Technology and Engineering

Dynamic Real-timeMicroscopy of the Urinary TractUsing Confocal Laser EndomicroscopyKatherine Wu, Jen-Jane Liu, Winifred Adams, Geoffrey A. Sonn, Kathleen E. Mach,Ying Pan, Andrew H. Beck, Kristin C. Jensen, and Joseph C. Liao

OBJECTIVES To develop the diagnostic criteria for benign and neoplastic conditions of the urinary tract usingprobe-based confocal laser endomicroscopy (pCLE), a new technology for dynamic, in vivoimaging with micron-scale resolution. The suggested diagnostic criteria will formulate a guide forpCLE image interpretation in urology.

METHODS Patients scheduled for transurethral resection of bladder tumor (TURBT) or nephrectomy wererecruited. After white-light cystoscopy (WLC), fluorescein was administered as contrast. Differ-ent areas of the urinary tract were imaged with pCLE via direct contact between the confocalprobe and the area of interest. Confocal images were subsequently compared with standardhematoxylin and eosin analysis.

RESULTS pCLE images were collected from 66 participants, including 2 patients who underwent nephrec-tomy. We identified key features associated with different anatomic landmarks of the urinarytract, including the kidney, ureter, bladder, prostate, and urethra. In vivo pCLE of the bladderdemonstrated distinct differences between normal mucosa and neoplastic tissue. Using mosaicing,a post hoc image-processing algorithm, individual image frames were juxtaposed to form wide-angle views to better evaluate tissue microarchitecture.

CONCLUSIONS In contrast to standard pathologic analysis of fixed tissue with hematoxylin and eosin, pCLEprovides real time microscopy of the urinary tract to enable dynamic interrogation of benign andneoplastic tissues in vivo. The diagnostic criteria developed in this study will facilitate adaptationof pCLE for use in conjunction with WLC to expedite diagnosis of urinary tract pathology,

particularly bladder cancer. UROLOGY 78: 225–231, 2011. Published by Elsevier Inc.

be

i8a

White-light cystoscopy (WLC), the standardapproach to evaluate the lower urinary tract,has several well-recognized shortcomings,

particularly in differentiating nonpapillary urothelial car-cinoma from inflammatory lesions and delineation oftumor boundaries.1 Although histology remains the cor-nerstone of cancer diagnosis, it is not readily available inreal time to guide surgical intervention. Significant in-terest exists in developing adjunctive imaging technolo-gies to overcome the shortcomings of WLC—particularlyto identify tumors with clinically aggressive features—thereby improving the quality and yield of transurethralresection and overall clinical outcomes.

From the Department of Urology and Department of Pathology, Stanford CancerCenter and the Bio-X Program, Stanford University School of Medicine, Stanford,California; and the Veterans Affairs Palo Alto Health Care System, Palo Alto,California

Reprint requests: Joseph C. Liao, M.D., Department of Urology, Stanford Univer-sity School of Medicine, 300 Pasteur Drive, S-287, Stanford, CA 94305-5118.

E-mail: jliao@stanford.edu

Submitted: October 22, 2010, accepted (with revisions): February 14, 2011

Published by Elsevier Inc.

Several optical imaging technologies have recentlybeen approved for the urinary tract, including fluores-cence cystoscopy (photodynamic diagnosis), narrow bandimaging, and optical coherence tomography.2-7 Probe-ased confocal laser endomicroscopy (pCLE) is anothermerging technology under investigation.8,9 Based on the

established principle of confocal microscopy, pCLE pro-vides in vivo microscopy of living tissues through minia-turized fiberoptic probes inserted through working chan-nels of standard endoscopes, including cystoscopes. Incontrast to other imaging technologies, pCLE has mi-cron-scale resolution, thus making it possible to visualizetissue microarchitecture and cellular morphology remi-niscent of histology. Currently, pCLE is approved forclinical use in the respiratory and gastrointestinaltracts.10-14 For Barrett’s esophagus, a precancerous entityn which WL diagnosis is also challenging, sensitivity of8% and specificity of 96% with excellent interobservergreement have recently been reported with pCLE.15

The goal of the current study is to expand upon ourinitial feasibility study to provide in-depth characteriza-

tion of the confocal microscopy features of benign and

0090-4295/11/$36.00 225doi:10.1016/j.urology.2011.02.057

Fe2st

tci

sIp

nCq

neoplastic regions of the urinary tract. Although buildingdiagnostic criteria for bladder cancer remains the focus,confocal features of urethra, prostate, ureter, and kidneyare also investigated. For image analysis, we evaluated acomputer algorithm called mosaicing, which provides apanoramic view of the microarchitecture.16-18 The result-ing collection of images and suggested diagnostic criteriaprovide the foundation for a prospective diagnostic ac-curacy study of pCLE for bladder cancer diagnosis.

MATERIAL AND METHODS

Probe-Based Confocal Laser EndomicroscopyConfocal microscopy of the urinary tract was performed withthe Cellvizio system (Mauna Kea Technologies, Paris, France),and consisted of a fiberoptic imaging probe, a laser scanningunit, and a desktop computer with the image acquisition andprocessing software. In addition to the previously described2.6-mm-diameter probe (GastroFlex UHD),8 a 1.4-mm probe(AlveoFlex) was used in this study. The 2.6-mm probe has aresolution of 1 �m and a field of view (FOV) of 240 �m,whereas the 1.4-mm probe has a resolution of 3.5 �m and anOV of 600 �m. The probes were sterilized after each use withither the System 1 (Steris, Mentor, OH) or the STERRAD00 (ASP, Irvine, CA) system. Images were collected as videoequences at 12 frames per second through direct contact of theissue with the probe.

Imaging ProtocolThe study was approved by the Stanford University Institu-tional Review Board and VAPAHCS Research and Develop-ment Committee. Consecutive patients scheduled for cysto-scopic biopsy or TURBT, in addition to 2 patients whounderwent nephrectomy, provided consent. After initial exam-ination with WLC, all patients received fluorescein (AlconLaboratories, Fort Worth, TX) intravesically and/or intrave-nously. Fluorescein (MW � 332.3 g/mL) is approved by FederalDrug Administration for intravenous use for ophthalmologicimaging of blood vessels and has recently been shown to be asafe contrast agent for pCLE in the gastrointestinal tract.19 Forintravesical administration, 300-400 mL of 0.1% fluoresceindiluted in normal saline was instilled into the bladder via aFoley catheter and left indwelling for 5 minutes. For intrave-nous administration, 0.5 mL of 10% fluorescein was givenimmediately before pCLE imaging.

Confocal images were obtained with the 2.6-mm probe in-serted through standard 26-Fr resectoscope or with the 1.4-mmprobe in standard rigid (21-Fr) or flexible (15-Fr) cystoscopes.WLC videos were recorded simultaneously and selected stillimages taken. By varying the probe pressure on the surface,different layers of the epithelium of the lower urinary tract wereimaged. Kidneys from 2 nephrectomy patients were used for exvivo confocal imaging of the renal cortex, pelvis, and proximalureter. Topical fluorescein was applied ex vivo in 1 kidney,while intravenous fluorescein was given before division of therenal hilum in the other. Imaged tissue locations were biopsiedor resected and sent for pathologic analysis with hematoxylinand eosin (H&E) staining. Acquired confocal video sequenceswere edited and analyzed after the procedure and compared

with H&E-stained tissue. e

226

Mosaicing AlgorithmThe mosaicing algorithm (Mauna Kea Technologies) is animage-processing technique that compiles a series of consecu-tive images into a single larger composite image that effectivelyincreases the FOV.16,17 It is based on a hierarchical frameworkhat recovers a globally consistent alignment of input frames toompensate for distortions induced by the relative motion ofmaged tissue with respect to the probe.18

Image Selection CriteriaAfter real-time confocal video sequences were acquired, videoswere processed offline before image characteristics were ana-lyzed. Video clips and still images of interest were extractedfrom the original sequences. Clips were selected based on clarityof images and probe contact time longer than 2 seconds. Mo-saics were created using the algorithm described above. Pro-cessed confocal images were reviewed with 2 pathologists(K.C.J., A.H.B.) and compared with corresponding H&E stains.Diagnostic criteria representative of different types of bladdercancer were selected based on reproducibility and correspon-dence to H&E histology. Imaging characteristics that wereeasily recognizable and reproducible by the urologist (J.C.L.)were chosen for use as diagnostic criteria. Representative imagesfor the atlas were chosen based on resemblance to correspond-ing H&E images as well as image quality, taking into accountmotion and lens artifacts and level of staining.

RESULTSTo compile the imaging atlas, 566 real-time confocalvideo sequences from 66 study participants (mean age 71years, range 28-90 years) were analyzed. The mean dura-tion of confocal image acquisition was 15.5 minutes perparticipant. Average probe dwell time for each area was11.4 seconds and the average number clips per area was4.3. There were no adverse events related to pCLE orfluorescein administration, other than transient fluores-cence-stained urine. No differences were noted in histol-ogy quality of tissue treated with IV or intravesical fluo-rescein.

We evaluated mosaicing, an image processing tech-nique that juxtaposes overlapping frames into a singlecomposite image, thus expanding the FOV while improv-ing overall image resolution. Figure 1 shows representa-tive mosaics of the umbrella cell layer of normal urothe-lium (Fig. 1A) and prostatic urethra (Fig. 1B). Comparedwith individual image frames (figure inset), the compos-ite mosaics expanded the FOV by up to 5-fold (from 0.24to 1.27 mm). The characteristic polygonal-shaped um-brella cells, visualized in vivo, were reminiscent of the exvivo images of umbrella cells obtained using more pow-erful desktop confocal microscopes.20,21 The glandulartructures visualized through the prostatic urethra afterV fluorescein administration are consistent with benignrostatic glands.Figure 2 shows representative confocal images of the

ormal anatomic landmarks from the entire urinary tract.onfocal microscopy of the lower urinary tract was ac-uired in vivo, whereas upper tract images were obtained

x vivo. With either intravesical or intravenous fluores-

UROLOGY 78 (1), 2011

lm�

cein administration, circulating erythrocytes were rou-tinely observed within the rich capillary network of thelamina propria. The results indicate that within 5 min-utes of intravesical instillation, fluorescein diffuses acrossthe transitional cell layers into the lamina propria. Nor-mal bladder urothelium was characterized by a polygonalumbrella cell layer, uniform smaller intermediate celllayer, and capillary networks with erythrocytes in thelamina propria (Figs. 1A and 2). Although the limiteddepth of penetration prohibited imaging beyond the lam-ina propria through intact urothelium, the fibers of themuscularis propria and the adipocytes of the perivesicalfat were imaged in the tumor resection bed in 14 partic-ipants. In the penile urethra, the lamina propria is char-acterized by branching thin dark bands that likely repre-sent small vessels. In this first reported application of

Figure 1. Confocal image processing with mosaicing algoriture with improved clarity compared to single image frameayer: note the characteristic large polygonal-shaped cellsultiple dark prostate glands within the white stroma. Note5-fold.

pCLE of the upper urinary tract, the renal cortex was

UROLOGY 78 (1), 2011

characterized by tubular structures consistent with con-voluted renal tubules, while renal pelvis and proximalureter showed urothelial cells and lamina propria similarto bladder urothelium.

Confocal features of low grade urothelial carcinomawere defined by analysis of 64 confocal videos from 22pathologically confirmed tumors, most of which werepapillary by WLC. Figure 3A-F shows the diagnostic con-focal features of low grade tumors, including densely packedurothelial cells that are homogenous and monomorphic.Other characteristics include cellular papillary structuresand fibrovascular stalks from tumor neoangiogenesis thatwere not observed in normal or inflamed areas. Duringreal-time imaging, fibrovascular stalks were easily visualizedbecause of the constant flow of erythrocytes.

For high grade urothelial carcinoma, 73 confocal vid-

that generates panoramic view of the tissue microarchitec-set). (A) Normal bladder urothelium showing umbrella cell) Normal prostate imaged through prostatic urethra withmosaics significantly increased the overall field of view by

thms (in. (Bthe

eos from 27 pathologically confirmed tumors were ana-

227

vesic

lyzed. With WLC, high-grade tumors were variable inappearance, ranging from papillary, to sessile, to flat.Microscopically, confocal criteria for high grade tumorsoverlap with those of low grade, including densely packedcells and presence of fibrovascular stalks. Consistently,however, high grade tumors exhibited a far more dis-torted microarchitecture; these tumors exhibited a pleo-morphic population of irregularly shaped cells with loss ofcellular cohesiveness, indistinct cell borders, and disor-ganized vasculature (Fig. 3G-J). Interestingly, neither theumbrella cell layer nor the capillary network of the lam-ina propria commonly seen in normal urothelium werevisualized with pCLE of urothelial carcinoma, includingTa, T1, and CIS.

Among the most challenging lesions to diagnose onWLC, CIS appears typically as erythematous patches,which are difficult to differentiate from benign condi-tions, such as post-BCG granulomatous reaction, Foleycatheter reaction, urinary tract infections, or scarringfrom previous resection. In our study, 7 of the high-gradetumors imaged were confirmed to be CIS. Figure 4 showsa comparison between CIS and benign conditions. Con-

Figure 2. Probe-based confocal laser endomicroscopy (pCurinary tract were acquired in vivo, whereas upper tract imsimilarities of the urothelium (intermediate cells) betweennetwork of moving erythrocytes. Muscularis propia and peri

focal features of CIS include larger, more pleomorphic

228

cells compared to inflammatory conditions, indistinctcell borders, and extensive acellular areas that may bedue to denuded urothelium. In contrast, confocal imagingof tissue with inflammation showed loosely arranged ag-gregations of smaller monomorphic cells in the laminapropria consistent with local recruitment of inflammatorycells. Confocal images of scar tissue revealed character-istics similar to the intermediate cells and lamina propriaof normal urothelium.

DISCUSSIONIn this study, we have compiled a confocal imaging atlasof the lower and upper urinary tract and developed diag-nostic criteria for normal, inflammatory, and malignantlesions of the bladder. Expanding on our feasibility report(n � 27),8 we have confirmed the safety profile of fluo-rescein administration. Key criteria for normal and be-nign epithelium include uniform cellular architecture,and visualization of the lamina propria layer and itscapillary network. Low-grade carcinoma was character-ized by fibrovascular stalks, and high cellular density,

mages of the normal urinary tract. All images of the lowers (kidney cortex and ureter) were acquired ex vivo. Noter and bladder. Lamina propria is characterized by capillaryal fat images were obtained from the tumor resection bed.

LE) iage

urete

whereas high-grade carcinoma was characterized by irreg-

UROLOGY 78 (1), 2011

posmdptotttsw

licb

s

ular architecture and pleomorphic cells. These distinctcharacteristics were consistently observed in normalurothelium and bladder cancer. Development of objec-tive diagnostic criteria is essential for future studies onthe diagnostic accuracy (sensitivity, specificity, and pos-itive and negative predictive value) of pCLE in theurinary tract.

Previously, we had found the small FOV and motionsensitivity of the confocal probe to be limitations.8 In theresent study, the mosaicing algorithm was used duringffline analysis of confocal video sequences to createtatic images with wider FOV and to compensate forotion introduced by the operator or the subject (eg,

iaphragmatic motion or vascular pulsation). For exam-le, mosaicing of the prostatic urethra images enabled uso visualize a single prostatic gland within the context ofther glands in the stroma (Fig. 1B). If applied in realime during TURBT, this mosaicing algorithm may po-entially delineate tumor boundaries by highlighting theransitional area of the lesion with adjacent normal tis-ue, especially in the case of broad based tumors or CIShere the borders are not distinct on WLC.Another previously reported limitation of pCLE is the

ack of deflectability with the rigid cystoscope, resultingn difficulty establishing perpendicular contact of theonfocal probe with the mucosa in certain parts of theladder (eg, anterior).8 We have investigated the use of a

1.4-mm probe that fits inside the flexible cystoscope andreport the in-depth comparative analysis of the 2.6- and1.4-mm probes elsewhere.22 Our preliminary ex vivo ex-perience of applying pCLE to the upper urinary tract wasfavorable (Fig. 2). With the availability of confocalprobes less than 1 mm in diameter,23,24 in vivo applica-tion of pCLE for upper tract pathology will be investi-gated in the future. The possibility of in vivo opticalbiopsy of the upper urinary tract with pCLE is particularly

Figure 3. Characteristic in vivo microscopy features of lowing H&E sections. Low grade urothelial carcinoma: (A, B) Croa thickened endothelial layer. (E, F) Cross-sectional mosavascular core. High-grade urothelial carcinoma: (G, H) Pleomtalk with variably sized vascular cores.

attractive, given the challenges associated with obtaining

UROLOGY 78 (1), 2011

physical biopsies of high quality through standard uret-eroscopes.

Our study has limitations, including observer bias in-herent in our single center experience. Image analysiswas performed retrospectively after pathologic confirma-tion using H&E, which may have biased chosen confocalcharacteristics. As with all new technologies, there is anassociated learning curve, in part based on the differentplane of examination with the respective technologies.Because confocal microscopy emphasizes the cells withinthe superficial layer of the mucosa, the interpreter mustconsider cell size and planar cellular organization (ordisorganization) over the traditional histopathologic fea-tures of cellular stratification and nuclear size and pleo-morphism. In our ongoing diagnostic accuracy study,confocal images will be reviewed offline and character-ized with these diagnostic criteria by multiple blindedreviewers to assess interobserver variability and reducediagnostic bias from WLC and pathologic findings.

Images in this study were obtained after fluoresceinadministration intravesically, intravenously, or both. Al-though the intravenous route offers greater efficiency (ie,image acquisition can start within 1-2 minutes of admin-istration) and subjectively more consistent staining, ef-flux of fluorescent urine from the ureteral orifices occursas short as 3-5 minutes after intravenous administra-tion, which may potentially impair visualization duringTURBT. Due to this concern, intravesical administrationhas become our route of choice for bladder imaging.Intravenous fluorescein can still be administered afterintravesical administration, if topical staining was subop-timal and extravesical imaging (eg, prostate, urethra) isdesired. Interestingly, we have found that similar to in-digo carmine, intravenous fluorescein may be used toidentify ureteral orifices intraoperatively.

pCLE is inherently limited by reliance on macroscopic

igh grade urothelial carcinoma compared with correspond-g of uniform-appearing cells. (C, D) Fibrovascular stalk withw of the fibrovascular stalk containing erythrocytes in theic and distorted sheet of cells. (I, J) Distorted fibrovascular

and hwdin

ic vieorph

optical imaging (WLC in our study) to identify regions of

229

ras

CfA

Cbbnpha

interest for further microscopic characterization, as it isnot practical to survey the entire bladder with pCLE.Combining pCLE with other macroscopic, “red flag”technologies, such as fluorescence cystoscopy or narrowband imaging may be useful to better characterize flatlesions, such as CIS, which remain the most significantchallenge across all technology platforms. Nuclear mor-phology, which is critical for standard tumor gradingbased on H&E, is not possible since fluorescein onlystains extracellularly.

A promising approach to identify neoplastic lesionsendoscopically with molecular specificity is the use offluorescently labeled cancer-specific peptides in conjunc-tion with pCLE.25,26 With the well-established trackecord of intravesical therapy, the bladder provides anttractive route to investigate the feasibility of endo-

Figure 4. Images of erythematous flat lesions in the bladdystoscopic view of erythematous patches in 2 patients wiiopsy site. (B and E) Confocal images showing large pleomorders. (C and F) Characteristic histologic images ofuclear/cytoplasmic ratios, nuclear hyperchromasia, nucleatch on WLC with (H) small monomorphic cells infiltratinistology. (J) Patient with scar from previous TURBT showingnd (L) thickened lamina propria on histology.

copic molecular imaging. f

230

CONCLUSIONSWe have compiled representative confocal images fromthe entire urinary tract, including the first imaging of theupper tract, into a confocal imaging atlas. We proposethese confocal diagnostic imaging criteria for normal,inflammatory, and low and high grade cancer to enableadaptation of pCLE as an adjunctive imaging modalityfor bladder cancer. Current prospective diagnostic accu-racy studies are underway to assess the sensitivity andspecificity of these confocal criteria in the diagnosis ofbladder cancer.

Acknowledgments. We thank Stanford University Cancerenter Developmental Cancer Research Award (to J.C.L.) for

unding and Mauna Kea Technologies for technical support.ssistance from the VAPAHCS operating room staff is grate-

y WLC, pCLE, and corresponding H&E sections. (A and D)S. Cautery mark in A used to coregister pCLE imaging andic cells, loss of cellular cohesiveness, and indistinct cellularwith full-thickness high-grade cellular atypia, increased

leomorphism, and cellular discohesion. (G) Erythematouse lamina propria and (I) chronic lymphocytic infiltrate onnormal appearing intermediate cells with clear cell borders

er bth CIorphCIS,ar pg th(K)

ully acknowledged.

UROLOGY 78 (1), 2011

1

1

1

1

1

1

1

1

1

1

2

2

2

References1. Lee CS, Yoon CY, Witjes JA. The past, present and future of

cystoscopy: the fusion of cystoscopy and novel imaging technology.BJU Int. 2008;102:1228-1233.

2. Cauberg EC, de Bruin DM, Faber DJ, et al. A new generation ofoptical diagnostics for bladder cancer: technology, diagnostic ac-curacy, and future applications. Eur Urol. 2009;56:287-296.

3. Cauberg EC, Kloen S, Visser M, et al. Narrow Band ImagingCystoscopy Improves the Detection of non-muscle-invasive Blad-der Cancer. Urology. 2010;76:658.

4. Goh AC, Tresser NJ, Shen SS, et al. Optical coherence tomogra-phy as an adjunct to white light cystoscopy for intravesical real-time imaging and staging of bladder cancer. Urology. 2008;72:133-137.

5. Herr H, Donat M, Dalbagni G, et al. Narrow-band imaging cys-toscopy to evaluate bladder tumours—individual surgeon variabil-ity. BJU Int. 2010;106:53-55.

6. Schmidbauer J, Remzi M, Klatte T, et al. Fluorescence cystoscopywith high-resolution optical coherence tomography imaging as anadjunct reduces false-positive findings in the diagnosis of urothelialcarcinoma of the bladder. Eur Urol. 2009;56:914.

7. Witjes JA, Redorta JP, Jacqmin D, et al. Hexaminolevulinate-guided fluorescence cystoscopy in the diagnosis and follow-up ofpatients with non-muscle-invasive bladder cancer: review of theevidence and recommendations. Eur Urol. 2010;57:607.

8. Sonn GA, Jones SN, Tarin TV, et al. Optical biopsy of humanbladder neoplasia with in vivo confocal laser endomicroscopy.J Urol. 2009;182:1299-1305.

9. Sonn GA, Mach KE, Jensen K, et al. Fibered confocal microscopyof bladder tumors: an ex vivo study. J Endourol. 2009;23:197-201.

0. Banno K, Niwa Y, Miyahara R, et al. Confocal endomicroscopy forphenotypic diagnosis of gastric cancer. J Gastroenterol Hepatol.2010;25:712-718.

1. Dunbar KB, Okolo P, 3rd, Montgomery E, et al. Confocal laserendomicroscopy in Barrett’s esophagus and endoscopically inappar-ent Barrett’s neoplasia: a prospective, randomized, double-blind,controlled, crossover trial. Gastrointest Endosc. 2009;70:645-654.

2. Thiberville L, Moreno-Swirc S, Vercauteren T, et al. In vivoimaging of the bronchial wall microstructure using fibered confocalfluorescence microscopy. Am J Respir Crit Care Med. 2007;175:22-31.

3. Thiberville L, Salaun M, Lachkar S, et al. Human in vivo fluores-cence microimaging of the alveolar ducts and sacs during bron-

choscopy. Eur Respir J. 2009;33:974-985.

UROLOGY 78 (1), 2011

4. Wang TD, Friedland S, Sahbaie P, et al. Functional imaging ofcolonic mucosa with a fibered confocal microscope for real-time invivo pathology. Clin Gastroenterol Hepatol. 2007;5:1300-1305.

5. Wallace MB, Sharma P, Lightdale C, et al. Preliminary accuracyand interobserver agreement for the detection of intraepithelialneoplasia in Barrett’s esophagus with probe-based confocal laserendomicroscopy. Gastrointest Endosc. 2010;72:19.

6. Becker V, Vercauteren T, von Weyhern CH, et al. High-resolutionminiprobe-based confocal microscopy in combination with videomosaicing (with video). Gastrointest Endosc. 2007;66:1001-1007.

7. De Palma GD, Staibano S, Siciliano S, et al. In vivo characterisa-tion of superficial colorectal neoplastic lesions with high-resolutionprobe-based confocal laser endomicroscopy in combination withvideo-mosaicing: a feasibility study to enhance routine endoscopy.Dig Liver Dis. 2010;42:11:791-797.

8. Vercauteren T, Perchant A, Malandain G, et al. Robust mosaicingwith correction of motion distortions and tissue deformations for invivo fibered microscopy. Med Image Anal. 2006;10:673-692.

9. Wallace MB, Meining A, Canto MI, et al. The safety of intrave-nous fluorescein for confocal laser endomicroscopy in the gastro-intestinal tract. Aliment Pharmacol Ther. 2010;31:548-552.

0. Rickard A, Dorokhov N, Ryerse J, et al. Characterization of tightjunction proteins in cultured human urothelial cells. In Vitro CellDev Biol Anim. 2008;44:261-267.

1. Yu W, Khandelwal P, Apodaca G. Distinct apical and basolateralmembrane requirements for stretch-induced membrane traffic atthe apical surface of bladder umbrella cells. Mol Biol Cell. 2009;20:282-295.

2. Adams W, Wu K, Liu J-J, Hsiao ST, Jensen KC, Liao JC. Com-parison of 2.6 mm and 1.4 mm imaging probes for confocal laserendomicroscopy of the urinary tract. J Endourol. 2011; In Press.

23. Becker V, Wallace MB, Fockens P, et al. Needle-based confocalendomicroscopy for in vivo histology of intra-abdominal organs:first results in a porcine model (with video). Gastrointest Endosc.2010;71:7:1260-1266.

24. Meining A. Confocal endomicroscopy. Gastrointest Endosc Clin NAm. 2009;19:629-635.

25. Hsiung PL, Hardy J, Friedland S, et al. Detection of colonicdysplasia in vivo using a targeted heptapeptide and confocal mi-croendoscopy. Nat Med. 2008;14:454-458.

26. Li M, Anastassiades CP, Joshi B, et al. Affinity peptide for targeteddetection of dysplasia in Barrett’s esophagus. Gastroenterology.

2010;139:5:1472-1480.

231