Two-color widefield fluorescence microendoscopy enables multiplexed molecular imaging in the

15 Pages·2016·9.3 MB·English

Durham Research Online
Deposited in DRO:
28 April 2016
Version of attached (cid:28)le:
Published Version
Peer-review status of attached (cid:28)le:
Peer-reviewed
Citation for published item:
Krstaji(cid:1)c, Nikola and Akram, Ahsan R. and Choudhary, Tushar R. and McDonald, Neil and Tanner, Michael
G. and Pedretti, Ettore and Dalgarno, Paul A. and Schole(cid:28)eld, Emma and Girkin, John M. and Moore, Anne
and Bradley, Mark and Dhaliwal, Kevin (2016) ’Two-color wide(cid:28)eld (cid:29)uorescence microendoscopy enables
multiplexed molecular imaging in the alveolar space of human lung tissue.’, Journal of biomedical optics., 21
(4). 046009.
Further information on publisher’s website:
http://dx.doi.org/10.1117/1.JBO.21.4.046009
Publisher’s copyright statement:
(cid:13)c The Authors. Published by SPIE under a Creative Commons Attribution 3.0 Unported License. Distribution or
reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.
[DOI: 10.1117/1.JBO.21.4.046009]
Additional information:
Use policy
Thefull-textmaybeusedand/orreproduced,andgiventothirdpartiesinanyformatormedium,withoutpriorpermissionorcharge,for
personalresearchorstudy,educational,ornot-for-pro(cid:28)tpurposesprovidedthat:
• afullbibliographicreferenceismadetotheoriginalsource
• alinkismadetothemetadatarecordinDRO
• thefull-textisnotchangedinanyway
Thefull-textmustnotbesoldinanyformatormediumwithouttheformalpermissionofthecopyrightholders.
PleaseconsultthefullDROpolicyforfurtherdetails.
DurhamUniversityLibrary,StocktonRoad,DurhamDH13LY,UnitedKingdom
Tel:[phone]|Fax:[phone]
https://dro.dur.ac.uk
Two-color widefield fluorescence
microendoscopy enables multiplexed
molecular imaging in the alveolar
space of human lung tissue
Nikola Krstajic´
Ahsan R. Akram
Tushar R. Choudhary
Neil McDonald
Michael G. Tanner
Ettore Pedretti
Paul A. Dalgarno
Emma Scholefield
John M. Girkin
Anne Moore
Mark Bradley
Kevin Dhaliwal
Nikola Krstajic´, Ahsan R. Akram, Tushar R. Choudhary, Neil McDonald, Michael G. Tanner,
Ettore Pedretti, Paul A. Dalgarno, Emma Scholefield, John M. Girkin, Anne Moore, Mark Bradley,
Kevin Dhaliwal, “Two-color widefield fluorescence microendoscopy enables multiplexed molecular imaging
in the alveolar space of human lung tissue,” J. Biomed. Opt. 21(4), 046009 (2016), doi: 10.1117/1.
JBO.21.4.046009.
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/28/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
JournalofBiomedicalOptics21(4),046009(April2016)
Two-color widefield fluorescence microendoscopy
enables multiplexed molecular imaging in the
alveolar space of human lung tissue
Nikola Krstajic´,a,b,* Ahsan R. Akram,a Tushar R. Choudhary,a,c Neil McDonald,a Michael G. Tanner,a,d
Ettore Pedretti,a,c Paul A. Dalgarno,c Emma Scholefield,a John M. Girkin,e Anne Moore,a
Mark Bradley,a and Kevin Dhaliwala,*
aUniversityofEdinburgh,Queen’sMedicalResearchInstitute,MRCCentreforInflammationResearch,EPSRCIRC“Hub”
inOpticalMolecularSensingandImaging,47LittleFranceCrescent,EdinburghEH164TJ,UnitedKingdom
bUniversityofEdinburgh,SchoolofEngineering,InstituteforIntegratedMicroandNanoSystems,EdinburghEH93JL,UnitedKingdom
cHeriot-WattUniversity,InstituteofBiologicalChemistry,BiophysicsandBioengineering,EdinburghEH144AS,UnitedKingdom
dHeriot-WattUniversity,InstituteofPhotonicsandQuantumSciences,EdinburghEH144AS,UnitedKingdom
eDurhamUniversity,BiophysicalSciencesInstitute,DepartmentofPhysics,SouthRoad,DurhamDH13LE,UnitedKingdom
Abstract. Wedemonstrateafasttwo-colorwidefieldfluorescencemicroendoscopysystemcapableofsimulta-
neouslydetectingseveraldiseasetargetsinintacthumanexvivolungtissue.Wecharacterizethesystemfor
lightthroughputfromtheexcitationlightemittingdiodes,fluorescencecollectionefficiency,andchromaticfocal
shifts.Wedemonstratetheeffectivenessoftheinstrumentbyimagingbacteria(Pseudomonasaeruginosa)inex
vivo human lung tissue. We describe a mechanism of bacterial detection through the fiber bundle that uses
blinking effects of bacteria as they move in front of the fiber core providing detection of objects smaller than
the fiber core and cladding (∼3 μm). This effectively increases the measured spatial resolution of 4 μm. We
show simultaneous imaging of neutrophils, monocytes, and fungus (Aspergillus fumigatus) in ex vivo human
lung tissue. The instrument has 10 nM and 50 nM sensitivity for fluorescein and Cy5 solutions, respectively.
Lung tissue autofluorescence remains visible at up to 200 fps camera acquisition rate. The optical system
lendsitselftoclinicaltranslationduetohigh-fluorescencesensitivity,simplicity,andtheabilitytomultiplexsev-
eral pathological molecular imaging targets simultaneously. © The Authors. Published bySPIE under a Creative Commons
Attribution3.0UnportedLicense.Distributionorreproductionofthisworkinwholeorinpartrequiresfullattributionoftheoriginalpublication,including
itsDOI.[DOI:10.1117/1.JBO.21.4.046009]
Keywords:fluorescence;microendoscopy;molecularimaging;lung;respiratorymedicine.
Paper150740RRRreceivedNov.4,2015;acceptedforpublicationMar.24,2016;publishedonlineApr.27,2016.
1 Introduction andinflammatorycells.Themostcommonmethodtodetermine
thebacterialburdeninrespiratorycriticalcarereliesonthecul-
Clinical molecularimaging (MI) covers abroadrange oftech-
tureofbronchoalveolarlavagefluid(BALF).Themajorlimita-
niques, including fluorescence, positron emission tomography,
tionsofBALFincriticallyillpatientsarethetimetakentoyield
single photon emission computed tomography (CT), and mag-
neticresonanceimaging.1MIaimstoimprovediagnosisbyuti- aresult,whichcanbeupto48h,10andcontaminationbyproxi-
mal airways sampling. During this intervening time, patients
lizing targeted reporters (smartprobes) for specific disease
may be prescribed inappropriate therapy or deteriorate rapidly
targets in tissue. In this work, we focus on advancing fluores-
while awaiting a confirmatory diagnosis. Therefore, there is a
cence-based MI microendoscopy of the distal lung (gas-
exchanging alveolar regions).2 Optical molecular imaging clinicalunmetneed11forimmediateinformationthatcandeter-
minetheabsenceorpresenceofbacteriaandmarkersofinflam-
(OMI)hasproventobeeffectiveagainstseveraldiseasetargets,
mationsuchasinfiltratinginnateimmunecells(e.g.,monocytes
suchasbacterialinfection,3inflammation,4andcancer.5Several
and neutrophils) associated with suspected pneumonia in criti-
candidatesmartprobeshavebeendesignedtargetingneutrophil
cally ill patients.
recruitment,6bacterialdetection,7andfibrogenesis.8Ourevolv-
Our approach is based on using smartprobes coupled with
ingstrategyistodeploycustomsmartprobesalongsidepulmo-
miniaturefiber-optic imaging bundles12 to access distal alveolar
narymicroendoscopyinvivotoimagediseaserelevanttargets.
regionsandtransmittheimagefromthedistalendofthefiberto
One particular area where disruptive optical technologies
therestoftheoptics.Thetechniqueisusedinconjunctionwith
would potentially have high impact is in the rapid diagnosis
flexiblebronchoscopy(FB).13FBisabletoaccesstrachea,bron-
of lung inflammation and infection in critically ill-ventilated
chi, and bronchioles, but cannot access the distal lung. FB was
patients, which has high level of mortality, often above
introduced by Ikeda et al.,14 who pioneered fiber bundle based
70%.9 Postmortem, pneumonia is pathologically defined on
bronchoscopy in 1968 and subsequently worked on improving
biopsies ofalveolar(distal lung)tissueinfiltratedwithbacteria
the imaging by introducing charge couple device (CCD)
camerastothetip.Flexiblebronchoscopeshavetips4to6mm
indiameteranddeploywhitelightendoscopyfornavigation,with
*Addressallcorrespondenceto:NikolaKrstajic´,E-mail:[email protected];
KevinDhaliwal,E-mail:[email protected] somesystemsutilizingnarrow-bandspectralimagingtoincrease
JournalofBiomedicalOptics 046009-1 April2016 (cid:129) Vol.21(4)
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/28/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
Krstajic´etal.:Two-colorwidefieldfluorescencemicroendoscopyenablesmultiplexed...
contrastonsuperficialbloodvessels.15,16Autofluorescenceofthe spectral range prior to collecting the image on a color CMOS
lung is complementary towhite light endoscopy in providing a camera. Confocal microendoscopy is an alternative method30
convenientwaytonavigatetothesiteofinterest.Forillumination deployedbothinacademia31andclinics20–22andprovidesopti-
inthe460to490nmrange,thelungautofluorescenceemission cal sectioning, permitting higher contrast imaging, as tissue
spectrapeakbetween520and570nm.Autofluorescenceendos- background fluorescence is largely suppressed.32 Two-color
copy has been used to detect pathologies in tissue since 1990s, confocal microendoscopy has also been explored, demonstrat-
especially in the gut.17 However, interpatient variations in ing good contrast on avariety of tissues.33 However, the scan-
lung autofluorescence pose a serious problem in using ning required for confocal systems is inherently slower than
autofluorescence alone clinically,18 which significantly reduces camera-based imaging methods.
specificity.19 Autofluorescence from the lung originates from Whilesinglecolorconfocalmicroendoscopyhasbeenevalu-
elastin,20 and the autofluorescence spectra demonstrate that the atedforbronchialairwayanddistallungimaging,34multicolor
redregion(650to700nm)haslowerautofluorescenceintensity widefield camera-based microendoscopy remains to be fully
characterized. Fiber bundle imaging and bacterial dynamics,
(5to10timeslower)thanthegreenregion(500to600nm).This
in the case of mycobacteria, have also recently been
providesastrongmotivationforthedesignofsmartprobesinthe
studied.35Inthiswork,weshowthatcamera-basedsystemspro-
redandnear-infrared(NIR)spectralranges(600nmto1000nm).
vide a robust and economical route to multicolor fluorescence
Microendoscopy can be used in bronchoscopy by inserting
detection.Whitelightendoscopydominatesbronchoscopypro-
thenarrowfiberimagingbundlethroughtheworkingchannelof
cedures, andour aimistoenhancefluorescence endoscopy by
the bronchoscope. The diameter of the working channel avail-
integratingsmartprobesinmulticolorscenarios.Here,theability
ableonmodernflexiblebronchoscopesrangesfrom1to3mm
ofthewidefieldsystemtooperateinthedistallungbyimaging
depending on the manufacturer. For distal lung imaging, nar-
targetsranginginsizefrom2to100 μmsuchasbacterialcol-
rower microendoscope probes are necessary to enable safe
onies,monocytes,neutrophils,andthefungusAspergillusfumi-
access to the alveolar space. The distal alveolar regions are
gatus deposited in ex vivo human alveolar lung tissue is
reached by performing a transbronchial pass. Bronchoscopic
demonstrated. Also, the blinking effect of fluorescing objects
interventions usually last 20 to 30 min, and prior to the pro-
smallerthanthefibercoreandcladdingisexploitedtoaugment
cedure, navigation is often guided by x-ray radiographs or
bacterial imaging.
CT scans, allowing for targeted regional exploration. Optical
microendoscopyenablestheexplorationofpulmonarysegments
2 Materials and Methods
byallowingmultipleendobronchialpassesintothedistallung.
Thisisideallysuitedtoimagethediffuseconditionsseenwith
2.1 Optical System
suspectedpneumoniaorpulmonaryinflammation.Thismethod-
ology has been demonstrated in ex vivo large animal models The diagram of the optical system is shown in Fig. 1. The
usingsmartprobes8andinvivoinhumansforlabelfreeparen- architecture extends standard widefield endoscopy to dual
chymal assessment of the distal lung.20–22 colorwhiletakingadvantagesofimprovedimagingandillumi-
Previousworkincolonoscopy,23ovariancancer,24detection, nation technologies.29 All components are off-the-shelf com-
andpreclinicalmouseimaging25havebeenreported,andmulti- mercial items. The proximal end of the fiber imaging bundle
colorfluorescenceendoscopyhasalonghistory.26,27Ourinstru- (Alveoflex™, Mauna Kea Technologies, Paris, France) is
ment is based on a single-color fluorescence endoscopy brought to the focus of the objective by mounting it onto a
platform28,29 with the fluorescence filter-set exchanged for 30-mmcagedZ-axistranslationmount(SM1Z,Thorlabs)ena-
that of an integrated two color, green and red, fluorescence bling fine focusing and optimized coupling of the image from
Fig.1 Diagramofthetwo-colorfluorescencesystem.TwoLEDsarecombinedwithadichroicmirror,and
illuminationissenttothemicroscopeobjectiveviatheemissionfilterandanothertwo-banddichroic.
Fluorescence from the imaging bundle is focused onto the color CMOS camera via a tube lens
(200mmfocallength).
JournalofBiomedicalOptics 046009-2 April2016 (cid:129) Vol.21(4)
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/28/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
Krstajic´etal.:Two-colorwidefieldfluorescencemicroendoscopyenablesmultiplexed...
the distal end of the fiber to the rest of the optics. The from the setup onto the camera and filtered by the Bayer filter
Alveoflex™ imaging fiber has approximately 30,000 cores ontheCMOSchip.AsshownonFig.2(c),theBayercolorfilter
withoutanylensingonthetip(i.e.,worksbycontactimaging). overlaps with the green channel of the two-color filter-set used
Figure 2(a) shows the fiber cores of the Alveoflex™ imaging (XF454, Horiba, UK), while the overlap in the red channel is
bundle. Two LEDs at 470 and 625 nm (M470L3 and smaller. Smaller overlap in the red channel is offset by greater
M625L3, Thorlabs) are used as illumination sources coupled separationofthetwochannels(∼100 nm),whichreducesspec-
to achromatic condenser lenses (ACL2520-A, Thorlabs for tral mixing.
470-nm LED and ACL2520-B, Thorlabs for 625 nm LED),
see Fig. 2(b) for LED spectra. Thecollimated beams are com-
2.2 Optical System Characterization
bined using a dichroic beamsplitter (FF506-DI03-25X36,
Semrock). Two-color epifluorescence is achieved using a The system performance was characterized by measuring the
two-color filter-set (emission, dichroic, and excitation filters lateralspatialresolution,thespectralthroughputofillumination
inXF454,Horiba,UK)beforebeingdirectedtoaninfinity-cor- and fluorescence collection, chromatic focus shift, verification
rectedmicroscopeobjectivewithNA0.3andworkingdistance of the camera read noise, and performance at high-frame-rate
10 mm(RMS10x-PF, Thorlabs).Emission and excitation filter acquisition(e.g.,>30 fps).Theaimwastoachievetheoptimal
spectraareshowninFigs.2(b)and2(c),respectively,alongwith contrastontargetsranginginsizefrombacteria(0.5to2 μm)to
normalizedabsorptionandemissionspectraofexamplefluores- cells,suchasmonocytesandneutrophils(10-to15 μm),andto
centprobesusedinthisstudy(PKH67andCellvue®Claret;see fungi (100 μm or more). The lowest detectable light intensity
text below). defines the visibility of the object against the noise floor,
Fluorescenceexitingthemicroscopeobjectiveisimagedonto while chromatic aberrations potentially affect contrast levels
the color CMOS camera (GS3-U3-23S6C-C, Grasshopper3, for each spectral range detected. The required frame rate for
Point Grey Research, Canada). A color imaging camera used optimal assessment in fluorescence microendoscopy is sug-
inconjunctionwithanappropriatemultibandfiltersetwaschosen gested as 10 to 15 fps.37 While this is easily achievable with
tosimplifytheopticaldesignforapplicationinaclinicalenviron- CMOS camera technologies in terms of speed and noise,38 it
ment.Thecamerahasauniversalserialbus3(USB3)connection is imperative to assess how such frame rates compromise the
tothecomputerallowingupto162framespersecond(fps)atfull limit of detection. High frame rates could be applied to tissue
resolution, 1900×1200. The color filter is the standard Bayer elastography and assessing alveolar compliance. In this
patternfilter[seeFig.2(c)]deployedinmanyconsumer,machine study, frame rates of 12–20 fps and camera resolution
vision,andmicroscopycameras.TheCMOScamerasensorisa 960×600 pixels were chosen to readily permit real-time
SonyIMX174whichhassevenphotoelectrons(e−)readnoise.36 (videorate)visualizationbyoperators.Commercialmicroendo-
We refer to the green and red channel as the green (510 to scopy systems operate at this rate.20 Higher camera resolution
560 nm) and red fluorescence (660 to 700 nm) as captured (1920×1200 pixels) was used whenever fiber core visibility
Fig.2 Alveoflex™fiberbundleasacquiredbythecamera(a)withazoomedinsettotherightshowing
fibercores(bright)andcladding(darkregionsbetweenthecores).Imagein(a)wasacquiredatfullres-
olution(1920×1200)inmonochromaticmodetoimprovesamplingofthecores.LEDsourcesspectra
andemissionfiltersabsorptioncurvesareshownin(b),whilein(c)thefluorescenceemissioncurvesare
overlaidwithemissionfilterandBayerfilters.
JournalofBiomedicalOptics 046009-3 April2016 (cid:129) Vol.21(4)
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/28/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
Krstajic´etal.:Two-colorwidefieldfluorescencemicroendoscopyenablesmultiplexed...
was needed. All tissue imaging was done at 960×600, unless asthesystemgain41toderivethereadnoiseine−.Wealsoused
stated otherwise. Higher frame rates up to 200 fps were also themonochromeversion ofthesensorforreadnoisemeasure-
evaluated, which is the upper limit imposed by the hardware, ments (part number GS3-U3-23S6M-C, Point Grey Research,
notably the camera, memory transfer speeds, and hard disk Canada). Dark noise was measured by taking a 3 s exposure
space available on the computer. time image in dark room conditions as above. The average
Optimal spatial resolution was measured by contacting the bias frame was subtracted from the 3 s exposure frame, and
distal end of the Alveoflex™ fiber to a standard chrome on the standard deviation of the resulting image was divided by
glass USAF 1951 target (Thorlabs, part number R1DS1N). 3 to obtain the dark noise for a 1 s exposure.
This measurement was done in transmission (through the Forallexperiments,abackgroundimagewasacquiredprior
USAFtarget)byallowingthefluorescencefromatarget(fluo- to imaging sessions and subtracted from the acquired images.
rescent block, Chroma) in the spectral range 510 to 560 nm Theresultingimagewascontrastenhancedtoallowforoptimal
(Thorlabs, MF510-42) to pass through the test target having illustration of image SNR. No frame averaging or image
been illuminated via the excitation delivered through the processingwasperformed,althoughthepotentialimprovement
Alveoflex™fiberandthetargetimagedbackthroughtheoptical of image processing is discussed further below.
systemintheconventionalmanner.Anintensitylineprofilewas
taken over the image obtained from the test target to estimate 2.3 Microsphere and Tissue Experiments
contrast reduction with increasing resolution.
Spectral throughput measurements were undertaken by Currentmethodstodeterminethebacterialburdeninsuspected
inserting a fiber probe connected to the spectrometer (200 μm ventilator-associated pneumonia (VAP) rely on the culture of
BALF. The growth of colony forming units (CFU) above
fiber connected to a USB2000 spectrometer, Ocean Optics) on
the following locations: (1) distal side of Alveoflex™ (in front 104 CFU∕ml42,43 or 105 CFU∕ml44,45 is considered diagnostic
of the light source), (2) proximal side of Alveoflex™, and of VAP. These are the accepted clinically relevant limits, and
(3) the camera imaging plane. The distal end of Alveoflex™ we have based our system on being able to perform at this
limitofdetection.Asbroncholavolarlavageinvolvestheinstil-
was illuminated by a white LED lamp. Fluorescence collection
lation and retrieval of fluid, a dilution effect is seen. The true
efficiency was measured by illuminating the distal end of the
bacterial burden in the infected distal lung is likely to be 105
fiberbyawhiteLEDmeasuringtheopticalpowerat(1)theproxi-
mal side of Alveoflex™ and (2) the camera imaging plane. to 106 CFU∕ml.9 Therefore, to establish the suitability of the
instrumentforthelowerlimitofbacterialdetectioninthedistal
Thelimitofdetection(LOD)forfluorescein(Sigma-Aldrich)
lung, we initially performed experiments in two scenarios
and Cy5 (Sigma-Aldrich) solutions was determined. 1, 10, 50,
wherebyweusedmicrospheresassurrogatesforbacteria:(a)im-
and 100 nM water solutions were prepared. The Alveoflex™
agingofmicrospheresaloneinblackEppendorftubes(toavoid
was inserted into the solution, and 10 images were taken for
tube fluorescence) and (b) imaging of microspheres in ex vivo
eachconcentrationwiththeexposuretimesetto80ms.Theback-
lung tissue placed in a well plate. 0.3% Inspeck™ 2.5 μm
groundimagewas obtained byilluminatingthe Alveoflex™by
microspheres (green Inspeck™ I-7219 and red Inspeck™ I-
bothLEDsasinnormaloperation,butwithoutanyfluorescence
7224)wereusedasafluorescenceemissionstandardindicative
or stray light entering the tip. Fifty background images were of labeled bacteria. Green 0.3% Inspeck™ microspheres have
savedandtheirmeansubtractedfromimagesacquiredfromsol-
equivalentemissiontogreenOMIsmartprobe7labeledbacteria
utions.A40×40 pixelareawasselectedfromthecentralareaof (Pseudomonas aeruginosa), and red 0.3% Inspeck™ micro-
the Alveoflex™ imaging fiber bundle. Mean and standard spheres are 1.5 times weaker than PKH red (Cellvue® Claret,
deviationwerecalculatedforthearea,forasingle80msexposure
Sigma-Aldrich) labeled bacteria (Pseudomonas aeruginosa).
time image indicating signal-to-noise (SNR).
Blackened microtubes were filled with concentrations
Chromaticeffectsareofparticularimportanceformulticolor 103∕ml, 104∕ml, 105∕ml, and 106∕ml to derive the detection
fluorescence microendoscopy. We assess the chromatic effects
limit of the system without the lung tissue present. Tubes
byloadingthedistalendoftheAlveoflex™with0.3%relative
were centrifuged for 10 s before each experiment. Excised
intensity standard fluorescence microspheres (green Inspeck™ lung tissue was placed in well plates with 0.3% Inspeck™
I-7219 and red Inspeck™ I-7224, Life Technologies). 2.5 μmmicrospheresdepositedatthefollowingconcentrations:
Fluorescencemicrospheresinsolutionreadilyattachtothedistal 103∕ml, 104∕ml, 105∕ml, and 107∕ml. Controls included
tip,andthisprovidesagoodtargetforevaluationofcontrastand excised lung tissue alone, 106 empty microspheres (unstained
contrastdegradation.Tooptimizethefocusoffluorescenceonto control beads supplied with green Inspeck™ I-7219, Life
thecamera,weadjustthepositionofthetubelenswithrespect Technologies) per milliliter in lung tissue, distilled water
to the CMOS sensor camera and the position of the proximal alone and lastly, 106 empty microspheres per milliliter in dis-
fiber end with respect to the microscope objective. The tube tilled water. The controls were used to minimize false positive
lenswasprealignedasinRef.26andkeptfixed.Theproximal detections,especiallyinthegreenchannel,wheresmallpartsof
end ofthe Alveoflex™was movedto threelocations. Optimal tissue could be mistaken for bacteria.
focusforgreenfluorescencewasfoundandmovedþ∕−5 μm Eachsamplewasimagedfor30s.Thefiberwasmovedevery
around this location. 2to3stoimageasmuchofthevolumeaspossibleinthelimited
Camerareadnoisewasmeasuredbytaking100biasframes observationtime.Detectionofweaklyfluorescentmicrospheres
(0 ms exposure time) in dark room conditions with a metal C- wasassessedvisuallythroughuserinterpretationofthereal-time
mountcaponthesensor.39Cameragainwassetto0dBandthe imagesacquiredduringthe30sexperiment.Asuccessfuldetec-
cameravideo mode to 7 (lowest noise mode for this particular tionwasdefinedasvisualidentificationofatleastasingletarget
camera40). The mean of all bias frames was subtracted from a with pattern recognition similar to those in the controls. The
samplebiasframeandthestandarddeviationprovidestheread camera settings were set as follows: 8 bit image acquisition,
noiseinanalog-to-digitalunits(ADU).Weused0.51 e−∕ADU gain 24 dB, exposure time 80 ms (12 fps).
JournalofBiomedicalOptics 046009-4 April2016 (cid:129) Vol.21(4)
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/28/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
Krstajic´etal.:Two-colorwidefieldfluorescencemicroendoscopyenablesmultiplexed...
2.4 Labeled Targets, Tissue, and Bacterial Imaging from the setup in Fig. 1 and 10 μL solution was placed on
Experiments theslideandcoveredbyacoverslip—and(2)withfiberbundle,
thatis,conventionalmicroendoscopyimaging.Constellation™
Excised human lung tissuewas placed in well plates with five
microspheres have a range of sizes and colors and are a good
experimentalconditions:(1)freshlyisolatedhumanmonocytes
candidate for cross comparison of imaging effects on small
labeled with PKH red (Cellvue® Claret, Sigma-Aldrich); (less than fiber core size ∼3 μm) and large objects (greater
(2) freshly isolated human neutrophils labeled with PKH than several fiber core sizes ∼10 μm). Fiber bundle imaging
green (PKH67, Sigma-Aldrich); (3) Aspergillus fumigatus ofConstellation™microsphereswasperformedinamicrotube,
labeled with PKH red; (4) labeled monocytes, neutrophils, ascontactwiththeglassslidecouldpotentiallydamagethefiber
and fungus (Aspergillus fumigatus) together in single well; bundle tip.
and (5) bacteria (Pseudomonas aeruginosa) at 107 CFU∕ml
labeled in PKH green (PKH67, Sigma-Aldrich). 3 Results
Neutrophils and monocytes were isolated by discontinuous
percollgradientsfromtheperipheralbloodofhumanvolunteers 3.1 Optical System Characterization
using methods previously described.46 As mentioned above,
Below, we present the measurement results for system spatial
bacteria used were Pseudomonas aeruginosa (ATCC 47085),
resolution, camera read noise, fluorescence collection, fluores-
and fungi used were Aspergillus fumigatus (clinical isolate,
cencespectralthroughput,chromaticfocusshift,andanevalu-
Royal Infirmary of Edinburgh, Scotland). Bacteria or fungi
ation of high-speed performance on ex vivo lung tissue.
were counterstained with PHK67 or Cellvue Claret dyes
Figure3(a)showstheUSAFtesttargetwithanevenfluores-
(Sigma-Aldrich, St Louis, MO) in accordance to the manufac-
cent illumination, and Fig. 3(b) shows the line profile across
turer’s instructions. Briefly, cells were washed three times in
group6elementswherethelinethicknessofthesmallestresolv-
PBS and re-suspended in 500 μL of Diluent C, then added to able line (group 6, element 6) is 4.38 μm. Individual test chart
500 μL of Diluent C containing 2 μL of stock dye.
lines are clearly distinguished at this resolution. Fluorescence
Following3min,bovineserumalbuminwasaddedtoterminate
ranging from 510 to 570 nm is imaged and focused in Fig. 3.
thereaction,andcellswerewashedafurtherthreetimesandre- Fiber core size, cladding size, and core to core coupling47
suspendedasrequiredforexperiments.Humantissuewasused limit the system’s spatial resolution and hence image quality;
with regional ethics committee (REC: 13/ES/0126) approval
see Fig. 2(a) for a demonstration of fiber core packing density.
and was retrieved from the periphery of specimens taken Read noise was measured to be 8 e− and dark noise 1 e−∕s.
from lung cancer resections. Monocytes and neutrophils were As the frame rates used are at least 12 fps, the influence of
isolatedandpurifiedfromperipheralbloodofhealthyvolunteers dark noise is minimal. The signal is read noise dominated for
(REC:08/S1103/38).Eachsamplewasimagedfor30s,andthe the detection of up to about 400 e−, where read noise is 35%
camera settings were as follows: 8 bit image acquisition, gain of the Poisson noise. The uniformity of the field of view
24 dB, exposure time 80 ms (12 fps). PKH red and green (FOV) is as follows: for the green channel, edge intensity is
were used as per manufacturer’s instructions. 54%ofthecentralintensity,andfortheredchannel,edgeinten-
Wellplate5withPKH greenlabeled Pseudomonasaerugi- sity is 76% of central intensity. While futurework will need to
nosaat107 CFU∕mlwasadditionallyimagedat5msexposure address the uniformity of the green channel, in the current sys-
time (200 fps) to study the dynamics of fiber core blinking temsthisdidnotseriouslyaffecttheimagingquality.Theoptical
events which are observed when imaging objects smaller power exiting the distal fiber and fluorescence collection effi-
than the fiber core. Furthermore, to study the effect further, ciency are summarized in Table 1. The maximum optical
we dissolved 5 μL of Constellation™ microspheres (C-14837, power coupling to Alveoflex™ for 470 nm LED is 2.4 mW,
Life Technologies) in 500 μL of distilled water and imaged butwelimitthisto1.8mWduetotheintrinsicredautofluores-
the solution in (1) free space—fiber bundle was removed cenceoftheAlveoflex™;seefiberbackgrounddiscussionbelow.
Fig.3 (a)ImageofaUSAF1951testtargetilluminatedwithanevenfluorescentsignaland(b)theline
profileacrossgroup6elements.Experimentperformedat510to560nmspectralrange,8bitimage,
80msexposuretime(12fps),gain15dB,1920×1200cameraresolution.
JournalofBiomedicalOptics 046009-5 April2016 (cid:129) Vol.21(4)
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/28/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
Krstajic´etal.:Two-colorwidefieldfluorescencemicroendoscopyenablesmultiplexed...
Table1 Illuminationandfluorescencecollectionefficiency. Table2 SNRforFITCandCy5solutions.
Measurement Result Experiment SNR
470nmLEDopticalpowerexiting 1.8mW 1nMfluorescein 1.5∶1
thedistalendofAlveoflex™
10nMfluorescein 2.3∶1
625nmLEDopticalpowerexiting 3.4mW
thedistalendofAlveoflex™ 50nMfluorescein 2.9∶1
100nMfluorescein 3.4∶1
Fluorescencecollection 53.1%
efficiencyongreenchannel 1nMCy5 1.4∶1
Fluorescencecollection 52.8% 10nMCy5 1.3∶1
efficiencyonredchannel
50nMCy5 1.6∶1
For625nmLED,theopticalpowermeasuredatthedistalendof 100nMCy5 2.5∶1
thefiberis3.4mW.Totalopticalpoweriswithinacceptablelimits
of maximum exposure as defined in Ref. 48. For fluorescence
collection, we measured 53% efficiency for both channels. green microspheres on the green channel. It indicates that the
SpectralthroughputisanalyzedinFig.4.Fiberfluorescence 0 μmpositionistheoptimalfocusforgreenmicrospheres,judg-
throughputisabove85%acrossthespectralrangestudied,500 ingbycontrast.Figure5(c)showsthelineprofileacrossthered
to750nm.Systemfluorescencethroughputwasmeasuredatthe microspheres on the red channel. It demonstrates that 5 μm is
camera image plane and includes attenuation by thefiberbun- theoptimalfocusfortheredmicrospheres.Thisaxialdisplace-
mentfortheredchanneloverthegreenchannelislikelydueto
dle, objective, dichroic, emission filter, and the mirror (see
chromatic aberration. As indicated previously,49 some imaging
Fig. 1). It reflects what would be expected from the dichroic
fiber bundles have detectable fiber background autofluores-
and emission filters used. Figure 4 shows Bayer filter spectra
cence. The modulation on Fig. 5(c) for the 5 μm position is
superimposed to illustrate expected reduction of the signal
duetoredfluorescencefromourfibercoresandprovidesfurther
onto the camera.
evidencethatthisistheoptimalfocusfortheredchannel.The
SNR analysis of the uniform solutions of fluorescein and
compromise is to place the proximal end of the Alveoflex™
Cy5aregiveninTable2.FluoresceinhasaLODconcentration
between the two optimal focal locations, minimizing the chro-
of 10 nM, andCy5 hasa LODconcentration of50 nM. Other
matic effect with our 0.3 NA microscope objective. The pres-
solutions,suchasinRef.21,obtainlowerdetectionlimits,pre-
ence of chromatic aberration is well understood and is, in
dominantly due to deploying several EMCCD cameras with
part, a consequence of a single channel detection platform.
optimalspectralrangesoneachcamera,whereaswehaveasin-
However, by placing the proximal end of the Alveoflex™
glecolorcamerawithahigherinherentreadnoise.Weverified
betweenthetwooptimalfocallocations,weminimizethechro-
this by using a scientific CMOS camera (PCO.edge mono
matic effect with our 0.3 NA microscope objective,and image
4.2LT, PCO AG, Germany) and LOD improved to 1 nM for
reconstructionisnotcompromisedsignificantlyforeitherofthe
fluorescein and 10 nM for Cy5.
imaging channels.
Figure5(a)showsthecontrastdegradationwithdefocusfor
Figure6demonstratesfastimageacquisitionofexvivolung
greenandredfluorescence.Defocusisobtainedbymovingthe autofluorescence.While3%Inspeck™microspheresarevisible
proximalfiberside.Figure5(b)showsthelineprofileacrossthe
atboth20and200fps(seearrowsinFig.6),theweakerfluo-
rescing 0.3% microspheres are barely visible at both 20 and
200fps.Despitethefactthatfasterframeratesresultinscaling
downthefluorescence capturedon thesensor,the autofluores-
cenceandtissuestructurearestillvisibleat200fps.Fasterframe
ratesrequirefasterdisplayhardware(displayrateonmonitorsis
currentlylimitedto60to120Hz).Oneinterestingobservation
fromthe200fpsvideosequence(seeaccompanyingVideo1)is
the possibility to extend microendoscopic imaging to tissue
elastography.50 Also, contrast often increases with increasing
frameratedespitelowerfluorescencesignal.Thisisduetobetter
localization of the object. Slower frame rates can “smear” the
contrast ofanobjectunderstudyduetothefast movingendo-
scopic environment.
3.2 Microsphere and Tissue Experiments
For both red and green 0.3% Inspeck™ beads, the LOD was
determined to be 104 microspheres per ml. Tables 3 and 4
Fig.4 Spectralthroughputanalysis.Althoughspectralattenuationis
notsignificanttothecameraplane,theBayerfilterdoesattenuatethe showthatLODsforgreenandred0.3%Inspeck™microspheres
redchannel. embedded in the excised human lung tissue are equivalent to
JournalofBiomedicalOptics 046009-6 April2016 (cid:129) Vol.21(4)
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/28/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
Krstajic´etal.:Two-colorwidefieldfluorescencemicroendoscopyenablesmultiplexed...
Fig.5 Reductionincontrastforindividualredandgreenmicrospheresisshown.(a)Visualappearance
ofdotsatdifferentfocusshifts(proximalfiberside).(b)Thelineprofileofthegreenchannelacrossthe
greenmicrosphere.(c)Thelineprofileoftheredchannelacrosstheredmicrosphere.Exposuretimewas
setto40ms(25fps),gain24dB,8bitimage,960×600cameraresolution.
105∕mland104∕ml,respectively.Thecontrolsshownodetected However,oncethedistallungisreached,itwouldbedesirableto
microspheres in both cases. The higher LOD for green micro- removethebackgroundautofluoresenceiftheobjectwearetry-
spheresisduetohighleveloflungtissueautofluorescenceinthe ingtoimagehasbroadlyequivalentexcitationwavelengths.In
green.Autofluorescenceofthehumanlungelastinandcollagen fact, this is the major reason to resort to designing molecular
enablesanatomicalnavigationbetweentheairwaysandalveoli. probesinredandNIRinthelungcoupledwithmicroendoscopy
Fig.6 (a,c)20fpsand200fpsimagingwithweaklyfluorescingmicropsheres,0.3%Inspeck™and(b,d)
morefluorescingmicrospheres,3%Inspeck™,inexvivolungtissue.Lungautofluorescencecanbeimaged
at200fpsandpossiblymore.SeetheaccompanyingVideo1illustratingvisibilityoflungtissueandred3%
Inspeck™microspheres(d)at200fps.Cameragainandresolutionweresetto24dBand960×600,respec-
tively(Video1,MOV,10.5MB[URL:http://dx.doi.org/10.1117/1.JBO.21.4.046009.1]).
JournalofBiomedicalOptics 046009-7 April2016 (cid:129) Vol.21(4)
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/28/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
Krstajic´etal.:Two-colorwidefieldfluorescencemicroendoscopyenablesmultiplexed...
Table 3 Well plate experiment for green 0.3% Inspeck™ Figure 7(a) shows PKH redlabeled monocytes inex vivolung
microspheres. tissue. Figure 7(b) shows PKH green labeled neutrophils in ex
vivo lung tissue. Figure 7(c) shows the strong signal from
Aspergillusfumigatus.Duetoitslargesize,Aspergillusfumiga-
Numberof Numberof
tusisarelativelyeasytargettoimage.Simultaneousdetectionof
duplicatewell successful
monocytes, neutrophils, and Aspergillus fumigatus is demon-
Experiment plates detections
stratedintheaccompanyingVideo2.Astissueautofluorescence
103 green0.3%Inspeck™ 4 0 has a strong green component, MI of targets in green is more
microspherespermlinlung difficult. However, for cellular targets 15 μm in size or more,
thisisentirelyfeasibleprovidingsignaltobackgroundisgreater
104 green0.3%Inspeck™ 4 0
than 10%.51
microspherespermlinlung
Smallerobjectssuchasbacteriaarealsovisibleathighsignal
105 green0.3%Inspeck™ 4 2 tobackground;howeverthefastmovingmicroendoscopyenvi-
microspherespermlinlung ronmentrequiresalargenumberofdotsforthemtobeimmedi-
atelyobservablebyaclinician.Figure8showstwoimagesofex
106 green0.3%Inspeck™ 4 4 vivolung,onewithoutbacteria,Fig.8(a),andonewithbacteria,
microspherespermlinlung Fig. 8(b). Bacteria arevisible as blinking fluorescing dots; see
Videos3and4.ThenumberofblinkingdotsislessinVideo3
Lungalone 2 0 (no bacteria) when compared to Video 4 (with bacteria).
Blinking can be explained as follows: If a fluorescing object
Distilledwater 2 0
is smaller than the fiber core and cladding, then the inherent
106 emptymicrospheres 2 0 relativemotionofthedistalendofthefiberbundlewithrespect
permlinlung tothetissuewillinducetheobjecttomoveinfrontofthecore
(visible)andcladding(notvisible).Inordertoexplorethephe-
106 emptymicrospheres 2 0 nomenonofblinkingofdynamictargetsoverthefibercores,we
permlindistilledwater
also evaluated frame rates up to 200 fps in imaging ex vivo
human lung tissuewith PKH green labeled Pseudomonas aer-
Table4 Wellplateexperimentforred0.3%Inspeck™microspheres. uginosa.Figure9demonstratesthismechanisminthreesequen-
tial images, each 5 ms exposure time, from Video 5. A core
lightsupintheleftimageofFig.9,thengoes:off:inthemiddle
Numberof Numberof image, and then lights up the neighboring core in the right
duplicate successful image. Figure 10 illustrates the effect further. Constellation™
Experiment wellplates detections
microsphereswereutilizedtodemonstratethatinfreespaceim-
103 red0.3%Inspeck™ 4 1 agingtheblinkingeffectisabsentinbothlargeandsmallfluo-
microspherespermlinlung rescing objects. However, when imaged through the fiber
bundle, blinking dominates the imaging of smaller objects.
104 red0.3%Inspeck™ 4 4 Objects larger than the core and cladding do not blink. Top
microspherespermlinlung and bottom figures in Fig. 10 are not on the same FOV.
Videos 6 and 7 demonstrate the effect clearly.
105 red0.3%Inspeck™ 4 4
This mechanism can be put to good use providing the bac-
microspherespermlinlung
terial colonies are small enough and bright enough. One can
106 red0.3%Inspeck™ 4 4 even think of engineering fiber cores and cladding (or grids
microspherespermlinlung for ex vivo study of assays) to match the size of the object
thatneedsdetecting,thusmaximizingblinking.Thekeyattrac-
Lungalone 2 0 tionisthat,toourexperience,blinkingisreadilydetectableby
observersatSNRsevenlessthan1.Onealsohastobecautious;
Distilledwater 2 0 anyobjectsmallerthanthefibercladdingwilleventuallyblink,
whether bacteria or not, so false positivesare likely. The same
106 emptymicrospheres 2 0
appliestoscenarioswhereultrafastimagingisperformed,which
permlinlung
will accentuate blinking due to motion effects and low SNR
106 emptymicrospheres 2 0 (blinking due to shot noise). Additionally, many bacterial col-
permlindistilledwater onies are large in size (>20 μm), and these will not blink.
Despite limitations, we believe the technique could be useful
if appropriately integrated into MI algorithms involving
asthebackgroundfluorescencedropsdramaticallywithhigher microendoscopy.
wavelengths.
4 Discussion
3.3 Labeled Targets, Tissue Experiments, and
Akeystrategyforfutureworkistouseautofluorescenceforstruc-
Observations on Bacterial Blinking
turalimagingintheinvivohumanlunganddeploysmartprobes
Figure7illustratesthecapabilityofthetwo-colorwidefieldsys- toevaluatedetectionofspecificpathology.7,8Thelimithereisthat
tem to perform fluorescence imaging of multiple targets of greenautofluorescenceinterfereswithgreenfluorescingsmartp-
immediaterelevancetopulmonaryinflammationandinfection. robes,creating challengestodisentanglethetwosignalsinvivo
JournalofBiomedicalOptics 046009-8 April2016 (cid:129) Vol.21(4)
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/28/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx