accuracy and precision in quantitative fluorescence

8/8/2019 Accuracy and Precision in Quantitative Fluorescence

http://slidepdf.com/reader/full/accuracy-and-precision-in-quantitative-fluorescence 1/14



JCB: FEATURE

1141QUANTITATIVE MICROSCOPY • Waters

Read noise is usually expressed in themanu acturer’s technical speci cationsas a number o electrons, meaning thatthe measured voltage will have a vari-ance equal to that number o electrons(i.e., the lower the value, the lower thenoise). Detectors that use signal ampli i-cation (e.g., PMTs and electron multi-plying [EM] CCDs) introduce additionalnoise during the ampli ication process.For example, EM-CCD cameras ampli ysignal di erences su ciently to revealclock-induced charging—stochastic vari-ations in the trans er o charge rom onepixel to another during read operations(Robbins and Hadwen, 2003; Moomaw,2007). When possible, collecting morephotons rom the specimen to increasethe signal (see Table I) will result in ahigher SNR image than ampli ying a

smaller number o collected photons.The various sources o noise add as thesum o the squares:

N N N N total Poisson read thermal = + + +2 2 2 ....

The resulting total noise in the digitalimage de nes a minimum expected vari-ance in the measured intensity values.Di erences in measurements that liewithin the expected variance due to noise

cannot be attributed to the specimen.Pawley (1994) provides a thorough re-view o the di erent sources o noise indigital microscopy images.

Noise is not a constant, so it cannotbe subtracted rom a digital image. How-ever, i multiple images o the same eldo view are collected and averaged to-gether (“ rame averaging”), the noisewill average out and the resulting meanintensity values will be closer to the“real” intensity values o the signal plusthe background (Cardullo and Hinch-cli e, 2007). Frame averaging is veryuse ul when imaging xed specimenswith a higher noise instrument like apoint-scanning con ocal, but is usuallyimpractical or quantitative imaging o live fuorescent specimens that are dy-namic and susceptible to phototoxicityand photobleaching. For quantitative fu-orescence imaging, noise added to thedigital image during acquisition shouldbe reduced as much as possible through

choice o detector and acquisition set-tings (Table I).

Resolution

In digital microscopy, spatial resolutionis de ned by both the microscope andthe detector, and limits our ability to ac-curately and precisely locate an objectand distinguish close objects as separate

rom one another (Inoue, 2006; Rasnik et al., 2007; Waters and Swedlow, 2007).Objects that cannot be detected in animage cannot be resolved, so spatialresolution is dependent on image SNR(Pawley, 2006c). When imaging a dy-namic specimen over time, accuracyo quantitation may be urther limitedby temporal resolution (the rate o image acquisition; see Dorn et a l., 2005;Jares-Erijman and Jovin, 2006 or

an example).Resolution in the optical

image. Lateral resolution o the opticalimage is de ned as the distance bywhich two objects must be separated inorder to distinguish them as separate

rom one another, which is equal to theradius o the smallest point source in theimage (de ned as the rst minimum o the airy disk; Inoue, 2006). Lateral reso-lution ( r ) in epifuorescence microscopyis given by

r NA=( . )0 61 λ

,

where is the wavelength o emissionlight and NA is the numerical aperture o the objective lens. Numerical aperture isusually marked on the objective lens bar-rel, just a ter the magni cation (Keller,2006). Resolution in the z-axis ( z) isworse than lateral resolution in the lightmicroscope, and is given by

z NA

=

22

λη,

where is the re ractive index o thespecimen. It is important to understandthat these equations give the theoreticalresolution limits o a per ect lens used toimage an ideal specimen; real lenses andspecimens o ten introduce aberrations inthe image that reduce resolution (Egnerand Hell, 2006; Goodwin, 2007). Thebest way to know the resolution limit o your imaging system is to measure it em-pirically (Hiraoka et al., 1990). These

equations de ne the theoretical resolu-tion limits in most cases; it should benoted that a hand ul o very talentedmicroscopists have ound it possible tosurpass these limits using specialized“super-resolution” imaging techniques ( orreview see Evanko, 2009).

It is a common misconceptionamong cell biologists that con ocal mi-croscopy should be used to obtain thehighest resolution images. Althoughcon ocal microscopy is very e ective atincreasing contrast in specimens with sig-ni cant out-o - ocus fuorescence (Pawley,2006b; Murray et al., 2007), the obtain-able resolution o con ocal microscopyis essentially the same as conventionalwide- eld fuorescence microscopy(Inoue, 2006). In addition, Murray et al.(2007) recently demonstrated that the

photon collection e ciency and SNR o wide- eld fuorescence is generally higherthan con ocal microscopy or specimenswith limited out-o - ocus fuorescence.Con ocal microscopy becomes necessaryand avorable or specimens with high lev-els o out-o - ocus fuorescence becauseout-o - ocus fuorescence adds to back-ground fuorescence, and there ore de-creases the capacity to collect the signal o interest (Fig. 1; Murray et al., 2007).

Resolution in the digital

image. The optical image is sampled by a

detector to create a digital image. The reso-lution o a digital image acquired with aCCD camera depends on the physical sizeo the photodiodes that make up the chip(Rasnik et al., 2007), whereas in point-scanning con ocal resolution is determinedby the area o the specimen that is scannedper pixel (Pawley, 2006c). The pixel sizeshould be at least two times smaller thanthe resolution limit o the microscope op-tics, so that the smallest possible object inthe image (de ned as the diameter o theairy disk) will be sampled by 4 pixels(Fig. 3; Pawley, 2006c). There is a trade-o between resolution o the digital imageand signal intensity because magni cationdecreases image intensity (Waters, 2007)and smaller pixels generally collect ewerphotons (Fig. 3). To compensate or loss o signal due to smaller pixel size, longercamera exposure times or more intense il-lumination may be necessary (Fig. 3 A). I the pixel size is too large, the optical imagewill be under-sampled and detail will be

Published June 29, 2009



JCB • VOLUME 185 • NUMBER 7 • 20091142

lost in the digital image (Fig. 3 C). In live-cell imaging, it is o ten avorable to give upsome resolution (by binning pixels, orexample) to increase image SNR and/ordecrease photobleaching and photo dam-age (Waters, 2007).

Optical resolution need not

limit accuracy in localization or

counting. How does the resolutionlimit a ect our ability to quantitate

using fuorescence microscopy? Clearly,the size o an object that is below theresolution limit cannot be accuratelymeasured with the light microscope.However, objects that are below the res-olution can be detected and an image o the object ormed by the microscope, i the imaging system is sensitive enoughand the object is bright enough. Al-though the size o the object in the im-age will be inaccurate, the centroid o ahigh SNR image o the object can beused to locate the object with nanome-ter precision, ar beyond the resolutionlimit (Churchman et al., 2005; Yildizand Selvin, 2005; Huang et al., 2008;Manley et al., 2008).

In fuorescence microscopy, the res-olution limit does not limit our ability toaccurately count fuorescently labeled ob-

jects, even i the objects are below theresolution limit. I the objects are all o similar size, are all labeled with the samenumber o fuorophores, and the intensity

o one object can be accurately deter-mined; then intensity values can be usedto count multiple objects that are too closeto one another to spatially resolve. Thesetypes o measurements are very challeng-ing to per orm with accuracy, and requirea thorough understanding o , and atten-tion to, every possible pit all (Pawley,2000)—but they are possible. For exam-ple, the measured intensity o proteins

conjugated to fuorescent proteins hasbeen used to accurately and preciselycount the number o labeled proteins lo-calized to the kinetochore (Joglekar et al.,2006) and proteins involved in cytokine-sis (Wu and Pollard, 2005).

Additional threats to

accuracy and precision in

quantitative microscopy

Specimen preparation. Fluorescencerom a fuorophore tagged to a molecule

o interest is o ten used to measure thequantity o the tagged molecule. Fluor-escent proteins expressed in live cellsare excellent or quantitation; becausethere is a constant number o fuoro-phores per labeled protein, the numbero photons emitted can be an accuratemeasure o the quantity o fuorescentlylabeled protein (Shaner et al., 2005;Straight, 2007; Joglekar et al., 2008).Small molecules that bind with higha nity to their target, such as calcium

indicator dyes, are also reliable orquantitation (Johnson, 2006). Quantita-tion o live versus xed cells is gener-ally pre erable because the xation andextraction process can remove taggedproteins, quench the luorescence o

luorescent proteins, and change thesize and shape o cells (Allan, 2000;Straight, 2007).

One should use caution when using

immuno luorescence to measure thelocal concentration o a protein o inter-est, particularly with soluble proteins.The xation and extraction necessary toget antibodies into cells can change thequantity and localization o detectableepitopes (Melan and Sluder, 1992). Multi-valent antibodies also bind with highera nity to multiple epitopes, which canmake areas with high concentration o the epitope label more e ciently thanareas o low concentration (Mason andWilliams, 1980). In addition, penetrationo the antibody may not be consistent indi erent areas o the tissue, cells, andsubcellular compartments (Allan, 2000).There ore, although accurate quantita-tion o the emission photons rom fuor-escently labeled antibodies is possible,that number o photons may not accu-rately refect the number o epitopes inthe specimen. It is possible to use rigor-ous controls to demonstrate that an im-munofuorescence protocol is an accurate

Figure 4. Non-uni orm illumination results in nonuni orm uorescence.All images were collected using a microscope (model TE2000E; Nikon), a Plan-Apochromat 20x 0.75 NA objective lens, a camera (ORCA-AG; Hamamatsu Photonics), and MetaMorph so tware. (A) An image o a feld o uorescenbeads, using wide-feld illumination. Individual beads contain a similar concentration o uorophore (clumps o beads appear brighter, as is seen near thecenter o the image). A pseudo-color displaying the range o intensity values (see inset) was applied. Note that beads in the top le t have di erent intensvalues than the beads in the bottom right. (B) An image o a uni orm feld o uorophore taken with the same microscope optics and conditions as A, showing uneven illumination across the feld o view. This nonuni orm illumination explains the nonuni orm uorescence rom the beads o similar uoroconcentration shown in A. (C) A ter at-feld correction (Zwier et al., 2004; Wol et al., 2007), the image intensity values more accurately re ect the real

uorescence in the specimen. This image was obtained using the image arithmetic unction in image processing so tware (in this case, MetaMorph) tdivide the image in A by the image in B. Bar = 50 µm.




JCB: FEATURE


measure o a particular epitope o inter-est (Mortensen and Larsson, 2001).

Measuring fuorescence deep intospecimens can also be problematic or

measurements o signal intensities. Inbiological specimens, light scatteringand optical aberrations increase with dis-tance rom the coverslip and decreasesignal intensity (Murray, 2005). Thesee ects are di cult to characterize andcorrect or in inhomogeneous biologicalsamples. Although the e ects o lightscattering are minimized when usingmulti-photon illumination (Rocheleauand Piston, 2003), the accuracy o quan-titation o intensity values is limited byoptical aberrations and dramatically in-

creased photobleaching in the ocal plane(Patterson and Piston, 2000).

Non-uni orm illumination. Fluorescence emission is generally pro-portional to the intensity o the illumi-nating light (except when fuorophoreground state depletion occurs; see Tsienet al., 2006; Murray et al., 2007). There-

ore, i a uni orm fuorescent sample isunevenly illuminated, the resulting fuor-escence will usually be uneven as well.Uneven illumination can be extremelydetrimental to quantitative measure-ments because it may cause the intensityo an object in one area o the eld o view to measure di erently than the in-tensity o an object o equal fuorophoreconcentration in another area o the eldo view (Fig. 4 A). To reduce uneven il-lumination, the wide- eld fuorescencemicroscope should be care ully aligned

or Koehler illumination (Salmon andCanman, 2001). Scrambling the imageo the light source be ore it enters the

microscope (using a liquid light guide,or example) can increase the uni ormity

o illumination across the eld o view(Nolte et al., 2006). In many cases, com-

pletely uni orm illumination is impossi-ble to achieve, and one must insteadcorrect or uneven illumination be oremaking quantitative measurements (Fig. 4;Table II; Zwier et al., 2004; Wol et al.,2007). To per orm the correction, an im-age o a uni orm fuorescent sample iscollected to reveal the uneven illumina-tion pattern (Fig. 4 B). The image to becorrected is then divided by the image o the uni orm fuorescent sample (usingthe image arithmetic unctions availablein most image-processing so tware) to

obtain the fat- eld corrected image(Fig. 4 C). Because the pattern o illumi-nation may di er rom day to day, it isbest to collect a new image o a uni ormfuorescent sample at each imaging ses-sion. A protocol or fat- eld correctionis described by Wol et al. (2007).

Bleed-through and auto-

fuorescence. When choosing fuores-cence lter sets, maximizing excitationand emission collection should be bal-anced with minimizing bleed-through(also called crosstalk) and autofuores-cence. Bleed-through o one fuoro-phore’s emission through the lter set o another fuorophore can occur when aspecimen is labeled with multiple fuoro-phores whose excitation and emissionspectra overlap (Fig. 5; Ploem, 1999;Bolte and Cordelieres, 2006; Rietdor and Stelzer, 2006). Many biologicalspecimens contain native autofuorescenceo similar wavelengths to the emission o many commonly used fuorophores

(Aubin, 1979; Tsien et al., 2006).For quantitative measurements, bleed-through and auto luorescence shouldbe avoided when possible, and measured

and subtracted rom measurements whenunavoidable (Rietdor and Stelzer, 2006).Avoid bleed-through and autofuores-cence by care ully choosing fuoro-phores and lter sets. Bleed-throughbetween two fuorophores can be de-tected and measured by using the fuoro-phore “A” lter set and camera acquisitionsettings to collect an image o a controlsample labeled with only fuorophore“B”. Autofuorescence can be detectedand measured by collecting images o acontrol specimen that is identical to the

experimental specimen except or theaddition o exogenous fuorophores.

Image registration. Images o di erent wavelengths emitted rom a sin-gle plane o a specimen may not be coin-cident in the optical images (Fig. 6 A).Shi ts between wavelengths in X, Y,and Z can be introduced by a variety o sources, including wedges in fuores-cence lters (Rietdor and Stelzer, 2006)and chromatic aberrations in the objec-tive lens (Keller, 2006). I registrationbetween images o di erent wave-lengths is important to your quantitativeanalysis, you should check or shi ts be-tween wavelengths in your microscopeusing submicron beads in used withmultiple luorophores (such as Tetra-speck beads, Invitrogen; protocol inWol et al., 2007). Consistent shi ts be-tween wavelengths can then be cor-rected using image-processing so tware(Fig. 6 B) or ( or axial shi ts) by using a

ocus motor to adjust ocus between

Figure 5. Bleed-through can cause inaccuracy in intensity measurements. (A and B) Images o a cell (outlined in white)labeled with DAPI (nuclei) and Bodipy-FL phalloidin (actin).Both images were collected using the same microscope (model80i; Nikon), a Plan-Apochromat 100x NA 1.4 oil objectivelens, the same camera (ORCA R2; Hamamatsu Photonics), andMetaMorph so tware. The same camera acquisition settingswere used or both images, but they were collected using twodi erent flters designed or imaging DAPI. (A) An image col-lected with a DAPI flter set containing a long-pass emissionflter, which allows bleed-through o the Bodipy-FL signal in thecytoplasm. The bleed-through o the actin in the cytoplasm isjust barely visible by eye in the image. The average intensityvalue o the cytoplasm in this image is 205. (B) An image othe same cell as in A, collected with a DAPI flter set containinga band-pass emission flter, which blocks bleed-through o theBodipy-FL signal in the cytoplasm. The average intensity valueo the cytoplasm in this image is 91, over 50% less than theimage in A. Bar = 10 µm.





wavelengths (see Murray et al., 2007 oran example).

Focus. Focus is critical to accu-rate and precise quantitation o fuores-cence intensity values (Murray, 2005;Table II). The distribution o intensityvalues along the z-axis o the opticalimage depends on the size o the fuores-cently labeled object and the point spread

unction o the microscope. For smallobjects imaged with high resolution op-tics, small changes in ocus can havelarge e ects on measured intensity val-ues. Joglekar et al. (2008) describe amethod o determining the error intro-duced by imprecise ocus when measur-ing objects that are thinner than thedi raction limit o the optics. Measuring

in 3D is almost always necessary or ac-curately determining intensity o objectslarger than the di raction limit, andwhen tracking the movements o objectsthat occur in 3D (De Mey et al., 2008).

Photobleaching. Almost allfuorophores photobleach when exposedto excitation light in the microscope,including all o the luorescent pro-teins. The rate o photobleaching isspeci c to the fuorophore, its environ-ment, and the intensity o the illuminat-ing light (Diaspro et al., 2006). Forsome specimens and fuorophores, anti-photobleaching reagents can be addedto the mounting medium to reduce therate o photobleaching (Diaspro et al.,2006; Tsien et al., 2006). Depending

on the rate o photobleaching, excitingluorescent specimens be ore collect-

ing images that will be used orquantitation may introduce error inmeasurements o signal intensity. Ini-tial ocusing and scanning o the speci-men is best done using techniques suchas phase or di erential inter erencecontrast (DIC) microscopy (Inoué andSpring, 1997) because the halogen lightsources typically used or transmittedlight illumination usually will not bleach

luorophores. To accurately measureluorescence intensity in the same eld

o view over time, one should measureand correct or photobleaching thatoccurs while imaging (Rabut andEllenberg, 2005).

Figure 6. Shi ts in image registration can a ect colocalization results.(A and B) Images o 100-nm Tetra-Speck beads (Invitrogen; mounted in glycerol) thatuoresce multiple colors including red and green, collected with a microscope (model 80i; Nikon) and camera (ORCA R2; Hamamatsu Photonics) using a

Plan-Apochromat 100x NA 1.4 oil objective lens and MetaMorph so tware. One image o the beads was collected using a flter set or green uorescence(FITC) and a second image o the beads was collected using a flter set or red uorescence (TRITC); all other microscope optics were the same betweethe two images. The two images were pseudo-colored and merged using MetaMorph so tware. The scatter plots (generated in MetaMorph) display thecorrelation between the intensity values o the red and green pixels in the images. (A) The merged image, showing a registration shi t o several pixebetween the red and green images. 10 sets o images were collected to determine that the shi t is repeatable, and there ore most likely caused by the fltersets (not depicted). The correlation coe fcient or these red and green images is only 0.72, even though the red and green images represent the exact samebeads. Bar = 1 µm. (B) The same images as in A, a ter correction or the shi t in registration using MetaMorph image processing so tware. The correlaticoe fcient increased to 0.97 a ter the correction.





intense ocused illumination is used tophotobleach fuorophores in a select re-gion o a specimen. I the fuorescentlylabeled molecules are mobile, un-

bleached molecules will move into thebleached region while the bleached mol-ecules move out. Recovery o the photo-bleached region over time can then beused to detect mobility/immobility, andto measure di usion or association/ dissociation kinetics o the fuorescentlylabeled component (Snapp et al., 2003;Rabut and Ellenberg, 2005; Sprague andMcNally, 2005).

As the specimen is illuminated tocollect images o the recovery process,fuorophores will continue to photobleach

at a rate dependent (in part) on the level o illumination. To get an accurate measureo recovery o the photobleached region,one must measure and correct or photo-bleaching that occurs during image acqui-sition. This can be done, or example, bymeasuring the fuorescence o an un-bleached region in the same or a neigh-boring cell (Rabut and Ellenberg, 2005).

The accuracy o FRAP analyses iscompromised by the act that fuoro-phores can reversibly photobleach(Diaspro et al., 2006). It is there ore im-portant to choose a fuorophore or FRAPexperiments that is less likely to enterillumination-induced dark states that canspontaneously recover to the fuorescentstate. EGFP, or example, is less likely toundergo reversible photobleaching thanYFP variants. Reversible photobleachingcan be detected by measuring the fuores-cence intensity o entire cells a ter bleach-ing because it will cause the intensity toincrease within a ew seconds. The extent

o the e ect o reversibility o photo-bleaching on FRAP measurements de-pends on the rate o image acquisition,and can be controlled by collecting a

number o pre-bleach images to reach asteady state o fuorophores in the dark state (Rabut and Ellenberg, 2005).

Presenting quantitative

microscopy measurements

Reporting error. No matter how care-ul you are when collecting images and

making measurements, every quantitativeanalysis has a level o uncertainty thatmust be reported (Table II; Cumming etal., 2007; Wol et al., 2007). Error is mostcommonly reported by stating or showing

(as error bars) the standard deviation orstandard error o the mean o the mea-sured values. The appropriate way to re-port error depends on your data and theconclusions you would like to make. This

journal recently published a thorough anduser- riendly review on reporting error inquantitative measurements (Cumming etal., 2007). When per orming arithmeticon multiple quantitative measurementsthat have independent sources o error(e.g., subtracting an average backgroundintensity value B ± b rom a measuremento signal intensity S ± s), the error in theindividual measurements should be prop-agated to the nal value. A general or-mula or error propagation was derived byWol et al. (2007).

Writing the materials and

methods. When publishing quantita-tive microscopy data, you should providethe reader with the in ormation they needto judge whether you used the appropri-ate equipment and acquisition parameters

or the experiment (Table III; Waters andSwedlow, 2007). Even with this in orma-tion, it may be di cult or the reader toassess whether your system was opti-

mized and operated to obtain the bestpossible SNR and resolution. Easy openaccess to raw image les and data used

or published quantitative analyses willthere ore be critical to the continuedgrowth o the eld o quantitative fuores-cence microscopy (Moore et al., 2008).

Further reading

This review is an introduction to the is-sues surrounding accurate and precisequantitation o fuorescence microscopydigital images. A thorough appreciation

o both the power and the limitations o quantitative microscopy is best obtainedthrough care ul attention to how these is-sues a ect your own data. The interestedreader is encouraged to learn more romthe many excellent books, reviews, andquantitative analyses that are re erencedthroughout this review.

The author wishes to thank the aculty and studeno the Analytical and Quantitative Light Microcopy course at the Marine Biological Laboratory(www.mbl.edu/education) or many illuminatindiscussions—especially Jason Swedlow, DavidWol , and John Murray; Jason Swedlow andWendy Salmon or help ul comments and discusions on this manuscript; Cassandra Rogers opreparing bead slides used to acquire images orthe fgures; and Dan Bolton or patient support during the preparation o this manuscript.

Re erencesAdams, M.C., W.C. Salmon, S.L. Gupton, C.S.

Cohan, T. Wittmann, N. Prigozhina, andC.M. Waterman-Storer. 2003. A high-speedmultispectral spinning-disk con ocal micro-scope system or fuorescent speckle micros-copy o living cells. Methods. 29:29–41.

Table III.In ormation to include in the Materials and methods or fgure legend

Follow the journal’s instructions or authors, and/or include the ollowing:

• Manu acturer and model o microscope• Objective lens magnifcation, NA, and correction or aberration (e.g., 60x 1.4 NA Plan-Apochromat)• Fluorescence flter set manu acturer and part number and/or the transmission and bandwidth (e.g., 490/30)• Illumination light source (including wavelength or laser illumination)• Camera manu acturer and model

• So tware program(s) and version• Manu acturer and model o other acquisition hardware, including con ocal, flter wheels, ocus motors, motorized stage, shutters, etc.• Image acquisition settings including exposure times, gain, and binning• Other acquisition parameters, including ocus step size ( or z-series), time between images ( or time lapse), etc.• Description o image processing routine used to create fgures• Description o segmentation and image analysis routine, and method o validation




JCB: FEATURE


Allan, V.J. 2000. Protein Localization by FluorescenceMicroscopy: A Practical Approach. Ox ordUniversity, New York. 256 pp.

Art, J. 2006. Photon detectors or con ocal micros-copy. In Handbook o Biological Con ocalMicroscopy. J.B. Pawley, editor. Springer-Verlag New York, Inc., New York. 251–262.

Aubin, J.E. 1979. Autofuorescence o viablecultured mammalian cells. J. Histochem.Cytochem. 27:36–43.

Axelrod, D., N.L. Thompson, and T.P. Burghardt.1983. Total internal infection fuorescentmicroscopy. J. Microsc. 129:19–28.

Bolte, S., and F.P. Cordelieres. 2006. A guided tourinto subcellular colocalization analysis inlight microscopy. J. Microsc. 224:213–232.

Cardullo, R.A., and E.H. Hinchcli e. 2007. Digitalmanipulation o bright eld and fuorescenceimages: noise reduction, contrast enhance-ment, and eature extraction. Methods Cell

Biol. 81:285–314.Chen, H., H.L. Puhl III, S.V. Koushik, S.S. Vogel,

and S.R. Ikeda. 2006. Measurement o FRET e ciency and ratio o donor to ac-ceptor concentration in living cells. Biophys.

J. 91:L39–L41.Churchman, L.S., Z. Okten, R.S. Rock, J.F. Dawson,

and J.A. Spudich. 2005. Single moleculehigh-resolution colocalization o Cy3 andCy5 attached to macromolecules measuresintramolecular distances through time. Proc.

Natl. Acad. Sci. USA. 102:1419–1423.Cumming, G., F. Fidler, and D.L. Vaux. 2007. Error

bars in experimental biology. J. Cell Biol. 177:7–11.

Day, R.N. 2005. Imaging protein behavior inside theliving cell. Mol. Cell. Endocrinol. 230:1–6.

De Mey, J.R., P. Kessler, J. Dompierre, F.P.Cordelières, A. Dieterlen, J.L. Vonesch, J.B.Sibarita, and F.S. Kevin. 2008. Fast 4D mi-croscopy. Methods Cell Biol. 85:83–112.

Diaspro, A., G. Chirico, C. Usai, P. Ramoino,and J. Dobrucki. 2006. Photobleaching.

In Handbook o Biological Con ocalMicroscopy. J.B. Pawley, editor. Springer-

Verlag New York, Inc., New York. 690–699.Dorn, J.F., K. Jaqaman, D.R. Rines, G.S. Jelson,

P.K. Sorger, and G. Danuser. 2005. Yeastkinetochore microtubule dynamics analyzedby high-resolution three-dimensional mi-croscopy. Biophys. J. 89:2835–2854.

Dorn, J.F., G. Danuser, G. Yang, and F.S. Kevin.2008. Computational processing and analy-sis o dynamic fuorescence image data.

Methods Cell Biol. 85:497–538.Egner, A., and S.W. Hell. 2006. Aberrations in

con ocal and multi-photon fuorescence mi-croscopy induced by re ractive index mis-match. In Handbook o Biological Con ocalMicroscopy. J.B. Pawley, editor. Springer-Verlag New York, Inc., New York. 404–412.

Evanko, D. 2009. Primer: fuorescence imaging underthe di raction limit. Nat. Methods. 6:19–20.

Giepmans, B.N., S.R. Adams, M.H. Ellisman, andR.Y. Tsien. 2006. The fuorescent toolbox

or assessing protein location and unction.Science. 312:217–224.

Goodwin, P.C. 2007. Evaluating optical aberrationusing fuorescent microspheres: methods,analysis, and corrective actions. MethodsCell Biol. 81:397–413.

Gordon, G.W., G. Berry, X.H. Liang, B. Levine, andB. Herman. 1998. Quantitative fuorescenceresonance energy trans er measurementsusing fuorescence microscopy. Biophys. J. 74:2702–2713.

Hell, S.W., and E. Stelzer. 1995. Lens aberra-tions in con ocal fuorescence microscopy.

In Handbook o Biological Con ocal Micros-copy. J.B. Pawley, editor. Springer-VerlagNew York, Inc., New York. 347–353.

Hiraoka, Y., J.W. Sedat, and D.A. Agard. 1990.Determination o three-dimensional imag-ing properties o a light microscope system.Partial con ocal behavior in epifuorescencemicroscopy. Biophys. J. 57:325–333.

Ho man, D.B., C.G. Pearson, T.J. Yen, B.J. Howell,and E.D. Salmon. 2001. Microtubule-dependent changes in assembly o micro-tubule motor proteins and mitotic spindlecheckpoint proteins at PtK1 kinetochores.

Mol. Biol. Cell. 12:1995–2009.Huang, B., W. Wang, M. Bates, and X. Zhuang.

2008. Three-dimensional super-resolutionimaging by stochastic optical reconstructionmicroscopy. Science. 319:810–813.

Inoue, S. 2006. Foundations o con ocal scannedimaging in light microscopy. In Handbook o Biological Con ocal Microscopy. J.B.Pawley, editor. Springer-Verlag New York,Inc., New York. 1–19.

Inoué, S., and K.R. Spring. 1997. Video Microscopy:The Fundamentals. Plenum Press, NewYork. 770 pp.

Jares-Erijman, E.A., and T.M. Jovin. 2003. FRETimaging. Nat. Biotechnol. 21:1387–1395.

Jares-Erijman, E.A., and T.M. Jovin. 2006. Imagingmolecular interactions in living cells byFRET microscopy. Curr. Opin. Chem. Biol. 10:409–416.

Joglekar, A.P., D.C. Bouck, J.N. Molk, K.S. Bloom,and E.D. Salmon. 2006. Molecular architec-ture o a kinetochore-microtubule attach-ment site. Nat. Cell Biol. 8:581–585.

Joglekar, A.P., E.D. Salmon, K.S. Bloom, and F.S.Kevin. 2008. Counting kinetochore proteinnumbers in budding yeast using geneticallyencoded fuorescent proteins. Methods Cell

Biol. 85:127–151.Johnson, I. 2006. Practical considerations in the

selection and application o fuorescentprobes. In Handbook o Biological Con ocalMicroscopy. J.B. Pawley, editor. Springer-Verlag New York, Inc., New York. 353–364.

Keller, H.E. 2006. Objective lenses or con ocalmicroscopy. In Handbook o BiologicalCon ocal Microscopy. J.B. Pawley, editor.Springer-Verlag New York, Inc., New York.145–160.

Manley, S., J.M. Gillette, G.H. Patterson, H. Shro ,H.F. Hess, E. Betzig, and J. Lippincott-Schwartz. 2008. High-density mappingo single-molecule trajectories with photo-activated localization microscopy. Nat.

Methods. 5:155–157.Mason, D.W., and A.F. Williams. 1980. The kinetics

o antibody binding to membrane antigensin solution and at the cell sur ace. Biochem.

J. 187:1–20.Melan, M.A., and G. Sluder. 1992. Redistribution

and di erential extraction o soluble proteins

in permeabilized cultured cells. Implicationsor immunofuorescence microscopy. J. CellSci. 101:731–743.

Moomaw, B. 2007. Camera technologies or lowlight imaging: overview and relative advan-tages. Methods Cell Biol. 81:251–283.

Moore, J., C. Allan, J. Burel, B. Loranger, D.MacDonald, J. Monk, J.R. Swedlow, andF.S. Kevin. 2008. Open tools or storageand management o quantitative image data.

Methods Cell Biol. 85:555–570.Mortensen, K., and L.I. Larsson. 2001. Quantitative

and qualitative immunofuorescence studieso neoplastic cells trans ected with a con-struct encoding p53-EGFP. J. Histochem.Cytochem. 49:1363–1367.

Murray, J.M. 1998. Evaluating the per ormanceo fuorescence microscopes. J. Microsc. 191:128–134.

Murray, J.M. 2005. Con ocal microscopy, deconvo-lution and structured illumination methods.

In Live Cell Imaging: A Laboratory Manual.R.D. Goldman and D.L. Spector, editors.Cold Spring Harbor Laboratory, Cold SpringHarbor, NY. 239–279.

Murray, J.M., P.L. Appleton, J.R. Swedlow, and

J.C. Waters. 2007. Evaluating per ormancein three-dimensional fuorescence micros-copy. J. Microsc. 228:390–405.

Nolte, A., J. Pawley, and L. Horing. 2006. Non-laserlight sources or three-dimentional micros-copy. In Handbook o Biological Con ocalMicroscopy. J.B. Pawley, editor. Springer-Verlag New York, Inc., New York. 126–144.

North, A.J. 2006. Seeing is believing? A beginners’guide to practical pit alls in image acquisition.

J. Cell Biol. 172:9–18.Patterson, G.H., and D.W. Piston. 2000. Photo-

bleaching in two-photon excitation micros-copy. Biophys. J. 78:2159–2162.

Pawley, J. 2000. The 39 steps: a cautionary tale o quantitative 3-D fuorescence microscopy.

Biotechniques. 28:884–886, 888.Pawley, J.B. 1994. Sources o noise in 3D micro-

scopical data sets. In Three-DimensionalCon ocal Microscopy: Volume Investigationo Biological Systems. J.K. Stevens, L.R.Mills, and J.E. Trogadis, editors. AcademicPress, Inc., San Diego. 48–90.

Pawley, J.B. 2006a. Fundamental limits in con o-cal microscopy. In Handbook o BiologicalCon ocal Microscopy. J.B. Pawley, editor.Springer-Verlag New York, Inc., New York.20–42.

Pawley, J.B. 2006b. Handbook o BiologicalCon ocal Microscopy. Springer-Verlag NewYork, Inc., New York. 985 pp.

Pawley, J.B. 2006c. Points, pixels, and gray lev-els: digitizing image data. In Handbook o Biological Con ocal Microscopy. J.B.Pawley, editor. Springer-Verlag New York,

Inc., New York. 59–79.Periasamy, A., and R.N. Day. 2005. Molecular

Imaging: FRET Microscopy and Spectroscopy.Ox ord University Press. 336 pp.

Ploem, J.S. 1999. Fluorescence microscopy. In Fluorescent and Luminescent Probes orBiological Activity. W.T. Mason, editor.Academic Press, London. 3–13.

Rabut, G., and J. Ellenberg. 2005. Photobleachingtechniques to study mobility and molecular dy-namics o proteins in live cells: FRAP, iFRAP,and FLIP. In Live Cell Imaging: A LaboratoryManual. R.D. Goldman and D.L. Spector,editors. Cold Spring Harbor Laboratory, ColdSpring Harbor, NY. 101–126.

Rasnik, I., T. French, K. Jacobson, and K. Berland.2007. Electronic cameras or low-light mi-croscopy. Methods Cell Biol. 81:219–249.

Rietdor , J., and E.H.K. Stelzer. 2006. Special opticalelements. In Handbook o Biological Con ocalMicroscopy. J.B. Pawley, editor. Springer-Verlag New York, Inc., New York. 43–58.

Robbins, M.S., and B.J. Hadwen. 2003. The noiseper ormance o electron multiplying charge-coupled devices. IEEE Trans. Electron. Dev. 50:1227–1232.

Rocheleau, J.V., and D.W. Piston. 2003. Two-photonexcitation microscopy or the study o livingcells and tissues. Curr. Protoc. Cell Biol. Chapter 4:Unit 4.11.

Russ, J.C. 2007. The Image Processing Handbook.CRC/Taylor and Francis, Boca Raton, FL.647 pp.


accuracy and precision in quantitative fluorescence

Documents