force in-formation will be recorded by the PC screen. Figure 18 showsthe examples of the experimental results for the three thresh-olds (0.38 N, 0.61 N, 0.75 N). From the results we can see thatthe catheter will be released quickly for each thresholdwhen the measured force is bigger than the protectionthreshold. Even in the moment of collision protection,the vascular tissue is not damaged, because the trigger-ing force is within the safety threshold of the vasculartissue’s endurance. That is, the plots indicate that the collisionprotection mechanism can take effect within the safety opera-tion threshold.

To verify the stability of the CPM, the experiments for thethree regions were conducted 10 times, respectively. The re-sults are shown in Fig. 19. From the results we can see thateach triggering force occurs within the respective safety oper-ation interval. This proves the stability and effectiveness of theCPM for the single threshold. However, there is an error be-tween the triggering force and the setting threshold. The av-erage error and variance have been summarized inFig. 20. This bar chart displays the stability of the av-erage error for each setting threshold. That is, the reasonfor this error is caused by the system inertia or the regulationaccuracy of the current, rather than by human actions.Moreover, such a small error is acceptable for most vasculartissues except for some special cases. In these special casessuch as needle insertion or palpation, the ‘alarm force’ will be

Fig. 16 The experimental setupfor getting the relationshipbetween the threshold and thecurrent

Fig. 17 The relationship between the threshold and the current

Biomed Microdevices (2018) 20:22 Page 9 of 13 22

changed according to the different characteristics of tissues(Yin et al. 2015).

4.3 Real-time adjustment of the protection threshold

In vascular interventional surgery, the catheter often needs topass through several bending regions consecutively to reachthe lesion target. To prevent any bending regions from beingperforated, the collision protection threshold needs to be ad-justed in real time.

4.3.1 Validation trials

A random insertion experiment was carried out to test the real-time adjustment capability of the CPM. The experimental setupis shown in Fig. 10. At the beginning of the experiment, theinitial collision protection threshold was set to 0.38 N.Figures 21 and 22 display the experimental process and results

(a) The setting protection threshold is 0.38N

(b) The setting protection threshold is 0.61N

(c) The setting protection threshold is 0.75N

Fig. 18 Examples of experimental results for collision protectionmechanism Fig. 20 The average error between the triggering force and the setting

threshold

Fig. 19 The triggering forces of ten times experiments for collisionprotection

22 Page 10 of 13 Biomed Microdevices (2018) 20:22

for real-time adjustment. From the two figures we can see that,at the 37 s of the experiment, the tip of the catheter collided withthe region A (as shown in Fig. 13), and the collision protectionfunction was triggered. Then we retracted the catheter and ro-tated the catheter tip to do the second insertion. At the 48 s, thecatheter passed through the region A without tip collision. Atthe same time, the collision protection threshold was adjusted to0.61 N. At the 56 s, the catheter reached region B (as shown inFig. 13) and the second tip collision occurred. We adjusted thecatheter as the previous steps for the next insertion. At the 78 s,the catheter passed through the region B without tip collision.Also, the collision protection threshold was changed to 0.75 Nin the meantime. At the 83 s, the catheter passed through theregion C (as shown in Fig. 13) without collision protection,because the catheter tip did not collide with the region C andthe measured force did not reach the setting threshold.

The experimental results indicate that the CPM takes effectin the two catheter tips collision, and the triggering forces areboth within the respective safety operation thresholds. TheCPM can not be triggered when the tip collision does not occur.

For verifying the effectiveness and stability of the collisionprotection performance by real-time adjustment of the protec-tion threshold, three different subjects (non-medical) wereasked to perform the catheter insertion with real-time adjust-ment of the protection threshold. Each of them performed fivetrials respectively, and the angle of catheter insertion changedat each random trial. Table 1 records the triggering forces ofcollision protection in region A, B and C. The results showthat the triggering forces in each trial are all within the safetythreshold of each region except S1-T4-B, S2-T3-B and S3-T3-B. The reason may be due to the larger rigid curvature ofregion B and the manual adjustment error of the control cur-rent. During the actual operation, this situation may be im-proved because the blood vessels have certain toughness,and the curvature will decrease when the catheter passesthrough the curved region. In addition, the number of collisionprotection in region B is significantly higher than that in re-gion A and region C, which is also due to the large curvatureof region B. This may lead to multiple collision protectionduring the angle adjustment of the catheter tip, which takes along time. For all that, there are still 42 of the 45 collisionpoints can be effectively triggered, and the effectiveness of thereal-time adjustment is 93.3% in terms of the validation re-sults. In other words, the CPM can effectively reduce tissuedamage and prevent vascular perforation, thus increasing thesafety of the endovascular catheterization.

4.3.2 Performance evaluation

To further demonstrate the performance of the real-time ad-justment of the CPM, it was compared with the haptic feed-back and no haptic feedback. Trials were performed with 4different subjects (non-medical) who had no prior experience

with either setup. This design of trials provided a clear, unbi-ased performance comparison because of removing any learn-ing bias. The experimental conditions were considered asthree modes. Each mode contained the visual feedback byan IP camera to facilitate the direction adjustment of the cath-eter tip. Mode1: collision protection mechanism, haptic feed-back and visual feedback (CHV);Mode2: haptic feedback andvisual feedback (HV); Mode3: visual feedback (V). Each sub-ject performed trials 10 times with each mode for statistics.The experimental task was that each subject operated the cath-eter safely through each region with a perforated risk (A, Band C). The following two important metrics were taken intoconsideration: (1) the success rate of the task and (2) averageelapsed time of task accomplishment.

Based on the results summarized in Fig. 23, the mode ofCPM and haptic feedback (Mode1-CHV) has made great con-tribution to remit the collision trauma. Themaximal success rate

Fig. 21 The experimental process for real-time adjustment

Fig. 22 Experimental results for the real-time adjustment

Biomed Microdevices (2018) 20:22 Page 11 of 13 22

is 90% achieved by the subject2, which is higher than that ofmode2 and mode3. Such great improvement indicates that theCPM played an important role in tissue protection. Especially incerebrovascular surgery, the wall of the blood vessel is easy topierce because of its fragility. Therefore, the safe operation be-tween the catheter and the vessel should be considered.

Figure 24 shows average elapsed time of task accomplish-ment for all participating subjects. The elapsed time of theunaccomplished task is not included in the statistics. The sta-tistical results display that the average elapsed time of mode1is longer than that of mode2 and mode3. The repeated angleadjustment after collision protection is considered to be theprimary reason. In addition, it takes time to manually openand adjust the collision protection function. However, evenwith long time execution, it is acceptable to compare withreducing the risk of vascular perforation. Perhaps a highlyefficient and short-time robot-assisted catheter system withcollision protection will be developed in the future.

5 Discussion

Endovascular robotic technology is an effective and revolu-tionary method to reduce X-ray radiation and fatigue of a

surgeon for endovascular catheterization. It can also improvethe effectiveness of the procedure by precise positioning of thecatheter and the force information. Nevertheless, few designshave taken the collision protection of the vessel walls and thecatheter tip into account.

In the present work, a novel robot-assisted catheter systemwith CPMwas proposed. The collision protection threshold canbe adjusted by the current. Their relationship, shown in Fig. 17,was obtained by the experiment. Figure 18 (a), (b) and (c)showed the evaluation results of the collision protection mech-anism when the protection threshold was set as 0.38 N, 0.61 Nand 0.75N, respectively. Although there was an error (shown inFig. 20) between the triggering force and the protection thresh-old, it was still within the threshold of safety operation. Thatsuggests the CPM has taken effect to a certain extent for thetissue protection. And the stability of the collision protectionmechanism had been verified as show in Fig. 19. In the actualprocedure, the collision protection threshold needs to be adjust-ed in real time, because a catheter sometimes passes throughseveral curved areas and reaches the lesion area. Figure 22displayed the experimental results for the real-time adjustmentin a vascular model. The effectiveness of the real-time adjust-ment was summarized in Table 1. The further performance

Table 1 Validation of the collision protection performance by real-time adjustment of protection threshold (N)

Subject S1 Trigger region S2 Trigger region S3 Trigger Region

Trial A B C A B C A B C

T1 0.31 0.67 0.82 B,C 0.41 0.64 0.73 A,B 0.48 0.70 0.73 A,B

T2 0.23 0.66 0.74 B 0.35 0.59 0.82 C 0.36 0.66 0.72 B

T3 0.47 0.62 0.70 A,B 0.29 0.75 0.86 B,C 0.32 0.74 0.79 B,C

T4 0.28 0.76 0.86 B,C 0.28 0.68 0.70 B 0.35 0.70 0.74 B

T5 0.23 0.65 0.83 B,C 0.42 0.65 0.73 A,B 0.42 0.67 0.71 A,B

The bold indicates that the triggering force exceeds the safety threshold for collision protection

Fig. 24 The average elapsed time of task accomplishmentFig. 23 The success rate of the task accomplishment

22 Page 12 of 13 Biomed Microdevices (2018) 20:22

evaluation was performed in terms of success rate and elapsedtime. And the results were shown in Figs. 23 and 24, respec-tively. It indicates that the CPM can react quickly to play aprotective role for improving the safety operation of a surgery.

In the previous researches, some researchers tried to avoidthe tip collision by simulation to plan the insertion path of thecatheter (Lenoir et al. 2006; Tang et al. 2012). And otherscontributed to robot-assisted catheter systems. Yin et al.(2015) proposed a haptic interface based on MR fluid, whichcan dynamically amplify the collision force information as thealarm to remind the surgeon to retract or rotate the catheter.Also, the research on the visualization and haptic forceequipped robot-assisted catheter system had been developed(Wang et al. 2016). It took VR simulator and haptic interfaceto help the novice realize safety operation of a catheter. Thedirective notification module (DNM) will determine whetherthe catheter tip is in a safety operation area by the collisiondetection algorithm. And the signs (safe, warning or danger-ous) will be transmitted to the operator by the form of tactilesensation. Although the vision and touch enable the operatorto reduce collision frequency, they don’t avoid the risk ofcollision perforation. In view of this, Wang et al. (2017) pro-posed a speed adjustable mechanism (SAM) on the basis ofthe previous training system. This mechanism adopts the prin-ciple of continuously variable transmission (CVT) to adjustthe insertion speed of a catheter at the master side of thetraining system. And the evaluation results show that the col-lision frequency is greatly decreased. The goal of these previ-ous studies is to alert the operator to do the response when acollision has occurred or is imminent. However, the operatorssometimes could not respond fast enough to deal with suchsituations. According to the experimental results, the proposeddevice in this research is capable of tissue protection and pre-vents vascular perforation, even if the operator does not takeenough fast protection action.

Despite the extremely promising results, it is important tonote that the study is limited by the fact that the experiments ofan in vitrowere used to conduct the performance evaluation ofthe collision protection; the limitation is manifested in twoways. Firstly, the glass vascular model has great rigidity andcan not produce vascular deformation when the catheter tipcollides with the model. In addition, the curvature of the bendcan not be changed during the catheterization, which will leadto the increasing of the contact area between the wall of thecatheter and the wall of the model, thereby increasing thefriction. Secondly, in this research, the viscous resistance ofthe blood is neglected, and it is an important component of themeasuring force during the actual operation. And the frictionbetween the catheter and the blood vessel also increases be-cause of the lack of blood lubrication. Therefore, the directionfor future research is to perform experiments in vivo or withartificial vascular models which are more similar to the realones to evaluate the performance of the CPM.

6 Conclusions

Vascular tissue protection is one of the important issues in therobot-assisted catheter system. In addition to providing visualand tactile information during the endovascular catheteriza-tion, it is very essential to provide the collision protectionfunction to decrease the collision trauma and avoid vesselpuncture effectively. Therefore, in this paper, we developeda novel master-slave robot-assisted catheter system with thecollision protection mechanism (CPM) for the safety opera-tion of endovascular catheterization. The CPM is based onelectromagnetic braking to realize the vascular tissue protec-tion. Once the measuring force exceeds the protection thresh-old which can be adjusted by the current, the catheter will bereleased quickly. And the relationship between the thresholdvalue and the current has been obtained by the experiment.

Moreover, the performance evaluation experiments of thecollision protection for the robot-assisted catheter system havebeen carried out. Based on the results of these evaluationexperiments, the CPM equipped the robot-assisted cathetersystem has made great contribution to remit the collision trau-ma. It can effectively improve the security of operation duringthe endovascular catheterization.

Acknowledgments This research is partly supported byNationalHighTech.Research and Development Program of China (No.2015AA043202), andSPSKAKENHI Grant Number 15 K2120.

References

F. Ahmadkhanlou, G.N. Washington, S.E. Bechtel, Modeling and controlof single and two degree of freedommagnetorheological fluid-basedhaptic systems for telerobotic surgery. J. Intell. Mater. Syst. Struct.20(10), 1171–1186 (2009)

A. Bettini, P. Marayong, S. Lang, A.M. Okamura, G.D. Hager, Vision-assisted control for manipulation using virtual fixtures. IEEE Trans.Robot. 20(6), 953–966 (2004)

J. Blake, H. Gurocak, Haptic glove with MR brakes for virtual reality.IEEE/ASME Trans. Mechatron. 14(5), 606–615 (2009)

N.A. Che Zakaria, T. Komeda, C.Y. Low, K. Mahadhir, Development offoolproof catheter guide system based on mechatronic design. Prod.Eng. 7(1), 81–90 (2013)

S. Condino, R. Piazza, F. Micheletti, F. Rossi, R. Pini, R. Berchiolli, A.Alberti, V. Ferrari, M. Ferrari, Electromagnetic guided in-situ laserfenestration of endovascular stent-graft: endovascular toolssensorization strategy and preliminary laser testing. Proceedings ofthe 2016 international conference on medical imaging and virtualreality (2016), pp. 72–83

N. Deshpande, L. S. Mattos, D. G. Caldwell, New motorized microma-nipulator for robot-assisted laser phonomicrosurgery. Proceedings ofthe 2015 I.E. International Conference on Robotics and Automation(2015), pp. 4755–4760

Y. Fu, A. Gao, H. Liu, S. Guo, The master-slave catheterization systemfor positioning the steerable catheter. Int. J. Mechatron. Autom. 1(3–4), 143–152 (2011)

J. Guo, S. Guo, N. Xiao, X. Ma, S. Yoshida, T. Tamiya, M. Kawanishi, Anovel robotic catheter system with force and visual feedback for

Biomed Microdevices (2018) 20:22 Page 13 of 13 22

vascular interventional surgery. Int. J. Mechatron. Autom. 2(1), 15–24 (2012)

J. Guo, S. Guo, L. Shao, P. Wang, Q. Gao, Design and performanceevaluation of a novel robotic catheter system for vascular interven-tional surgery. Microsyst. Technol. 22(9), 2167–2176 (2015a)

J. Guo, L. Shao, S. Guo, Y. Yu, Q. Gao, A multidimensionalinformationmonitoring method for a novel robotic vascular inter-ventional system. Proceedings of the 2015 I.E. international confer-ence on information and automation (2015b), pp. 32–37

J. Guo, S. Guo, T. Tamiya, H. Hirata, H. Ishihara, A virtual reality-basedmethod of decreasing transmission time of visual feedback for ateleoperative robotic catheter operating system. Int. J. Med. Rob.Comput. Assisted Surg. 12(1), 32–45 (2016a)

J. Guo, S. Guo, Y. Yu, Design and characteristics evaluation of a novelteleoperated robotic catheterization system with force feedback forvascular interventional surgery. Biomed. Microdevices (2016b).https://doi.org/10.1007/s10544-016-0100-0

J. C. Gwilliam, M. Mahvash, B. Vagvolgyi, A. Vacharat, D. D. Yuh, A.M. Okamura, Effects of haptic and graphical force feedback onteleoperated palpation. Proceedings of the 2009 I.E. internationalconference on robotics and automation (2009), pp. 677–682

P. Hlivák, H. Mlčochová, P. Peichl, R. Čihák, D. Wichterle, J. Kautzner,Robotic navigation in catheter ablation for paroxysmal atrial fibril-lation: Midterm efficacy and predictors of Postablation arrhythmiarecurrences. J. Cardiovasc. Electrophysiol. 22(5), 534–540 (2011)

E. Johnson, Y. Zhang, K. Shimada, Estimating an equivalent wall-thickness of a cerebral aneurysm through surface parameterizationand a non-linear spring system. Int. J. Numer. Methods Biomed.Eng. 27(7), 1054–1072 (2011)

E.M.Khan,W. Frumkin,G.AndreNg, S. Neelagaru, F.M.Abi-Samra, et al.,First experience with a novel robotic remote catheter system: Amigo™mapping trial. J. Interv. Card. Electrophysiol. 37, 124–129 (2013)

M. Khoshnam, R.V. Patel, A pseudo-rigid-body 3R model for a steerableablation catheter. Proceedings of the 2013 I.E. international confer-ence on robotics and automation (2013), pp. 4427–4432

J. Lenoir, S. Cotin, C. Duriez, P. Neumann, Interactive physically-basedsimulation of catheter and guidewire. Comput. Graph. 30(3), 416–422 (2006)

D. Lloyd-Jones, R. Adams, T. Brown, et al., Executive summary: Heartdisease and stroke statistics-2010 update: A report from theAmerican Heart Association. Circulation 121(7), 948–954 (2010)

X. Ma, S. Guo, N. Xiao, S. Yoshida, T. Tamiya, Evaluating performanceof a novel developed robotic catheter manipulating system. J. Micro-Bio Robot. 8(3), 133–143 (2013)

A. Mohapatra, R. Greenberg, T. Mastracci, M. Eagleton, B. Thornsberry,Radiation exposure to operating room personnel and patients duringendovascular procedures. J. Vasc. Surg. 58(3), 702–709 (2013)

G. Niemeyer, C. Preusche, G. Hirzinger, in Telerobotics. Spring handbookof robotics, ch. 31 (Springer-Verlag, New York, 2008), pp. 741–757

A.M. Okamura, C. Simone, M.D. O’Leary, Force modeling for needleinsertion into soft tissue. IEEE Trans. Biomed. Eng. 51(10), 1707–1716 (2004)

Y. Okumura, S.B. Johnson, T.J. Bunch, B.D. Henz, C.J. O’brien, D.L.Packer, A systematical analysis of in vivo contact forces on virtualcatheter tip/tissue surface contact during cardiac mapping and inter-vention. J. Cardiovasc. Electrophysiol. 19(6), 632–640 (2008)

C. J. Payne, H. Rafii-Tari and G. Yang, A force feedback system forendovascular catheterisation. Proceedings of the 2012 IEEE/RSJinternational conference on intelligent robots and systems (2013),pp. 1298–1304

R. Piazza, S. Condino, A. Alberti, D. Giannetti, V. Ferrari, M. Gesi, M.Ferrari, Using of 3D virtual reality electromagnetic navigation forchallenging cannulation in FEVAR procedure. Proceedings of the2017 international conference on augmented reality, virtual realityand computer graphics (2017), pp. 221–229

R. Rizzo, N. Sgambelluri, E. Scilingo, M. Raugi, A. Bicchi, Electromagneticmodeling and design of haptic interface prototypes based onmagnetorheological fluids. IEEETrans.Magn. 43(9), 3586–3600 (2007)

N. Sgambelluri, R. Rizzo, E. Scilingo, M. Raugi, A. Bicchi, Free handhaptic interfaces based on magneorheological fluids. Proceedings of14th symposium on haptic interfaces for virtual environment andteleoperator systems (2006), pp. 367–371

K. Takashima, R. Shimomura, T. Kitou, H. Terada, K. Yoshinaka, K.Ikeuchi, Contact and friction between catheter and blood vessel.Tribol. Int. 40(2), 319–328 (2007)

W. Tang, T. Wan, D.A. Gould, T. How, N.W. John, A stable and real-timenonlinear elastic approach to simulating guidewire and catheter in-sertions based on Cosserat rod. IEEE Trans. Biomed. Eng. 59(8),2211–2218 (2012)

M. Tavallaei, D. Gelman, M. Lavdas, A. Skanes, D. Jones, J. Bax, M.Drangova, Design, development and evaluation of a compacttelerobotic catheter navigation system. Int. J. Med. Rob. Comput.Assisted Surg. 12(3), 442–452 (2016)

T. Tsujita, K. Sase, A. Konno, M. Nakayama, X. Chen, K. Abe, M.Uchiyama, Design and evaluation of an encountered-type hapticinterface using MR fluid for surgical simulators. Adv. Robot.27(7), 525–540 (2013)

T.Wang,D. Zhang, L.Da, Remote-controlled vascular interventional surgeryrobot. Int. J. Med. Rob. Comput. Assisted Surg. 6(2), 194–201 (2010)

Y. Wang, S. Guo, T. Tamiya, H. Hirata, H. Ishihara, X. Yin, A virtual-reality simulator and force sensation combined catheter operationtraining system and its preliminary evaluation. Int. J. Med. Rob.Comput. Assisted Surg. 13(3), e1769 (2016)

Y.Wang, S. Guo, Y. Li, T. Tamiya, Y. Song, Design and evaluation of safetyoperation VR training system for robotic catheter surgery. Med. Biol.Eng. Comput. (2017). https://doi.org/10.1007/s11517-017-1666-2

R. J. Webster III, J.S. Kim, N. J. Cowan, G. S. Chirikjian, A. M.Okamura, Nonholonomic modeling of needle steering, Int. J.Robot. Res., 25(5–6), 509–525 (2006)

S. Willems, D. Steven, H. Servatius, B.A. Hoffmann, I. Drewitz, et al.,Persistence of pulmonary vein isolation after robotic remote-navigated ablation for atrial fibrillation and its relation to clinicaloutcome. J. Cardiovasc. Electrophysiol. 21(10), 1079–1084 (2010)

X. Yin, S. Guo, H. Hirata, H. Ishihara, Design and experimental evalua-tion of teleoperated haptic robot-assisted catheter operating system.J. Intell. Mater. Syst. Struct. 27(1), 3–16 (2014)

X. Yin, S. Guo, N. Xiao, T. Tamiya, Safety operation consciousnessrealization of a MR fluids-based novel haptic Interface forTeleoperated catheter minimally invasive neuro surgery. IEEE/ASME Trans. Mechatron. 21(2), 1043–1054 (2015)

T. Yogesh, S. Jeffrey, W. David, D. Maria, Design and performance eval-uation of a remote catheter navigation system. IEEE Trans. Biomed.Eng. 56(7), 1901–1908 (2009)

D. Zhang, T. Wang, D. Liu, G. Lin, Vascular deformation for vascularinterventional surgery simulation. Int. J. Med. Rob. Comput.Assisted Surg. 6(2), 171–177 (2010)

L. Zhang, S. Guo, H. Yu, Y. Song, Design and principle analysis forelectromagnetic brake clamping mechanism of a novel slave manip-ulator. Proceedings of the 2016 I.E. international conference onmechatronics and automation (2016), pp. 490–495

L. Zhang, S. Guo, H. Yu, Y. Song, Performance evaluation of a strain-gauge force sensor for a haptic robot-assisted catheter operatingsystem. Microsyst. Technol. 23(10), 5041–5050 (2017a)

L. Zhang, S. Guo, H. Yu, S. Gu, Y. Song, M. Yu, Electromagneticbraking-based collision protection of a novel catheter manipulator.Proceedings of the 2017 I.E. international conference onmechatronics and automation (2017b), pp. 1726–1731

C. Zhou, L. Xie, X. Shen, M. Luo, Z. Wu, L. Gu, Cardiovascular-interventional-surgery virtual training platform and its preliminaryevaluation. Int. J. Med. Rob. Comput. Assisted Surg. 11(3), 375–387 (2015)

22 Page 14 of 13 Biomed Microdevices (2018) 20:22