Journal of Dermatological Science (2006) 43, 135—141

www.intl.elsevierhealth.com/journals/jods

The real-time, three-dimensional analyses ofbenign and malignant skin tumors byconfocal laser scanning microscopy

Ichiro Ono *, Akiko Sakemoto, Jiro Ogino, Takafumi Kamiya,Toshiharu Yamashita, Kowichi Jimbow

Sapporo Medical University School of Medicine, Department of Dermatology,Chuoku, South-1 West-16 Sapporo, Hokkaido 060-8543, Japan

Received 4 March 2006; received in revised form 17 April 2006; accepted 1 May 2006

KEYWORDSSkin tumors;Clinico-pathologicalanalysis;Basal cell carcinoma;Malignant melanoma;Microstructure;Multiplanarreconstruction

Summary

Background: In obtain images of skin tumors non-invasively with real-time, confocallaser scanning microscope (CLSM) is introduced.Objective: Reconstructed images of given horizontal sections were converted intothree-dimensions using the data set of a large number of tomograms in the horizontaldirections.Methods: To develop themultiplaner reconstruction images of skin tumors in verticaldirections and three-dimensionally reconstructed images of tumors will be obtainedfrom the continuously collected horizontal image data sets.Results: Three-dimensional analyses of the skin tumors from reconstructed images ofthe CLSM scanning have provided the information as to their physiological character-istics as well as the extent of deep invasion in real-time with non-invasive manner.High performance three-dimensional conversion software was effective in displayingthree-dimensional construction of skin tumors.Conclusion: The CLSM scanning images followed by three-dimensional reconstructionusing them can provide the real-time and non-invasive diagnoses of skin tumors andanalyze the radial growth phase of tumors and the three-dimensional growth char-acteristics.# 2006 Japanese Society for Investigative Dermatology. Published by Elsevier IrelandLtd. All rights reserved.

* Corresponding author. Tel.: +81 11 611 2111x3458; fax: +81 11 613 3739.E-mail address: ichiro@sapmed.ac.jp (I. Ono).

0923-1811/$30.00 # 2006 Japanese Society for Investigative Dermatology. Published by Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.jdermsci.2006.05.001

136 I. Ono et al.

1. Introduction

Recently, the dermoscope has been introduced andits diagnostic significance has been confirmed [1—7].Its diagnostic significance is especially high for pig-mented tumors such as melanocytic nevi, malignantmelanoma and pigmented basal cell carcinoma[3,6—8]. However, it has not been feasible to non-invasively diagnose the innovational depth andextent of skin tumors at the level of the individualcell in real-time [9].

The confocal laser scanning microscope (CLSM)has been developed, and is being introduced forclinical use recently [10—13]. With this method, alaser beam is irradiated to an in vivo or ex vivotissues. The reflected light is processed in a confocalmanner, and horizontal images of skin tumors fromcross sections can be generated in real-time andobserved [10—12,14,15]. In this study, from CLSMsectional images of skin tumors, images in verticaldirections are generated and then their image dataare reconstructed three-dimensionally [16]. Theclinical significance of those CLSM images are dis-cussed by comparing these images with informationobtained by other conventional strategies.

2. Materials and methods

2.1. Devices used and data collection

The CLSM used in this study was a Vivascope 1500(Lucid Inc., Rochester, NY, USA). This device emitsinfrared radiation by using a diode laser as an oscilla-tion source.The reflected light is guided toadetectorby a split mirror and processed in a confocal manner.Thus, a horizontal sectional imageof skin in a rangeof500 mm � 500 mm is generated in real-time. Thedisplayed pixel count is 1000 pixels � 1000 pixelsand the display gray scale is 8 bit (256 scales). Thewavelength of the infrared laser beam is 830 nm. Thereflected light is finely moved in the direction of theX- and Y-axes with a motor driven imaging modulecontrolled by a computer using a water immersionobjective lenswith a focal lengthof 5.3 mm.Further-more, it is possible to generate images deep into theskin by repeating the procedure described above andmoving the imaging module in the direction of the Z-axis with the motor at minimum intervals of 1.6 mm.In this case, the resolution decreases in deeper areasof the skin due to scattering laser light [12]. However,it is possible to generate images up to a depth ofabout 350 mm,which a laser beam can reach by usingthe Viva Stack repeatedly. In this study, up to 220slices of data were collected by this method withinterval of 1.6 mm.

2.2. Data processing

The images obtained by the CLSM are immediatelydisplayed on the monitor of the device. At the sametime, theyare automatically savedon theharddisc ofthe Vivascope 1500 in the BMP format. In this study,the data were transferred to an external hard disc oftheanalytical computer of thedevice.Then, thedatawere analyzed with a MacBook Pro computer (Applecomputer Inc., Cupertino CA, USA). The softwareused for analysis in this study included PhotoshopCS (Adobe Systems Incorporated, San Jose, CA, USA),Image J version 1.33 u (National Institute of Health,Washington, WA USA), QuickTime version 7.03 (Applecomputer Inc., Cupertino, CA, USA), and OsiriX Med-ical Imaging Software version 2.3.1 (David GeffenSchool of Medicine, University of California LosAngeles, Los Angeles, CA, USA).

To observe each image in a multiple image set,the data set was imported into Image J. If there is anaxial deviation in the vertical direction due topatient movement, it is corrected automaticallyby the StackReg, which is also plug-in software forImage J. Then, three-dimensional images were gen-erated by OsiriX. This software can generate notonly the reconstructed images in the vertical planeimage, but also the three-dimensional images byusing the volume rendering method after settingwindow level and window width. OsiriX supports forimage fusion, so images obtained during the samedata sets with altering window level and windowwidth will be automatically aligned by OsiriX. If thetwo image series with differing window level andwindow width have the same number of slices andthe same slice spacing, then stack synchronizationcan be useful in linking the two data sets differentcolors. The fusion intensity is controlled with the‘‘Fusion Percentage’’ slider from the toolbar tounderstand the microstructure of skin tumors, espe-cially the distribution of melanin containing cells.

Tumor tissues, includingmelanocytic nevus, basalcell carcinoma and malignant melanoma, were col-lected from our patients.

3. Results

The resolution of the horizontal sectional imagesthrough CLSM observation was such that each cellcould be discriminated. Cells containing a largeamount of melanin pigments were imaged in whitecolor because the reflection coefficient ofmelanin ishigh. It is apparent that the quality of this method isgood enough to identify cells in normal skin andtumor cells, such as the neoplastic aggregates andstroma in BCC that contain melanin pigments

Analyse

sofskin

tumors

with

real

time,co

nfocal

laserscan

ningmicro

scopy

137

Table 1 CLSM images of normal skin and skin tumor

Normalskin

Melanocyticnevi congenital

Melanocyticnevi aquired

Seborrheickeratosis

Basal cellcarcinoma

Malignantmelanoma

Continuity ofbasal layer

+ + + + + �

Indivisualcell shape

(Uniform) Uniform Uniform Uniform Large, but uniform Large, dysplastic

Melanindistribution

Basalcells

Basal cells;nevus cells

Basal cells;nevus cells

Entire epidrmis Melanocytessurroundingtumor cells

Melanoma cells;at various locationcasting-off cellsintra epidermis

Tumor cellnesting

� + + � + +

Other features Irregular rete ridges;melanin containingnevus cells at apexof dermal pappila

Enlarged papillarydermis; melanincontaining nevuscells at apexof dermal pappila

Acantosiscomposed ofmelanin containingkeratinocytes; darkspherical or ovalcell structures

Dark zone completelysurrounding tumor;palisade arrangementwith leaf-like structure

Presence of nestswith various size;melanin-containingcells in the dermis.

Fig.

1Case

1:melan

ocytic

nevu

s:an

18ye

ar-old

lady

with

melan

ocytic

nevu

sonherface

.(a)

Horizo

ntal

CLSM

image

sofbasal

layerleve

l(scale

bar

100mm).

(b)Tw

o-

dim

ensio

nalM

PRim

agewith

data

takenbyCLSM

.Enlarge

dpap

illaryderm

ispacke

dwith

nevu

sce

lls.(c)

Three-

dim

ensio

nal

CLSM

image

,which

was

mad

ebyremovin

gthedata

ofepiderm

isfro

mdata

takenbyCLSM

.Atthis

fusio

nim

age,melan

inco

ntain

ingnevu

sce

llsin

unifo

rmsize

atap

exofderm

alpap

illaan

dmelan

inco

ntain

ing

keratin

ocyte

swere

observe

das

red.

138 I. Ono et al.

Fig. 2 Case 2: basal cell carcinoma: an 72 year-old manwith a 2.5 cm � 1.8 cm nodule slightly elevated on hisback. (a) Horizontal CLSM images of basal layer level(scale bar 100 mm). The images by the CLSM showed aleaf-shaped tumor nest that continued to the basal celllayer and expanded to the upper dermis. Additionally, adark zone existed around the tumor nest. (b) Two-dimen-sional MPR image with data taken by CLSM. (c) Three-

[10,18]. In addition, it was possible to collect con-tinuous images from the surface of the skin to thedeeper dermal part of the skin, which can generateimages continuously. However, clear images couldnot be reconstructed by reconstructing any givensection in the vertical direction, even if data setswere collected in conditions under ex vivo with lessmovement. They had even less diagnostic signifi-cance in the in vivo data set with patient’s move-ment. In contrast, images were reconstructed in thevertical direction by using data sets modified withStackReg. This improves image definition signifi-cantly, especially. It was possible to display thethree-dimensional construction of tumor tissuesby analyzing the data sets with high-performancethree-dimensional conversion software such asOsiriX, increasing the usability of the method espe-cially using fusion procedures under.

Representative results of the comparison of theimage of skin tumors using the CLSM before andafter the surgical excision are summarized in tableand representative findings from each types oftumors and shown them both the histopathologicalcomparison (Table 1).

3.1. Case 1

Melanocytic nevus: an 18 year-old lady with melano-cytic nevus on her face, which diagnosed as acquiredmelanocytic nevus with Miesher type clinically. Nestsof cells with a high reflectance and even size, whichwere slightly larger than the basal cells wereobserved in some parts of the papillary dermis bythe horizontal sectional images with the CLSM.Reconstructed images in the vertical direction wereprepared using the data set corrected by StackReg.There was a clump of cells with a high reflectance,whichwas comparatively larger and even sized in theupper dermis. In this case, a clearer cell nest in thepapillary dermis was seen from the vertical planewhen observed with the ex vivo CLSM. The cells gotsmaller in the deeper dermis, a phenomenon called‘‘maturation’’. No dysplasia or nuclear divisionimages were observed also in the conventional his-topathological images as well as those with the exvivo CLSM of vertical plane as well as multuplanarreconstructed images. The fusion image revealedthat melanin containing nevus cells in uniform sizeat apex of dermal papilla and melanin containingkeratinocytes were observed as red (Fig. 1a—c).

dimensional CLSM fusion image, which was made byremoving the data of epidermis from data taken by CLSM.The tumor nest was leaf like shaped and surrounded bymelanin containing cells, which indicated red, and socalled ‘‘dark zone’’.

Analyses of skin tumors with real time, confocal laser scanning microscopy 139

Fig. 3 Case 3: malignant melanoma: a 72 year-old ladywith an irregular shaped black colored tumor on her rightsole. (a) Horizontal CLSM images of horny layer level (scalebar 100 mm). Large dysplastic cells containing differentamount of melanin of were observed from the level of thehorny cell layer in horizontal section. (b) Two-dimensional

3.2. Case 2

Basal cell carcinoma: an 72 year-old man with a2.5 cm � 1.8 cm nodule slightly elevated on hisright cheek. The images with the CLSM showed aleaf-shaped tumor lump, which continued to thebasal epidermal layer and expanded to the upperdermis. There was a dark zone with a low reflec-tance surrounding the tumor nest. Not only thehorizontal sectional images, but also the recon-structed images in the vertical direction showeddark zones surrounding the tumor lumps. Three-dimensional CLSM ‘‘fusion image’’ from data takenby CLSM revealed that the tumor nest was leaf likeshaped and surrounded by melanin containing cells,which indicated red, and so called ‘‘dark zone’’.(Fig. 2a—c).

3.3. Case 3

Malignant melanoma: a 76 year-old lady with anirregular shaped black colored tumor on her rightsole. In the CLSM images, large dysplastic cells withdifferent sizes variable melanin contents wereobserved from the level of the horny cell layer inhorizontal section, and many clumping cells thatwere supposed to be ‘‘casting off’’ of melanin-con-taining cells were observed in the epidermis as wellas horny layer. Uneven sized tumor nests, appar-ently different from the melanocytic nevus cells,which have complex branching dendrites, typical ofmelanoma on acral sites, and disarray of the normalarchitecture of the superficial epidermal layers,were also observed. The same observations weremade in the image from vertical-sectional recon-struction images. The histopathological imagesshowed the clumping cells with nuclei rich in nearlyround chromatin of different sizes, and melaninpigments forming large and small alveolar config-urations. In some regions, images of ‘‘casting-off’’were observed. The reconstructed images revealedalveolar configuration consisting of various large andsmall cells with different contents of melanin. Inaddition, cells containing a large amount of melanin

MPR image showed an alveolar configuration consisting ofvarious large and small cells with different contents ofmelanin in dermis. Many clumping cells that were sup-posed to be cast off were observed in the epidermis as wellas the horny layer. (c) Three-dimensional CLSM imageshowed the various sized tumor cell nests in addition tomelanin clumping cells that were supposed to be cast offwere observed in the epidermis as well as the horny layerin addition to tumor nest in papillary dermis. At this fusionimage, melanin containing melanoma cell in various sizenests were observed as redwhich are also different in size.

140 I. Ono et al.

corresponding to the ‘‘casting-off’’ cells wereobserved in the horny layer and epidermis by‘‘fusion image’’ (Fig. 3a—c).

4. Discussion

Recently, it has been recognized that the dermo-scope and CLSM have diagnostic significance to skintumor cells, which contain melanin pigments [8—15,17,19—21]. The introduction of these two meth-ods has allowed us to observe microstructures inreal-time non-invasively. In particular, although thedepth of penetration is limited up to 300 mm, theCLSM can confirm non-invasively the extent of skintumors at the cellular level [16,18,22]. In thisrespect, the echogram is also a technology to befocused on [9]. However, it is difficult to performanalyses with the resolution of cellular level.Increased clinical application of CLSM is anticipated[15,23—27]. In this study, reconstructed images ofgiven sections were converted into three-dimen-sions using the data set of a large number of tomo-grams in the horizontal directions. This kind ofmethod is already common in the field of computedtomography (CT) [28—30].

In this study, sectional images in a verticaldirection generated by reconstructing the imagedata of a skin malignancy obtained by the CLSM,and a method to generate multiplaner reconstruc-tion and three-dimensionally processed imagesand their clinical significance are discussed mainlyby comparing them with information obtained byconventional strategies. However, there is noknown report of performing three-dimensionalimaging of skin tumors by using the data set ofthe CLSM collected from living individuals. Fromthis strategy, visualization of basal layer and nor-mal epidermo—dermal junction in three-dimen-sional information of skin tumors are obtainedand the clinical significance of those analyticalmethod is proved high. Especially, three-dimen-sional reconstructed images provided the extent ofvarious sized tumor nests at skin which containmelanin, will contribute the understanding thenature of the tumor and its way of invasion [16]in addition to analyze the radial growth phase oftumors and the growth characteristics from theoseimages.

In conclusion, although, not all lesions image wellat present, and much experience is required for acorrect interpretation and to correlate the resultsobtained with the divise, the CLSM scanning imagesfollowed by three-dimensional reconstruction canbe a more sophisticated tool to diagnose skin tumorsnon-invasively [2,26,31].

References

[1] Fitzpqtrick TB, Bernhard JD, Cropley TG. The sytuctures ofskin lesions and fundamentalds of diagnosis. In: FitzpqtrickIM, Eisen AZ, Wolff K, Austen KF, Goldsmith LA, Katz SI,Fitzpqtrick TB, editors. Fitzpqtrick’s dermatology in generalmedicine. 5th ed., New York: McGraw Hill; 1999. p. 13—41.

[2] Charles CA, Marghoob AA, Busam KJ, Clark-Loeser L. Mela-noma or pigmented basal cell carcinoma: a clinical-patho-logic correlation with dermoscopy, in vivo confocal scanninglaser microscopy, and routine histology. Skin Res Technol2002;8:282—7.

[3] Soyer HP, Kenet RO, Wolf IH, Kenet BJ, Cerroni L. Clinico-pathological correlation of pigmented skin lesions usingdermascopy. Eur J Dermatol 2000;10:22—8.

[4] Soyer HP, Smolle J, Holdi S, Pachernegg H, Kerl H. Surfacemicroscopy: a new approach to the diagnosis of cutaneouspigmented tumors. Am J Dermatopathol 1989;11:1—10.

[5] Yadav S, Vossaert KA, Kopf AW, Silverman M, Grin-JorgensenC. Histopathologic correlates of structures seen on dermas-xopy (epiluminescence microscopy). Am J Dermatol 1993;15:297—305.

[6] Argenziano G, Soyer HP, Chimenti S, Talamini R. Dermoscopyof pigmented skin lesions: results of a consensus meeting viathe Internet. J Am Acad Dermatol 2003;48(5):679—93.Review.

[7] Crotty KA, Menzies SW. Dermoscopy and its role in diagnosingmelanocytic lesions: a guide for pathologists. Pathology2004;36:470—7.

[8] Pellacani G, Cesinaro AM, Longo C, Grana C, Seidenari S.Microscopic in vivo description of cellular architecture ofdermascopic pignent network in nevi and melanomas. ArchDermatol 2005;141:147—54.

[9] Skvara H, Teban L, Fiebiger M, Binder M, Kittler H. Limita-tions of dermascopy in the recognition of melanoma. ArchDermatol 2005;141:155—60.

[10] Rajadhyakha M, Grossman M, Esterowitz D, Webb RH, Ander-son RR. In vivo confocal laser microscopy of human skin:melanin provides strong contrast. J Invest Dermatol1995;104:946—52.

[11] Rajadhyakha M, Gonzalez S, Zavislan JM, Anderson RR,WebbRH. In vivo confocal scanning laser microscopy of human skinII: advances in instrumentation and comparison with histol-ogy. J Invest Dermatol 1999;113:293—303.

[12] Huzaira M, Rius F, Rajadhyakha M, Anderson RR, Gonzalez S.Topographic variations in normal skin, as viewed by in vivorefrectance confocal microscopy. J Invest Dermatol 2001;116:846—52.

[13] Lsngley RGB, Rajadhyakha M, Dwyer PJ, Sober AJ, Flotte TJ,Anderson RR. Confocal scanning laser microscopy of benighnand malignant melanocyteic skin lesions in vivo. J Am AcadDermatol 2001;45:365—76.

[14] Gonzalez S, Tannous Z. Real-time, in vivo confocal reflec-tance microscopy of basal cell carcinoma. J Am Acad Der-matol 2002;47:869—74.

[15] Chung VQ, Dwyer PJ, Nehal KS, Rajadhyakha M, Manaker GM,Charles C, et al. Use of ex vivo confocal scanning lasermicroscopy during Moes surgery for nonmelanoma skin can-cers. Dermatol Surg 2004;30:1470—8.

[16] Masters BR, Gonnord G, Corcuff P. Three-dimensional micro-scopic biopsy of in vivo human skin: a new technique based ona flexible confocal microscope. J Microsc 1997;185:329—38.

[17] Rallan D, Harland CC. Skin imaging: is it clinically useful? ClinExp Dermatol 2004;29:453—9.

[18] Yamashita T, Kuwahara T, Gonzalez S, Takahashi M. Non-invasive visualization of melanin and melanocytes by reflec-

Analyses of skin tumors with real time, confocal laser scanning microscopy 141

tance-mode confocal microscopy. J Invest Dermatol 2005;124:1—6.

[19] Gerger A, Koller S, Kern T, Massone C, Steiger K, Richtig E,et al. Diagnostic applicability of in vivo confocal laserscanning microscopy in melanocytic skin tumors. J InvestDermatol 2005;124:493—8.

[20] Busam KJ, Hester K, Charles C, Sachs DL, Antonescu CR,Gonzalez S, et al. Detection of clinically amelanotic malig-nant melanoma and assessment of its margins by in vivoconfocal scanning laser microscopy. Arch Dermatol 2001;137:923—9.

[21] Nori S, Rius-Diaz F, Cuvas J, Goldgeier M, Jaen P, Torres A,et al. Sensitivity and apecificity of reflectance-mode con-focal microscopy for in vivo diagnosis of basal cell carci-noma: a multicenter study. J Am Acad Dermatol 2004;51:923—30.

[22] Ruocco E, Argenziano G, Pellacani G, Seidenari S. Noninva-sive imaging of skin tumors. Dermatol Surg 2004;30:301—10.

[23] Rajadhyakha M, Manaker G, Flotte T, Dwyer PJ, Gonzalez S.Confocal examination of nonmelanoma cancers in thick skinexcisions to potentially guide Mohs micrographic surgerywithout frozen histopathology. J Invest Dermatol 2001;117:1137—43.

[24] Curiel-Lewandrowski C, Williams CM, Swindells KJ, TahanSR, Astner S, Frankenthaler RA, et al. Use of in vivo confocalmicroscopy in malignant melanoma. Arch Dermatol 2005;140:1127—32.

[25] Sauermann K, Gambichler T, Wilmert M, Rotterdam S. Inves-tigation of basal cell carcinoma [correction of carcionoma]by confocal laser scanning microscopy in vivo. Skin ResTechnol 2002;8:141—7.

[26] Marghoob AA, Charles CA, Busam KJ, Rajadhyaksha M, Lee G,Clark-Loeser L, et al. In vivo confocal scanning laser micro-scopy of a series of congenital melanocytic nevi suggestiveof having developed malignant melanoma. Arch Dermatol2005;141:1401—12.

[27] Pellacani G, Cesinaro AM, Seidenari S. In vivo confocalreflectance microscopy for the characterization of melano-cytic nests and correlation with dermoscopy and histology.Br J Dermatol 2005;152:384—6.

[28] Vannier MW, Marsh JL, Warren JO. Three dimensional CTreconstruction images for craniofacial surgical planning andevaluation. Radiology 1984;150:179—84.

[29] Ono I, Gunji H, Suda K, Kaneko F. Method for preparing anexact size model using helical volume scan computed tomo-graphy. Plast Reconstr Surg 1994;93:1363—71.

[30] Ono I, Abe K, Shiotani S, Hirayama Y. Producing a full-scalemodel from computed tomography data with the rapidprototyping technique using the Binder Jet method, -acomparison with the laser lithography method using a dryskull. J Craniomaxillofac Surg 2000;11:527—37.

[31] Gonzalez S, Tannous Z. Real-time, in vivo confocal reflec-tance microscopy of basal cell carcinoma. J Am Acad Der-matol 2002;47:869—74.