• Non ci sono risultati.

Component based design of a drug delivery capsule robot

N/A
N/A
Protected

Academic year: 2021

Condividi "Component based design of a drug delivery capsule robot"

Copied!
9
0
0

Testo completo

(1)

ContentslistsavailableatScienceDirect

Sensors

and

Actuators

A:

Physical

jo u r n al hom e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / s n a

Component

based

design

of

a

drug

delivery

capsule

robot

Marco

Beccani

a,∗

,

Gregorio

Aiello

c

,

Nikolaos

Gkotsis

a

,

Hakan

Tunc

b

,

Addisu

Taddese

a,b

,

Ekawahyu

Susilo

a

,

Péter

Völgyesi

b

,

Ákos

Lédeczi

b

,

Elena

De

Momi

c

,

Pietro

Valdastri

a

aSTORMLab,DepartmentofMechanicalEngineering,VanderbiltUniversity,Nashville,TN37235-1592,USA

bInstituteforSoftwareIntegratedSystems,DepartmentofElectricalEngineeringandComputerScience,VanderbiltUniversity,Nashville,TN37212,USA cDepartmentofElectronic,InformationandBiomedicalEngineering,PolitecnicodiMilano,Milano20133,Italy

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received30December2015 Receivedinrevisedform4April2016 Accepted13April2016

Availableonlinexxx Keywords: Surgicalrobotics

Medicalcyber-physicalsystems Drugdelivery

MedicalCapsuleRobots Designenvironment Embeddedmicro-systems

a

b

s

t

r

a

c

t

Since the introduction of Wireless Capsule Endoscopy (WCE) researchers have started exploring the design space of Medical Capsule Robots (MCRs): embedded micro-systems that can operate autonomouslywithinthehumanbodyandcandiagnose,prevent,monitor,andcurediseases.Although theresearchintheareaofMCRsisanactivetopicandhasgrownexponentially,currentdevicesprovide onlylimitedfunctionalitiesbecausetheirdesignprocessisexpensiveandtimeconsuming.Toopenthis researchfieldtoawidercommunityand,atthesametime,createbetterdesignsthroughadvanced toolsupport,inourpreviousworkswepresentedadesignenvironmentfortherapiddevelopmentof MCRs.Inthispaper,thisenvironmentwasadoptedtodesignaDrugDeliveryCapsule(DDC)basedona coil-magnet-pistonmechanism.Theforceofthecoilactingonthemagneticpistonandthedrugrelease profileweremodeledandassessedonbench-topwithamaximumrelativeerrorbelow5%.Then,invivo trialswereperformedtovalidatetheDDCfunctionalitywithascheduleddrugreleaseprofilefora5h and24minprocedure.Theresultingdesignenvironmenttemplateisavailableopensourceforfurther developmentofdrugdeliveryapplicationsaswellastoserveasguidelineinprototypingnovelMCRs addressingotherclinicalneeds.

©2016PublishedbyElsevierB.V.

1. Introduction

Theintroductioninyear2000ofthePillCambyGivenImaging [1]showedthatembeddedmicro-systemscanbeusedtoreachthe moreremoteregionsofthehumanGastrointestinal(GI)tract.Since then,millionsofpatientshaveswallowedthisself-contained wire-lesscameratogetanon-invasivediagnosisofdiseasesinthesmall intestine.Unfortunately,thisapproachisnoteffectiveforother dis-trictsoftheGItract,suchasthecolonandthestomach,wheremore deadlyformsofcancerusuallydevelop[2].Themainreasonforthis isrelatedtothelackofadvancedfunctionalities,suchasactive loco-motion[3],advanceddiagnosisandtissuemanipulation[4],biopsy sampling[5],ordrugdelivery[6].

Inrecentyears,theresearchcommunityhasaddressedthisissue bydevelopinganumberofroboticsmartcapsules,referredinthis paperasMedicalCapsuleRobot(MCR).MCRstypicallyenterthe humanbodythroughnaturalorificesorsmallincisionsandcan per-formendoscopyandsurgerybyinteractingwiththesurrounding environmentwithsensorsand/oractuators.

∗ Correspondingauthor.

E-mailaddress:marco.beccani@vanderbilt.edu(M.Beccani).

ThedesignprocessofMCRshoweverischallengingbecauseit hastoaddressseverecross-cuttingconstraintssuchassize(togain non-invasiveaccess,MCRdiameterislimitedtoabout1cm),power consumption(limitedspaceforbatteryisavailableonboard),and failsafeoperation(MCRoperatesdeepinside thehumanbody). Therefore,MCRdesignanddevelopmentrequiressignificantskills andeffortsinembeddedsystems,miniaturizedelectronics, pack-aging,debugging,andmechanical miniaturizationofthedevice. Customcircuitboardsandmechanicalenclosureshavetobe engi-neered,whichmakesMCRdevelopmentanexpensiveandtime consumingprocessthatisonlyaccessibletoalimitednumberof groupsintheresearchcommunity[7].

Because many of the MCRs developed in the past share equivalenthardware andsoftware componentssuchasspecific sensors,dataprocessing,actuation,andwirelesscommunication, webelievethatitisindeedpossibletosystematizethedesignof MCRs.Thiswouldlowerthebarrierofentrytothisfieldtoresearch groupsthatdonothavetheresourcestodevelopMCRsfromthe groundup.

Inourpreviousworks,wepresentedourcontributionstowards the creation of an environment for the rapid design and development of MCRs [7–9]. The basic elements of the inte-grated design environment are shown in Fig. 1. At the time

http://dx.doi.org/10.1016/j.sna.2016.04.035

(2)

Fig.1.TheelementsoftheintegrateddesignenvironmentforMCR.

of writing, the design environment includes a library of hard-waremodulesbuiltupfromourexperiencewithexistingcapsule designs[10].Thehardwarecomponentsincludefundamental ele-mentsofMCRdesignsuchassensors(inertialsensors,encoders, vitalsignsensors),actuators(DCmotors,coils,illumination,servo motors,steppermotors),batteriesandvoltageregulators,wireless transceiversworkingatdifferentcarrierfrequencies,and micro-controllers(MCUs),whichserveasbuildingblocksforthesystem. Aflexiblebackboneenablesplug-n-playconnectivityofuptosix hardwaremodules.Eachofthehardwaremodulesiscoupledwith asoftwarecomponentandintegratedinanopen-sourceweb envi-ronmentavailableatpillforge.github.ioforthedeveloperto implementspecificMCRapplications.

Inthiswork,wepresenthowthedesignenvironmentcanbe adoptedto develop a Drug Delivery Capsule(DDC)based on a coil-magnet-pistonmechanismforcontrolleddeploymentofdrug doses over time. As a result of this work, we have created a designtemplatefortheDDC.ThistemplateaimstoprovideMCR researchersasolutionthatshortensthedevelopmenttimeofnew DDCsbasedonthesameorasimilaractuationprinciple.Template parameterssuchasthedrugviscosityanddeploymentschedules canbeadjustedaccordingthespecificsoftheapplication.Theforce exertedbythecoil-magnet-pistonmechanismandthedrugrelease profileovertimehavebeenmodeledandcharacterized experimen-tally.Beyondthespecificapplicationofcontrolleddrugdelivery, thisworkalsoillustrateshowtheopen-sourcedesignenvironment wehavedevelopedcanbeusedtorapidlydesignanMCR.

Thispaper isorganizedasfollows: Section2 brieflypresents thestateof theartfor drugdeliveryMCRs, Section3illustrates themechanical,electrical,andembeddedfirmwareandsoftware design considerationsfor thedeveloped MCR.This sectionalso describesthedesign environmentand thedesigntemplate that havebeencreatedfortheMCR.Section4presentstheexperimental assessmentonbenchandinvivoresults.Sections5and6include respectivelyadiscussionandconclusions.

2. Clinicalmotivationfordrugdeliverysystems

Theformulationofdrugsthatallowforreproducibleabsorption profilesfortargetingspecificareasoftheGItractrepresentsamajor bottleneckincurrentdevelopmentoforallyadministered medica-tions[11].Thisismainlyduetounpredictablegastrictransittime andphysiochemicalcharacteristicsoftheGItractthatdifferfrom patienttopatient[12].

Theavailabilityofacontrollabledrugdeliverysystems(DDS) mayreducetoxicityassociatedwithsystemicadministrationand preventtheformationofantibodiestodrugs,particularlyinthecase ofcoloncancerorinflammatoryboweldiseasetherapies[13].These aretwoextremelycommondiseases,withcolorectalcancerbeing

abletodeployaliquiddrug.Kimetal.[21]proposedathree-axis Helmholtzcoilcapsulewhichmoveswitharotationalmotionand deploysadrug.Woodsetal.[22]integratedaholdingmechanism insidethecapsuletoresisttoperistalsisandaneedleforlocaldrug deployment.TheMAARScapsule,presentedin[23],iscomposed ofmagneticsemi-hardmodulesanddeploysasingleshotofdrug uponademagnetizationprocess.Yoetal.[24]recentlyproposed anMCRwherethedrugreleasecanbetriggeredthroughareed switch.Finally,Hafezietal.[25]developedaningestiblesensorthat allowspatients,families,andphysicianstomonitorprescription complianceanddrug-adherencepatternsinrealtime.

While each of these capsules taken alone solves a specific problem,noneofthemoffersacompletesolutiontoachievean intelligentrelease ofa drugona specifictarget. Inorder todo that, multiple functionalities need to becombined and several designiterationsneedtobetested.Thankstothemodularapproach adoptedinthiswork,wehavedoneafirststepinthatdirection byimplementingacontrollablemechanismfordrugreleaseand anintelligentscheduler.Thesecomponentstakeonlypartofthe availableslotsontheflexiblecircuitbackbone,leavingspacefor addingotherfunctionalmodules,suchaslocalizationand/oractive positioncontrol.

3. Designconsideration

3.1. Principleofoperation–mechanismconsideration

ReferringtoFig.2,wheretheprincipleofoperationoftheMCR ispresented,theDDCmechanismconsistsofadrugchamber,two coils,and themagnetic piston.The drugchamber(d=7.94mm, l=6.35mm,volume=314.42mm3)ishostedinacylindrical

enclo-suretogetherwithanaxiallymagnetizedcylindricalpermanent magnet acting asa piston (D54-N52, K&JMagnetics,USA). The DDCshellandthedrugchamberwereprototypedwithanObjet30 3Dprinter(Stratasys,USA)withVerowhitematerial.The clear-ancebetweentheouterdiameterofthemagneticpistonandinner diameterofthechamberwas0.4mm.Thisvalueguaranteedalow frictionwiththemagnet,resultinginnoleakageofthedrugand theactuatingmechanism.Thedistalcollaredgeofthechamberhas twelvecircularholes(eachwitharadiusofrh=0.8mm)fromwhere

thedrugisreleasedintotheenvironment.Thenumberofholesand theirradiushasbeenchosensuchthatthedrugisdeployed uni-formlywithoutbeingaffectedbycapillarity.Incasetheapplication requiresadrugwithadifferentviscosity,thenumberofholesas wellastheirdimensioncanbeadjusted.Thecombinationofproper drugviscosityandaperturesizepreventsthedrugfromleakingfor anypossibleorientationofthecapsule.Benchtrialsshowedthat theproposeddesignisabletodeploysolutionswithaviscosityup to1000cP,whichishigherthanwater(0.89cP),blood(between3 and4cP)andair(0.018cP).

Toactuatethemagneticpiston,twocoilsaremountedatboth sidesofthedrugchambersuchthat,whentheyareactivated,the

(3)

Fig.2.PrincipleofoperationoftheDDC(a).Whenthecoilsareactivated(b)the magneticpistondeploysthedrugoutsideofthedrugchamber.

currentIinducestwostaticmagneticfieldsalignedalongthesame axis,havingthesamemagnitude,butoppositedirections.

TheforceF(dz)inducedbythetwocoilsonthemagnetic pis-ton can be expressed as a function of the relative position dz of thepiston with respectto thetwo coils.The trend of F(dz) wasmodeledwithCOMSOLmultiphisics(COMSOLMultiphysics, Stockholm,Sweden)andexperimentallyverifiedasdiscussedin Section4.ConsideringthesurfaceofthemagneticpistonS,wecan derivethepressureP(dz)inthechamberasafunctionofthe mag-neticpistonpositiondz:

P(dz)=F(dz)

S . (1)

UsingtheHagen-Poiseuille’sequation[26],wecanthenexpress thefluidflow ˙V asa functionofthepressureassumingthatthe radiusoftheholesissmallerthantheirlength,thedrugisa New-tonianfluid,andthemotionoftheresultingdrugflowacrossthe holesislaminar.Undertheseassumptions,wehave:

˙V=n·P(dz) Rh ,

(2)

whereRh=8l/r4isthehydraulicresistance,isthedynamic

vis-cosityofthedrug,whilel,randnarethelength,theradiusandthe numberofholesinthechamber,respectively.

Fromtheflux ˙V,it ispossibletoderivethequantityofdrug releasedMas

M= ˙V·tact. (3)

withtactbeingthetimeperiodwhenthecoilisactive.

CoildimensioningThe magneticforceacting onthepistonis proportionaltothemagneticfieldBgeneratedbythecoilsas

F= B2·A

 (4)

Fig.3. PictureoftheembeddedelectronicsoftheDDCbeforefoldingtheflexible circuit.Thelateralconnectorsattheoppositesidesoftheflexiblecircuitareused forprogramminganddebuggingthemodulesandareremovedbeforeintegration intothecapsuleshell.

whereFistheforce,Bisthemagneticfield,Aisthesectionofthe coil,andisthemagneticpermeabilityofthemedium(i.e.,air, therefore=0r0).SinceBisaproportionaltothecurrentI,

thecoildimensions,andthenumberofturns,weneedtooptimize thethreequantitiestoachieveenoughforcetopushthemagnet alongthechamberagainstthedrug.

Ontheotherhand,theamountofcurrentinthecoilislimited bythepowerdissipatedbyitsinternalresistanceasheat.Forsafety reasons,thecapsuletemperatureshouldnotincreasebymorethan 1◦C[27].Sincethecoilsaremountedinaparallelconfiguration, theirtotalresistanceisRTC=1/2RCwithRC resistanceofasingle

coil.Themaximumcurrentthatcanflowthroughthecoilwire with-outcausinga1◦Cincreaseisequalto800mA.Adoptingavalueof 700mAforthecurrent,andconsideringthatthecoilissuppliedat 3.3V,wehave:

RTC= VI =3.30.7 =4.71 (5)

thus RC=9.42 for each coil. Since the wire resistivity is

=17.421·10−9manditsissectionSw=1.5386·10−8m2,the

lengthofthewireLresultsin: L=R·Sw

 =8.32m. (6)

Consideringthatthewireiswrappedaroundacircumferenceof Cin=0.0157m,thenumberofturnsNresultsin

N= L Cin =

8.32

0.0157=530.394≈530 turns. (7) Oncewrapped,thecoilresultsinadiameterof10mm,alengthof 18mm,andaninductanceof1.5mH.

3.2. HardwareArchitecture

ThehardwarearchitectureforthepresentedMCRconsistsof twoseparatesystems:abasestationandtheDDCitself.Thebase station[7]wasusedtoexchangewirelessdatabetweentheMCR andtheuserinterface.Inthisparticularapplication,theusermight desiretochangethedeliveredtherapyscheduleafterthecapsule hasbeeningested,thusthebasestationistheonlywaytointeract withtheDDC.

TheDDCelectronicsconsistsofthreeembeddedmodules:the MCU,a433MHztransceiver,andthepoweringandbattery mon-itoringmodule.Themodules,aremountedonourflexiblecircuit. These components are shown in Fig. 3 beforethe flexible cir-cuitisfoldedtofit acapsule-shapedshell.Theremainingthree emptyslotsontheflexible circuitcanbeusedin thefuture to add other functional modules. The MCU moduleis based ona

(4)

municatesviaSerialPeripheralInterface(SPI)withthe433MHz transceiver(CC1101,TexasInstruments,USA)module.Thedriver for the coil actuation circuitry consists of a power p-channel MOSFET(STR2P3LLH6,STMicroelectronics,SUI),whichhasbeen mounteddirectlyontopoftheflexiblecircuit.Thetwocoilsare connectedtothedrainoftheMOSFET.Thismodulecanprovide upto800mA.TheMOSFETgateisdrivenbyaMCUdigital out-putwithaPulseWidthModulation(PWM)signal.Finally,power totheDDCisprovidedbya30mA/hLiPorechargeablebattery (LP-FR30,PlantracoLTD,USA).Accordinglytothedrugscheduler,the DDCdeploysthedrugonlyatspecifictimes.Thebatterylifetime thereforehastobeoptimizedtominimizetheMCRpower con-sumption.ThisisachievedbyrunningtheMCUinlowpowermode, andactivatingitsperipheralinterfaces(e.g.,wireless communica-tion,actuation)ondemand.Table1showsthepowerconsumption oftheDDCembeddedelectronicsduringdifferentoperatingmodes. All the documentation related to the base station as well totheotherminiaturehardwaremodules isavailable at pill-forge.github.io.

3.3. Designenvironmentandscheduler

Developinganembeddedapplicationcanbeintimidating for noviceusers.Applicationsareusuallycodedwithlow-level lan-guages like C, which presents a barrier for researchers whose expertiseliesintheapplicationsofMCR.Inaddition,settingup toolchainsandthenecessarysoftwarelibrariesforaspecific tar-getdeviceisusuallycumbersome.Toalleviatethesehardships,we havedevelopedaweb-basedvisualsoftwaredesignenvironment [7].

Thedesignenvironmentprovidesahigherlevelofabstraction forapplications,makingiteasierfordeveloperstoreasonabout theirsoftware.Ausercancompilehis/herapplicationthroughthe designenvironmentanddownloaditsbinariesthroughtheweb interface.Sincethecompilationisprocessedontheserverside,the userdoesnotneedtoinstallanyadditionaltools.

We applieda proven model-based methodology [28] tothe MCRdomain.Thismethodallowedustovisuallyrepresent soft-waremodelsspecifictoMCRandtogeneratethecorresponding software artifacts. The compositional features of model-based development’sprovideadditionalabstractionsthatcan encapsu-late the underlying complexities. The design environment was implementedusingtheWeb-basedGenericModelingEnvironment (WebGME), which is a modeling tool used to design domain-specificvisualmodelinglanguages[29].

WehavechosenTinyOSasthecoreoperatingsystemforthe MCR.TinyOSisanevent-driven,operatingsystemcommonlyused insmallresourceconstraineddevices[30].Itscomponentbased paradigmlendsitselfnaturallytothemodel-basedmethodology ofourdesignenvironment.Hence,amodelinglanguagewas cre-atedusingWebGMEtorepresentTinyOSlanguageconstructs.A onetoonemappingofTinyOSsoftwarecomponentstoWebGME’s notion of components allows users to create applications and

thebasestation,thescheduleandviscosityblocksareconnectedto atemplatedrugdeliverybaseapp,whiletheheartbeatratevalue blockisconnectedtoatemplatedrugdeliveryMCRapp’sport.MCR sendsastatusmessageperiodicallytothebasestation.Thestatus messagesusedforthecommunicationprotocolbetweenMCRand thebasestationaswellastosignaltothebasestationthattheMCR isstillalive(hencethenameheartbeat).Oncethecommunicationis establishedbetweenthedevices,theremainingpercentageofdrug isusedasthepayloadinthestatusmessages.Theheartbeatrate valuedeterminesthefrequencyofmessagessendfromtheMCR.In addition,thetemplateappshaveattributestosettheradioaddress forwirelesscommunication.

TheinternalstructureofthescheduleblockcanbeseeninFig.4. Theblueobjectrepresentsthebeginningofaschedule,thegreen elementsrepresentstheamountofdrugtobereleased.Blueand greenelementsareconnectedwithtimeconnections.Theschedule blockdictateswhatpercentageofdrugshouldbereleasedinthe specifiedtimeintervals.

Oncethescheduleandtheparametersareset,ausercan com-piletheapplicationbyrunningthecompilerpluginviathearrow buttonlocatedontopleftofthedesignenvironment.The genera-tortraversestheinputdataandfeedsittothetemplateapp.The appgetscompiledontheserverside,theresultingbinaries,inturn, canbedownloadedbytheuser.Thesoftwareisavailableathttps:// github.com/pillforge/drug.delivery.

4. Deviceexperimentalassessment

ExperimentaltrialsaimedtoassessandcharacterizetheDDC functionalitywereperformed.Theinteractionforceexertedfrom thetwocoilsonthemagneticpistonwasmodeledand experimen-tallyvalidated.WealsoverifiedthattheDDC,onceprogrammed withascheduler,deliveredtheamountofdrugexpectedfromthe releaseprofiledescribedinSection3.1.Finally,thereliabilityand functionalityofthedeviceweretestedinaninvivotrialsonacanine model.

4.1. Validationoftheinteractionforcebetweenthecoilsandthe magnet

Theinteractionforcebetweenthecoilsandthemagnethasbeen modeledusingfiniteelementanalysis(COMSOL)byassumingthe twocoilsasHelmholtzcoilswiththefollowinggeometricaland electricalparameters:numberofturnsN=531,resistivityofthe wire=17.421·10−9m2,radiusofthecoilr

c=2.5mm,current

flowingineachcoilIc=350mA,diameterofthewiredw=0.2mm,

andlengthofthecoilL=1cm. Themotionof themagnetic pis-tonoccurs along thezdirectionoftheDDC, thuswe extracted fromthesimulationthenumericalvaluesfortheinteractingforce F(dz)asafunctionofthepistonpositionintothedrugchamber. Thesamequantitywasthenexperimentallymeasuredwiththe setuppresentedinFig.5.Arapidprototypingpart(Objet30,Objet

(5)

Fig.4.ThedesignenvironmentshowingBaseandMCRappswiththehighlightoftheschedule.

GeometriesLtd,USA)wasdesignedtomounta3-axisforce sen-sor(NANO17,ATIIndustrialAutomation,USA)withthecoilsand thedrugchamber. Opposite tothe chamber,the magnetic pis-tonwasmountedattheendeffectorofasix-degree-of-freedom

Fig.5. TheexperimentalsetuptoassesstheforceF(dz)insidethedrugchamber whilethecoilsareactive.

roboticmanipulator(RV6SDL,MitsubishiCorp.,Japan).Whilethe coilswereactive,themagneticpistonwasmovedinsidethe cham-berforitsentirelength,whiletheresultingforcewascollected usingaUniversal(SerialBus(USB)dataacquisitionboard(NI-PCI 6224,NationalInstruments,USA)atasamplingrateof100Hz.

Thegraphofthemodeledforceandtheexperimentaldataare representedinFig.6.Theplotshowsaparabolicandsymmetric trendfortheforce,whichreachesitsminimumexactlyatthe cen-terofthechamber,halfwaybetweenthecoils.Therelativeerror betweenthefiniteelementsimulationandtheforcesensordata was3.4%(=±2.5mN).

Inordertovalidatetheeffectofgravityonthedrugdelivery mechanism,thetrialswereperformedwiththecapsuleassuming differentorientations.Sincethecontributeofgravitationalforceis negligiblecomparedtothemagneticforcegeneratedbythecoil, themechanismwasabletowithstanditregardlessofthecapsule orientation.

Fig.6.Experimentaldatameasuredfromtheloadcellcomparedtothefinite ele-mentanalysissimulation.

(6)

releasewastriggered,theMCRwasplacedonadigitalscale (Ameri-canWeighScales,Inc,USA,resolution0.01±0.005g)anditsweight recorded.Thequantityofsolutionreleasedwithtime wasthus derivedfromthecapsuleinitialweight,whenthedrugchamber wascompletelyfilledup.Tendifferentexperimentaltrialswere thenperformed.Theresultsofoneofthetrialforthetwodifferent viscosityisshowninFig.7,wheretheexpecteddrugdeployment profile(inblue)iscomparedtotheactualexperimentalprofiledata (inred).Repeatabilitytestsfordelivering10%ofthedrugloaded intothechamberresultedinanaveragedrugaliquotof72␮Lwith astandarddeviationof0.28␮L.

Agoodmatchingoftheexperimentaldatawiththeexpected drugreleaseprofilewasobserved.Overall,therelativeerrorforthe quantityofdrugreleasedvs.itsexpectationis2.4%±1.8%.Current consumptionofthedevicematchedthevaluesreportedinTable1 witharelativeerrorequalto3%.Inparticular,thecurrentinthe coilwas695mA.

4.3. Invivotrial

TheinvivotrialwasperformedatVanderbiltUniversityunder protocolM/14/014,thatincludedotherdevicesandexperiments, usinga25kgfemaledogundergeneralanesthesia.TheDDCdrug chamberwasfilledwithasolutionof60%water and40% lubri-cantwithagreendye,resultinginadynamicviscosityof0.05Pas. Thecapsulewasthenprogrammedtoscheduleadrugdeliveryof 7␮Levery15suntilallthedruginthechamberwascompletely deployedforatotalof6doses.

TheDDC wasinsertedfromthe anusintothecolon using a standardcolonscopeandleftinplacefortheamountoftimeneeded forthetrial.Thesamecolonscopewasusedtoobservethereleaseof thedrugunderinsufflationofthecolon.Fig.8showsthesequenceof theDDCdeployingascheduleddrugdoses:thedrugreleaseisfirst triggered(A),thenthedrugstartstoflowoutsideofthechamber (B)andreachesthecolontissue(C).

Fig.7. Predicteddrugdeliveryprofile(inblue)comparedwiththeexperimental data(inred)foroneofthetrial.(Forinterpretationofthereferencestocolorinthis figurelegend,thereaderisreferredtothewebversionofthisarticle.)

ThecurrentconsumptionoftheDDCduringthetrialwas esti-matedasin[31]usingthefollowingaveragecurrentconsumptions: ¯I =



iIiTi



iTi , (8)

where Iiand Ti arethe currentconsumption from Table1 and

thetimeintervalrelatedtoasingleactionperformedbytheDDC, respectively. For our particular application ¯I resulted equal to 28mA.Adopting a battery witha capacity of250mAh, wecan expect a maximum lifetimeof T=250/288hand 50min.The invivotrialshowedthatthedevicewasabletooperateforatleast 5hand24min.Afterthat,allthedrugwasdeployedandthetrial wasconsideredasconcluded.Incasethatalongeroperatingtime wereneeded,batterylifetimecanbefurtherextendedby increas-ingtheMCUsleepintervals.Crohn’sdiseaseandulcerativecolitis areinflammatorysmallboweldiseasesthataretypicallytreatedby releasingcontrolledquantitiesofdrugsoveraperiodoftime.While thetypicaltransittimefor50%offoodinthesmallbowelisupto threehours,humanclinicaltrialsinthisGItractforsimilarMCR,

(7)

Fig.9.(a)ApictureofthefinalDDCprototype,wheretheembeddedelectronicswasdevelopedviacustomdesign.Thedeviceis25mmlongforadiameterof16mm.(b) Thedrugstarstobereleasedand(c)iscompletelydeployed.

(e.g.theIntellicap)haveshownanaveragesmallboweltransitofup to5h.Theactualbatterylifetimeguaranteesenoughtimeforthe devicetooperatewhiletravelingacrossthesmallbowel.However, sincethedeliverytimeandthequantityofdrugdeployedcanbe adjustedwiththescheduler,differentdiseasescanbetargeted.The purposeoftheanimaltrialwastoqualitativelyassessthe function-alityoftheDDCinaclinicalscenario.Themainlimitationofthe trialcomparedtoaclinicalscenariowasthatthedevicewasnot swallowed,becauseoftheprotocoltrialtime.Thegoalofthetrials wastoassessifthedevicewasabletoproperlyoperateinsidea constrainedenvironment,beingabletorespondtowireless com-mands,andreliablyactuatethemechanism.Forthispurpose,we firsttriedthedevicewithoutinsufflation.Afterthescheduledtime ofdrugdelivery,weinsufflatedthecolonandwecheckedproper deploymentofthedrug.Atthispoint,werunatrialwithinsufflation tocaptureavideofootage.

5. Discussion

Asdiscussedin [7],themainpurposeofourmodulardesign environmentistoacceleratethetime-to-prototypefortheearly validationofdesignhypothesisinthefieldofMCRs.The time-to-prototypedependsprincipallyonthecomplexityofthehardware andsoftwareoftheMCRaswellastheskill-setandexperienceof thedeveloper.

Inparticular,hardware complexityconsistsofcircuitdesign, assemblyofminiaturecomponents,andtesting.Whileacustom approach requires the developer to go through these iterative stepsuntilthedesiredlevelofminiaturizationandperformance isachieved,ourmodulardesignenvironment providestheuser withminiaturizedhardwaremodulesandaflexiblecircuitfor eas-ierconnectivity.Thesamedegreeofminiaturizedconnectivitycan beachievedincustomdesignswithflexiblePCBcablesorhybrid PCBsolutions,buttheyareknowntobeexpensiveforthelow vol-umesthataretypicalforresearchandproof-of-conceptvalidation [32].

SoftwarecomplexityconsistsofwritingefficientembeddedC codetoimplementthevariousfunctionsofanMCRandto com-municatewithexternalcomponentssuchassensors,actuatorsand wirelessbasestations.Instead,ourmodulardesignenvironment providesthedeveloperwithasetofsoftwarecomponents,which canbe customizedaccording totherequirementsof the appli-cation.Ultimately,theexperienceandskill-setofthedevelopers involveddetermineshowefficientlythedescribedhardwareand softwarecomplexitiesaremanaged.Exactquantitativebenefitsof ourcomponentbaseddesignfurtherdependonthespecificsofthe application,however,togivetothereaderanideaofthetimeand costoutcomes,theswimmingcapsuleMCRdescribedin[7],has beendevelopedandassembledinlessthantwomonthswithan overallestimatedcostof15,000USD(includingmaterialsand sup-portforpersonnel)comparedtoninemonthsandanestimatedcost of50,000USDfortheMCRpresentedin[33].

Itisworthmentioningthatthehardwareplatformwepropose, beingasystemmadeofmodules,isnotoptimizedforanyparticular application.Thus,MCRsbuiltwithourplatformaregenerallylarger thantheircustomcounterparts,asshowninFig.3comparedto Fig.9.

6. Conclusions

Inthiswork,wepresentedacomponentbaseddesignofaDDC robotbasedonacoil-magnet-pistonmechanism.Thisdeviceaims toprovideacontrolleddeploymentofdrugdosesovertimewitha scheduler.Thedrugviscosityanddeploymenttimecanbesetinthe applicationtemplateforthedesignenvironmentaccordingtothe desireddrugreleaseprofile.Themagneticforceaswellasthedrug releaseprofilehavebeenmodeledandthenexperimentally veri-fied.Themodelsfortheforceandthereleaseddrugprofileshowed relativeerrorsbelow5%whencomparedtotheexperimentaldata. ThefunctionalityoftheDDCwasalsosuccessfullyassessedina clinicalscenarioviaaninvivotrial.

Inthiswork,wevalidatedthehypothesisthata coil-magnet-pistonmechanismisaviablesolutionforascheduleddrugrelease. However,theDDCthatwedevelopedbyassemblingthe preexis-tingmodulesfromourlibraryresultedinasizethatisimpossibleto swallow.Therefore,oncewewereconfidentwiththe coil-magnet-pistonmechanism operation, wescaleddownourdesign toan ingestiblesizebydesigningacustomflexiblecircuit.TheDDC rep-resentedinFig.9is16mmindiameterand25mminlengthwith adrugreservoirof201.06mm3.Thiscapsulecontainsascaled

ver-sionofthecoil-magnet-pistonmechanismdescribedinthispaper anditiscapableofdeployingdrugswithviscositiesupto700cP. TheDDCweightwasdecreasedfrom18gto12g,whilepower con-sumptionintheTxorRxmodeandintheDrugDeploymentmode werereducedfrom33.8mAto24.2mAandfrom748mAto418mA, respectively.

Beyondthespecificdesign,modeling,and assessmentof the proposedcoil-magnet-pistonmechanism,themaincontribution ofthisworkistoillustratehowourcomponentbaseddesign envi-ronmentcanbeleveragedfortheearlyvalidationofanovelMCR design.Allthehardwareandsoftwarecomponentsdescribedinthis paperareavailableopen-sourceatpillforge.github.ioandcan beusedasatemplatebytheMCRresearchcommunity.

FutureworkaimstoexpandtheDDCtemplatewithmodules formagneticlocomotion[3],anchoringforlocalizeddrugdelivery, andmagneticlocalization[34]towardtheendgoalofintelligent drugdeliveryintheGItract.

Acknowledgment

Thismaterialisbaseduponworksupportedinpartbythe Van-derbilt Institutein Surgery and Engineeringand in partby the NationalScienceFoundationundergrantsnumberCNS-1239355 and IIS-1453129. Any opinions, findings and conclusions or

(8)

roboticcapsuleforreleasingbioadhesivepatchesinthegastrointestinaltract,

J.Med.Dev.8(1)(2014)014503.

[5]W.H.Crosby,U.Army,H.W.Kugler,Intraluminalbiopsyofthesmallintestine,

Am.J.Dig.Dis.2(5)(1957)236–241.

[6]F. Munoz,G. Alici,W.Li,Areview ofdrug deliverysystemsforcapsule

endoscopy,Adv.DrugDeliv.Rev.71(2014)77–85.

[7]M.Beccani,H.Tunc,A.Taddese,E.Susilo,P.Volgyesi,A.Ledeczi,P.

Valdas-tri,Systematicdesignofmedicalcapsulerobots,IEEEDes.Test32(5)(2015)

98–108.

[8]A.Taddese,M.Beccani,E.Susilo,P.Völgyesi,A.Lédeczi,P.Valdastri,Toward

rapidprototypingofminiaturecapsulerobots,in:2015IEEEInternational

Con-ferenceonRoboticsandAutomation(ICRA),ICRA2015,2015,pp.4704–4709.

[9]M.Beccani,E.Susilo,C.DiNatali,P.Valdastri,SMAC–amodularopensource

architectureformedicalcapsulerobots,Int.J.Adv.Robot.Syst.11(2014)1–16.

[10]P.Valdastri,M.Simi,R.J.WebsterIII,Advancedtechnologiesforgastrointestinal

endoscopy,Annu.Rev.Biomed.Eng.14(5)(2012)397–429.

[11]S.Stegemann,F.Leveiller,D.Franchi,H.DeJong,H.Lindén,Whenpoorsolubility

becomesanissue:fromearlystagetoproofofconcept,Eur.J.Pharm.Sci.31

(5)(2007)249–261.

[12]J.F.Pinto,Site-specificdrugdeliverysystemswithinthegastro-intestinaltract:

fromthemouthtothecolon,Int.J.Pharm.395(1)(2010)44–52.

[13]P.J.vanderSchaar,J.F.Dijksman,H.Broekhuizen-deGast,J.Shimizu,N.van

Lelyveld,H.Zou,V.Iordanov,C.Wanke,P.D.Siersema,Anovelingestible

elec-tronicdrugdeliveryandmonitoringdevice,Gastrointest.Endosc.78(3)(2013)

520–528.

[14]D.M.Parkin,F.Bray,J.Ferlay,P.Pisani,Globalcancerstatistics,2002,CA:Cancer

J.Clin.55(2)(2005)74–108.

[15]N.K.Shah,B.Rane,N.Gujarathi,Developmentsincolonspecificdrugdelivery systems–areview,PharmaScienceMonitor5(2)(2014)95–110.

[16]A.Connor,Location,location,location:gastrointestinaldeliverysiteandits

impactonabsorption,Ther.Deliv.3(5)(2012)575.

[17]D.Becker,J.Zhang,T.Heimbach,R.C.Penland,C.Wanke,J.Shimizu,K.

Kul-matycki,Novelorallyswallowableintellicap®devicetoquantifyregionaldrug

absorptioninhumangitractusingdiltiazemasmodeldrug,AAPS

Pharm-SciTech15(6)(2014)1490–1497.

[18]Scintipharmawebsite.www.intuitivesurgical.com.

[19]I.Wilding,Theenterioncapsule:anoveltechnologyforunderstandingthe

biopharmaceuticalcomplexityofnewmolecularentities(NMES),DrugDeliv.

Technol.1(1)(2001)8–11.

[20]S.Yim,M.Sitti,Shape-programmablesoftcapsulerobotsforsemi-implantable

drugdelivery,IEEETrans.Robot.28(5)(2012)1198–1202.

[21]S.H.Kim,K.Ishiyama,Magneticrobotandmanipulationforactive-locomotion

with targeted drug release, IEEE/ASME Trans. Mechatron. 19(5) (2014)

1651–1659.

[22]S.P.Woods,T.G.Constandinou,Wirelesscapsuleendoscopefortargeteddrug

delivery:mechanicsanddesignconsiderations,IEEETrans.Biomed.Eng.60(4)

(2013)945–953.

[23]C.T.Dietzel,H.Richert,S.Abert,U.Merkel,M.Hippius,A.Stallmach,Magnetic

activeagentreleasesystem(MAARS):evaluationofanewwayfora

repro-ducible,externallycontrolleddrugreleaseintothesmallintestine,J.Control.

Release161(3)(2012)722–727.

[24]W.Yu, R.Rahimi,M.Ochoa,R. Pinal,B. Ziaie,Asmartcapsulewith

GI-tract-location-specificpayloadrelease,IEEETrans.Biomed.Eng.62(9)(2015)

2289–2295.

[25]H.Hafezi,T.L.Robertson,G.D.Moon,K.-Y.Au-Yeung,M.J.Zdeblick,G.M.Savage,

Aningestiblesensorformeasuringmedicationadherence,IEEETrans.Biomed.

Eng.62(1)(2015)99–109.

[26]C.Loudon,K.McCulloh,ApplicationoftheHagen-PoiseuilleEquationtofluid

feedingthroughshorttubes,Ann.Entomol.Soc.Am.92(1)(1999)153–158.

[27]EN45502-1:2015,Implantsforsurgery.Activeimplantablemedicaldevices,in:

GeneralRequirementsforSafety,MarkingandforInformationtobeProvided

bytheManufacturer,2015,pp.1–64.

[28]G.Karsai,J.Sztipanovits,A.Ledeczi,T.Bapty,Model-integrateddevelopment

ofembeddedsoftware,Proc.IEEE91(1)(2003)145–164.

[29]M.Maróti,T.Kecskés,R.Kereskényi,B.Broll,P.Völgyesi,L.Jurácz,T.

Leven-doszky,Á.Lédeczi,Nextgeneration(meta)modeling:web-andcloud-based

collaborativetoolinfrastructure,in:ProceedingsofMPM,2014,p.41.

[30]P.Levis,S.Madden,J.Polastre,R.Szewczyk,K.Whitehouse,A.Woo,D.Gay,

J.Hill,M.Welsh,E.Brewer,etal.,Tinyos:anoperatingsystemforsensor

networks,in:AmbientIntelligence,Springer,2005,pp.115–148.

EngineeringfromUniversityofPisa,Pisa,Italy,in2010 andthePh.D.degreeinMechanicalEngineeringfrom Van-derbiltUniversityin2015. Heis currentlyapostdoc researcherattheembeddedandrealtimesystemlab (MLAB)attheUniversityofPennsylvaniainPhiladelphia. Hisresearchinterestincludesembeddedsystemdesign forminiaturizedwirelessmedicalroboticsforcapsular endoscopyandroboticsurgery,andmodelbaseddesign ofmedicaldevices.

GregorioAielloreceivedtheB.Sc.,andM.Sc.degreesin BiomedicalengineeringfromPolitecnicodiMilano,Milan, in2014,and2015,respectively.Heiscurrentlyaresearch scholaratARMALab,mechanicalengineeringat Vander-biltUniversity,Nashville,TN,USA.During2015hewasan internstudentatSTORMlab,Mechanicalengineeringat VanderbiltUniversity,whereheworkedtothedesignof MedicalCapsuleRobot.

NikolaosGkotsisiscurrentlyanundergraduate Mechan-icalEngineeringstudentatVanderbiltUniversity.Forthe last12monthshehasbeendoingresearchattheSTORM LabofVanderbiltUniversity,withmainfocusonMedical CapsuleRobots.Hehasalsobeendoingresearchonsoft robotictissuepalpation.

HakanTuncisaPh.D.studentandaresearchassistantattheInstituteforSoftware IntegratedSystems,VanderbiltUniversity.Hisresearchfocusesondesign environ-mentforminiaturemedicalrobotics.

AddisuZ.TaddeseisaPh.D.studentatVanderbilt Univer-sity.Hisresearchfocusesonminiaturemedicalrobotics andtheirapplicationinsurgicalprocedures.HeisanNSF graduatefellowandstudentmemberofIEEEandACM.

EkawahyuSusiloisaPostdoctoralScholarin Mechan-icalEngineering,VanderbiltUniversity.Hismajorisin wirelessembeddedsystemdesignandminiaturization forlow-powerapplications.HereceivedhisPh.D.degree, withhonor,fromScuolaSuperioreSant’Anna(SSSA),Pisa, Italyin2009.HejoinedDecawaveLtd.in2010, develop-ingRTLSalgorithmusingIEEE802.15.4aUWBPHY/MAC compliantdeviceswith10cmprecisioninthree dimen-sionalcoordinates.Currently,heisjoiningSTORMLabat VanderbiltUniversity.

(9)

PeterVolgyesiisaResearchScientistattheInstitutefor SoftwareIntegratedSystemsatVanderbiltUniversity.He isoneofthearchitectsoftheGenericModeling Environ-ment,awidelyusedmetaprogrammablevisualmodeling tool,andWebGME–itsmodernweb-basedvariant.As PIontwoNSFfundedprojectsMr.Volgyesiandhisteam recentlydevelopedalow-powersoftware-definedradio platform(MarmotE) andacomponent-based develop-menttoolchaintargetingmulticoreSoCarchitecturesfor wirelesscyber-physicalsystems.

AkosLedecziisanAssociateProfessorofComputer Engi-neeringandaSeniorResearchScientistattheInstitutefor SoftwareIntegratedSystemsatVanderbiltUniversity.He hasanM.Sc.fromtheTechnicalUniversityofBudapestand aPh.D.fromVanderbilt,bothinElectricalEngineering.His researchinterestsincludewirelesssensornetworksand modelintegratedcomputing.

Elena De Momi, MSc in Biomedical Engineering in 2002,PhDinBioengineeringin2006,currentlyAssistant ProfessorinElectronicInformationandBioengineering Department(DEIB)ofPolitecnicodiMilano.Shewas co-founderoftheNeuroengineeringandMedicalRobotics Laboratory,in2008,being responsibleoftheMedical Roboticssection.ShehasbeenanAssociateEditorofthe JournalofMedicalRoboticsResearchandofthe Interna-tionalJournalofAdvancedRoboticSystems.

PietroValdastrireceivedtheMaster’s(Hons.)degreein ElectronicEngineeringfromUniversityofPisa,Pisa,Italy, in2002andthePh.D.degreeinBiomedicalEngineering fromScuolaSuperioreSant’Anna,Pisa.Afterfiveyears asAssistantProfessorwiththeDepartmentof Mechan-icalEngineering, Vanderbilt University, Nashville,TN, USA,heisnowFullProfessorandChairofRoboticsand AutonomoussystemsattheUniversityofLeeds,UK.Pietro ValdastriistheDirectoroftheSTORMLabandasenior memberofIEEE.Prof.ValdastrireceivedtheNational ScienceFoundationCareerAwardandtheRoyalSociety WolfsonResearchMeritAwardtostudyanddesign cap-sulerobotsformedicalapplications.

Riferimenti

Documenti correlati

The target protein can be expressed with different fusion partners, which has been developed to increase the expression yield and the solubility of the recombinant proteins, even if

levels; once income begins to increase, families begin to disperse their food budget shares on different foods, substituting low-quality foods with a more varied range of

The modest enhancements in SFF at tangent- point longitudes present when the AMCs are included largely dis- appear with their removal, leaving a flatter distribution of SFF

riprodotte integralmente ma non nella sequenza originale. Dunque la copia di età moderna, redatta per motivi di contenzioso, riproduce nell’insieme l’origina­ le andato

Infatti, al di là dell'indubbia rivoluzione copernicana che produrrebbe nel nostro ordina- mento la traduzione in legge di una delle pur diverse proposte oggi in di-

Gradients in Faraday depth with respect to position on the sky result in corresponding gradients in the polarization angle (if the intrinsic emitted angles are uniform across

La concezione della via verde come fenomeno di transizione che possa portare alla creazione di un mercato dominato, se non intera- mente costituito, dall’editoria ad accesso

Table 3 Summary of Cases Treated from 2005 to 2007 Patient Age Cause Type Materials 1 25 Sports Large Nonresorbable 2 83 Vehicle accident Small Resorbable 3 49 Assault