Bioheat Transfer Analysis of Cryogen SprayCooling During Laser Treatment of Port

Wine StainsT. Joshua Pfefer, PhD,1* Derek J. Smithies, PhD,2 Thomas E. Milner, PhD,1,2

Martin J.C. van Gemert, PhD,3 J. Stuart Nelson, MD, PhD,2 and Ashley J. Welch, PhD1

1Biomedical Engineering Program, The University of Texas at Austin, Austin, Texas, 787122Beckman Laser Institute and Medical Clinic, University of California, Irvine, California, 92715

3Laser Center, Academic Medical Center, Amsterdam, 1105 AZ, The Netherlands

Background and Objective: The thermal response of port wine stain(PWS) skin to a combined treatment of pulsed laser irradiation andcryogen spray cooling (CSC) was analyzed through a series of simula-tions performed with a novel optical-thermal model that incorporatesrealistic tissue morphology.Study Design/Materials and Methods: The model consisted of (1) a three-dimensional reconstruction of a PWS biopsy, (2) a Monte Carlo opticalmodel, (3) a finite difference heat transfer model, and (4) an Arrheniusthermal damage calculation. Simulations were performed for laserpulses of 0.5, 2, and 10 ms and a wavelength of 585 nm. Simulated cryo-gen precooling spurts had durations of 0, 20, or 60 ms and terminated atlaser onset. Continuous spray cooling, which commenced 60 ms beforelaser onset and continued through the heating and relaxation phases,was also investigated.Results: The predicted response to CSC included maximal pre-irradiation temperature reductions of 27°C at the superficial surfaceand 12°C at the dermoepidermal junction. For shorter laser pulses (0.5,2 ms), precooling significantly reduced temperatures in superficial re-gions, yet did not effect superficial vessel coagulation. Continuous cool-ing was required to reduce significantly thermal effects for the 10-mslaser pulse.Conclusions: For the PWS morphology and treatment parameters stud-ied, optimal damage distributions were obtained for a 2-ms laser pulsewith a 60-ms precooling spurt. Epidermal and vascular morphology aswell as laser pulse duration should be taken into account when plan-ning CSC/laser treatment of PWS. Our novel, realistic-morphology mod-eling technique has significant potential as a tool for optimizing PWStreatment parameters. Lasers Surg. Med. 26:145–157, 2000.© 2000 Wiley-Liss, Inc.

Key words: bioheat transfer; cryogen spray cooling; laser-tissue interaction; lightpropagation; Monte Carlo; numerical modeling; port wine stains; ther-mal damage

INTRODUCTION

Ideal treatment of port wine stains (PWS)and other cutaneous vascular lesions requires se-lective thermal necrosis of blood vessels withminimal damage to the normal overlying epider-mis and adjacent dermis [1]. Past research intolaser treatment of PWS has often involved inves-tigations of alternate laser parameters to reducecollateral damage [2]. Continuous wave lasers

Contract grant sponsor: Office of Naval Research Free Elec-tron Laser Biomedical Science Program; Contract grant num-ber: N00014-91-J-1564; Contract grant sponsor: Albert W.and Clemmie A. Caster Foundation; Contract grant sponsor:Whitaker Foundation; Contract grant number: WF-21025;Contract grant sponsor: National Institutes of Health; Con-tract grant number: AR43419.

*Correspondence to: T. Joshua Pfefer, PhD, Wellman Labo-ratories of Photomedicine, Massachusetts General Hospital,Bartlett 722, 55 Fruit Street, Boston, MA 02114.E-mail: jpfefer@helix.mgh.harvard.edu

Accepted 14 October 1999

Lasers in Surgery and Medicine 26:145–157 (2000)

© 2000 Wiley-Liss, Inc.

such as the argon, which produced extensive re-gions of dermal and epidermal coagulation, havebeen replaced by pulsed lasers to limit thermaldiffusion during irradiation [3].

Surface cooling has proven effective for re-ducing collateral thermal injury during vesselphotocoagulation in skin [4–6]. Because the epi-dermis contains the chromophore melanin, lightabsorption can result in the generation of hightemperatures that lead to necrosis and undesireddyspigmentation or scarring. By minimizing su-perficial temperatures and the duration of high-temperature exposure, thermal injury in the epi-dermis can be reduced or prevented. Epidermalcooling enables the use of higher radiant expo-sures and, thus, deeper coagulation of cutaneousvessels. Cooling may also reduce the temperaturein superficial blood vessels, decreasing the risk ofablation-induced vessel rupture.

Various skin cooling techniques such as ice[4,5,7], a water-chilled, transparent plate [8], di-rect water irrigation [5], and water spray [9] havebeen investigated for use in treatment of vascularlesions. These methods tend to require long expo-sure durations due to the low surface heat trans-fer; however, they are suitable for achieving tem-perature reduction deep within the tissue. A tech-nique that has been shown to be more appropriatefor spatially selective cooling of superficial skinregions is cryogen spray cooling (CSC).

In the 1950s, pressurized sprays of cryogen(low boiling point refrigerants such as freon) wereimplemented to prepare the skin for dermabra-sion procedures [10]. In 1983, Welch et al. [5] in-vestigated the use of a freon gas spray to reducethermal damage during irradiation of porcineskin in vivo with continuous wave argon andNd:YAG laser irradiation. Although this first ap-plication of spray cooling to photocoagulation ofcutaneous vessels was not optimized, results in-dicated that it could provide significant reductionof epidermal and dermal coagulation in compari-son to noncooled treatment or direct precoolingwith ice.

Treatments involving CSC have combinedspurts of liquid, nonchlorofluorocarbon refriger-ants (from tens to hundreds of milliseconds in du-ration) with laser irradiation (l 4 585 nm, tp ∼0.5 ms) [6,11]. These studies have involved therelease of a refrigerant such as tetrafluoroethane(C2H2F4; boiling point 4 −26.2°C) from a pressur-ized (∼5 atm) canister [11]. A mixture of cryogenand ice accumulate in a thin layer (10–20 mm) onthe skin surface. CSC begins to reduce skin tem-

perature within a millisecond of spray onset andthe ice/cryogen layer may remain on the skin forhundreds of milliseconds after a 100 ms spurt[12]. Heat transferred from the skin causes thecryogen to evaporate, creating a strong convectiveeffect [13]. Highly transient cooling of the super-ficial skin regions can lead to surface temperaturereductions of 30–40°C for cryogen spurt durations(Dcsc) of 5–100 ms [14,15]. These studies have in-dicated that CSC allows the use of increased ra-diant exposures without epidermal injury, thusallowing deeper coagulation of blood vessels andexpediting PWS clearance.

THEORETICAL MODELS OF CRYOGENSPRAY COOLING

The thermal mechanisms involved in thishighly transient process remain only partially un-derstood. Two types of models have been used toinvestigate CSC: those that simulated coolingalone [14,16] and those that simulated cooling inconjunction with laser heating [15,17]. Thesemodels have been one-dimensional (1-D) andsimulated tissue morphology by using homoge-neous layers. The usefulness of layered geometrymodels may be limited when attempting to simu-late thermal transients in PWS skin, becausePWS morphology is highly nonhomogeneous inthree dimensions. Specifically, these models werenot able to account for the unique optical andthermal effects in and around discrete blood ves-sels or predict the spatial distribution of thermaldamage. Previous models have also assumed ho-mogeneous thermal and damage propertiesthroughout the skin, accounted for the effects oflight scattering by using simple approximations,and neglected the effect of heat transfer duringthe laser pulse [14,15,17]. Previous simulations ofCSC/laser treatment have only been performedfor tp ∼ 0.5 ms, whereas clinical treatment in-creasingly uses lasers with pulse durations of1–10 ms. One previous study investigated the useof liquid nitrogen and water cooling during lasertreatment of PWS by using a 2-D, layered-geometry, optical-thermal numerical model [18].

According to Anvari et al., “. . . the optimumcooling strategy needs to be determined on [an]individual patient basis as density of melano-somes as well as depth of the PWS blood vesselsvary considerably” [15]. Therefore, modeling-based, patient-specific treatment planning maybe useful in optimizing cooling parameters. Addi-tionally, the use of true PWS morphology in simu-

146 Pfefer et al.

lating the effects of CSC provides a unique per-spective on basic optical and thermal laser-tissueinteractions during PWS treatment.

CSC is a novel and effective adjunct to PWStreatment. However, to achieve optimal applica-tion of CSC, it is important to know how thermaltransients and the extent of protection againstthermal injury vary with cryogen spurt durationand laser parameters. The purpose of this study isto investigate these basic mechanisms and to as-sess mathematically the effect of specific CSCspurt durations on irreversible damage in PWSskin components. This has been accomplished bysimulating PWS laser treatment (l 4 585 nm; tp4 0.5, 2, 10 ms) in conjunction with CSC (precool-ing) spurts (Dcsc 4 20, 60 ms) and continuousspray cooling.

MATERIALS AND METHODS

The modeling approach used in this study isnearly identical to that described and reported ina previous study of laser treatment of PWS with-out cooling [19]. As depicted in Figure 1, themodel includes a tissue grid that has been gener-ated from a PWS biopsy, which provides informa-tion about local tissue type (epidermis, dermis, orblood) and, thus, material property data for thelight propagation, heat transfer, and thermaldamage modeling components.

The high-resolution computer-based PWS bi-opsy reconstruction was described previously bySmithies et al. [20]. A portion of the reconstructedvolume measuring 268 mm × 312 mm × 460 mm(x,y,z, where z is depth) was used in the presentstudy. This material grid included 52 x–z sectionsof 134 × 230 voxels each, resulting in a total ofapproximately 1.6 million rectangular voxels (2mm × 6 mm × 2 mm). The material grid was iden-tical in size and dimension to the grids used forenergy absorption, temperature, and damage dis-tributions.

The 3-D, arbitrary morphology Monte Carlomodel used to simulate light propagation in thisstudy has been described and verified previously[21] and implemented by using the aforemen-tioned PWS biopsy reconstruction [19,22]. Theprimary optical properties — absorption coeffi-cient (ma), scattering coefficient (ms), scatteringanisotropy (g) and index of refraction (n) — usedin the present simulations are displayed in Table1. A single optical simulation is required to calcu-late the energy deposition rate distribution for anominal irradiance of 1 W/cm2. This distributionis then scaled linearly with irradiance to producethe source term (S) for each thermal simulation.

The explicit finite difference technique isused to solve the 3-D transient form of the Fourierheat conduction equation [13]:

rc­T­t

=­

­xSk­T­xD +

­

­ySk­T­yD +

­

­zSk­T­zD + S (1)

where T is temperature (°C), t is time (seconds), Sis the heat source term (W/m3), r is density (kg/m3), c is specific heat (J/kg/°C), and k is thermalconductivity (W/m/°C). The bioheat equation isnot used because of the very short time scalesand, thus, negligible perfusion effects involved.Adiabatic boundary conditions are applied to thefive internal tissue boundaries, whereas a stan-dard Robin convection condition is used to de-scribe heat flow between the superficial tissuelayer and the environment. A uniform initial skintemperature of 30°C was assumed. The thermalproperties of skin used in these simulations arelisted in Table 2 [23].

The primary advantage of the simulationsperformed for this study lies in the combination ofrealistic tissue morphology with a surface bound-ary condition that simulates CSC. In general, pro-cesses involving convection with phase changesuch as the boiling of a cryogen can be approxi-mated by using a convection coefficient (h) of 2.5

Fig. 1. Flow chart of the modeling process. PWS, port wine stain.

Cryogen Spray Cooling During PWS Treatment 147

to 100 kW/m2/°C [13]. For the present study, theconvection coefficient (4 kW/m2/°C) and ambienttemperature (−7°C) values from a previous inves-tigation of CSC are implemented [16].

The temporal progression of CSC/laser treat-ment of PWS was simulated in three phases: (1) aprecooling phase, involving cryogen cooling (andno laser irradiation) for a duration (Dcsc) of 0, 20or 60 ms; (2) a heating phase, involving an adia-batic surface and laser irradiation for a durationof (tp) 0.5, 2 or 10 ms; and (3) a relaxation phase,involving an adiabatic surface and no laser irra-diation until temperatures relaxed to below 60°C(at which point the simulation is ended becausethe damage accumulation rate is minimal). Theassumption that cooling terminates at the end ofthe CSC pulse is a first approximation. A residualice/cryogen layer may provide post-spray cooling[9,12] although the significance of this effect islikely dependent on spray duration, volume ofcryogen released, and other variables. The “con-tinuous cooling” simulation incorporated a secondtemporal structure in which the convectiveboundary was implemented starting 60 ms beforelaser onset and continuing through a 10-ms heat-ing phase and subsequent relaxation to below60°C.

The Arrhenius rate process equation is usedto calculate thermal damage:

V~t! = A*0

texpF1

Ea

RT~t!Gdt (2)

where A is the frequency factor or molecular col-lision rate (1/s), Ea is activation energy (J/mole),

and R is the universal gas constant (8.314 J/mole/K) [24,25]. The thermal damage calculation isimplemented as a subroutine in the thermal dif-fusion model. Two sets of damage coefficients (A,Ea) are used: one for blood (hemoglobin [26]) andanother for epidermis and dermis (bulk skin [27]).These values are displayed in Table 3 and Figure2 for comparison with other relevant data fromthe literature.

The thermal response of tissue to irradiationat a specific radiant exposure or irradiance levelis highly dependent on laser pulse duration. Thisis seen in the use of higher radiant exposure lev-els for longer laser pulse PWS treatments. To pre-sent more comparable situations, we have alteredthe radiant exposure based on the tissue thermalresponse, by using a temperature-based end pointin a manner similar to previous theoretical stud-ies [3,28,29]. For each of the three non-CSC cases(Dcsc 4 0 ms and tp 4 0.5, 2, 10 ms), a radiantexposure was found that produced a maximumend of laser pulse temperature of 150°C (Table 4).This radiant exposure was then used for the cor-responding cooling simulations (Dcsc 4 20, 60 ms)for that laser pulse duration. This temperaturewas chosen because it ensured that significantdamage would be produced in the region of theblood vessels without exceeding ablation-leveltemperatures reported for laser-irradiated tissue[30,31].

RESULTS

The predicted end of laser pulse tempera-tures and final (post-relaxation phase) coagulationdistributions are presented in Figures 3 and 4,respectively. Although results were computed forthe entire 3-D tissue volume, data are presentedhere for a single 2-D section. In Figures 3 and 4,data are presented for pulse durations of 0.5, 2,and 10 ms for the uncooled (Dcsc 4 0) and pre-cooled (Dcsc 4 20, 60 ms) cases.

The temperature distributions immediatelyafter irradiation (Figure 3) show a decreasinglevel of thermal confinement with increasing la-ser pulse duration. For tp 4 0.5 ms, large tem-perature gradients were predicted at the edge ofthe epidermis and blood vessels, whereas forlonger pulse durations, these gradients decreasedand perivascular heating increased. As tp length-ened, heat transfer from blood regions necessi-tated the use of higher radiant exposures to reachan end-point temperature of 150°C. As a result,epidermal temperatures increased with laser

TABLE 1. Optical Properties of Skin Components atl = 585 nm*

ma (cm−1) ms (cm−1) g n

Epidermis 18 470 0.79 1.37Dermis 2.4 129 0.79 1.37Blood 191 467 0.99 1.33

*From van Gemert et al. [3].

TABLE 2. Thermal Properties of Skin Components*

k (W/m/°C) r (kg/m3) c (J/kg/°C)

Epidermis 0.21 1200 3600Dermis 0.53 1200 3800Blood 0.55 1100 3600

*From Duck [23].

148 Pfefer et al.

pulse duration until they approached maximumvessel temperatures.

The effect of tp on the final (post-relaxationphase) distribution of thermal injury is shown inFigure 4. The duration of the relaxation phase(from the end of irradiation until the maximumtemperature fell to below 60°C and the simulationwas terminated) increased with laser pulse dura-tion from less than 1 ms for tp 4 0.5 ms to hun-dreds of milliseconds for tp 4 10 ms. Damage dis-tributions show some correlation with the post-irradiation temperatures shown in Figure 3.Coagulation was primarily limited to the bloodvessels for tp 4 0.5 ms, whereas longer pulse du-rations produced wider regions of perivascular co-agulation. Similarly, an increase in pulse dura-tion produced an increase in extent of epidermalcoagulation. For tp 4 10 ms, heat generated inthe epidermis and blood diffused into the dermis

during and after the laser pulse, causing regionsof epidermal and perivascular coagulation to de-velop into widespread necrosis.

The most obvious result of increased cryogenspurt duration was the elimination of epidermalcoagulation for tp 4 2 ms, from the noncooled caseto the Dcsc 4 20 ms case. The thickness of peri-vascular damage in superficial regions was re-duced by only a few microns from the uncooled toDcsc 4 60 ms case. Cooling reduced temperaturesand thermal injury only in superficial regions fortp 4 0.5 or 2 ms, whereas for tp 4 10 ms, it re-duced the extent of deeper dermal coagulation. Itis likely that much of the dermal injury in thenoncooled case was caused by heat that was gen-erated in the epidermis and superficial vesselsand diffused into deeper regions after the end ofthe laser pulse.

The predicted effect of CSC on end of precool-ing phase temperatures is presented in Figure 5for durations of 20 ms (black lines) and 60 ms(gray lines). To investigate the effect of spatialvariations in thermal properties, two sets of simu-lations were performed: one in which heterog-enous tissue properties were used (thick lines),and one in which the thermal properties of dermiswere implemented for all three tissue types (thinlines). In the heterogenous case, 20 ms of coolingcaused epidermal temperatures to be reduced by7–23°C, whereas a 60-ms CSC spurt caused re-ductions of 12–27°C. The effect of the 60 ms CSCspurt on temperatures below the dermoepidermaljunction was more than double that produced bythe 20-ms spurt. The magnitude of temperaturereduction and gradients in the epidermis weresignificantly reduced for the homogenous thermalproperties case due to the higher values of ther-mal conductivity and diffusivity in the epidermis.The increased rate of heat conduction led tohigher surface temperatures but lower tempera-tures at and below the dermoepidermal junctionfor the heterogeneous case. The magnitude ofthese effects increased with cooling duration.

TABLE 3. Rate Process Coefficients*

A (1/s) Ea (J/mole) Tth (°C) Reference

1. Hemoglobin 7.6 × 1066 455,000 93 [26]2. Bovine serum albumin 3.2 × 1056 379,000 90 [37]3. Aorta 5.6 × 1063 430,000 91 [38]4. Bulk skin 7.6 × 1076 550,000 74 [27]5. Bulk skin 3.1 × 1098 628,000 67 [39]

*In the present study, hemoglobin [26] data were used for blood regions andbulk skin [27] for dermis and epidermis. Threshold temperature for coagu-lation (Tth) is listed for a constant temperature exposure duration of 10 ms.

Fig. 2. Graph of time-temperature relationship defined byArrhenius equation and rate coefficients assuming step tem-perature changes and well-defined exposure durations.

TABLE 4. Laser Parameters to Produce a MaximumTemperature of 150°C Without Cooling

Pulse duration (ms) 0.5 2.0 10.0Irradiance (W/cm2) 6900 3025 1300Radiant exposure (J/cm2) 3.45 6.05 13.00

Cryogen Spray Cooling During PWS Treatment 149

Fig. 3. Temperature distributions immediately after the end of the laser pulse (arrow indicates location of data used in Figs.5–9). Region shown is 312 mm × 460 mm (width × depth). Radiant exposures are listed in Table 4. (a) tp 4 0.5 ms, no cooling;(b) tp 4 2 ms, no cooling; (c) tp 410 ms, no cooling; (d) tp 4 0.5 ms, Dcsc 4 20 ms; (e) tp 4 2 ms; Dcsc 4 20 ms; (f) tp 410ms Dcsc 4 20 ms; (g) tp 4 0.5 ms, Dcsc 4 60 ms; (h) tp 4 2 ms, Dcsc 4 60 ms; (i) tp 410 ms, Dcsc 4 60 ms. Arrow in a indicateslocation of data in Figures 6–8.

150 Pfefer et al.

Fig. 4. Distribution of native (light gray) and coagulated (dark gray) tissue after the end of the relaxation phase. Region shownis 312 mm × 460 mm (width × depth). Radiant exposures are listed in Table 4. (a) tp 4 0.5 ms, no cooling; (b) tp 4 2 ms, nocooling; (c) tp 410 ms, no cooling; (d) tp 4 0.5 ms, Dcsc 4 20 ms; (e) tp 4 2 ms; Dcsc 4 20 ms; (f) tp 410 ms Dcsc 4 20 ms;(g) tp 4 0.5 ms, Dcsc 4 60 ms; (h) tp 4 2 ms, Dcsc 4 60 ms; (i) tp 410 ms, Dcsc 4 60 ms.

Cryogen Spray Cooling During PWS Treatment 151

A more quantitative documentation of theend of heating phase temperature distributionsfor a specific data set (indicated by the arrow inFigure 3a) is presented in Figures 6–8. The mostsignificant CSC-induced temperature reductions(compared with noncooled simulations) occurredwithin 0.1 mm of the surface. The data in Figure7 indicate that the CSC-induced epidermal spar-ing seen for the tp 4 2-ms case (Fig. 4) was due todecrease in epidermal temperatures from 80 to60°C. The only cases for which temperature in-creased significantly with depth across the epi-dermis were tp 4 0.5 ms and cooling durations of20 and 60 ms. This gradient was a remnant fromthe end of the precooling phase (Fig. 5) but hadbeen reduced in magnitude by heat conductionduring the laser pulse. Although temperature re-duction at the surface varied significantly withpulse duration (25°C reduction for tp 4 0.5 ms;15°C for tp 4 10 ms), at a depth of 0.05 mm thisdiscrepancy was greatly reduced (12 and 13°C).

The results of a simulation involving con-tinuous cooling are presented in Figures 8 and 9.In this case, CSC began 60 ms before the onset ofa 10-ms laser pulse and continued until the veryend of the simulation (the end of the relaxationphase). This provided a temperature gradient atthe superficial surface before, during, and afterthe laser pulse, resulting in a significant reduc-tion in superficial temperatures and thermal in-jury compared with the Dcsc 4 60-ms case. Al-though continuous cooling resulted in a furtherreduction in temperature of nearly 60°C near the

skin surface, the difference near the dermoepider-mal junction was only about 5°C. As a result, onlyepidermal regions were spared (including regionscorresponding to Fig. 8 data). Note that the regionspared contains ridges, thus a higher local surfacearea and an increased rate of cooling.

Because thermal damage is a rate process,and, thus, a function of time and temperature,transient thermal behavior is a critical issue. Sur-face temperature dynamics are presented in Fig-ure 10 for two cases: tp 4 2 ms with no coolingand tp 4 2 ms with Dcsc 4 60 ms. This figureindicates that with or without CSC, temperaturesare predicted to decrease at a moderate rate overthe first hundred milliseconds after the laserpulse. The initial, relatively constant tempera-ture level in the CSC case was likely due to somediffusion of heat from deeper regions of the epi-dermis toward the surface.

DISCUSSION

Optical-Thermal Mechanisms

One of the benefits of an arbitrary-morphol-ogy model is the ability to account for realisticheterogeneity in thermal properties. Due to thelow water content of the epidermis, its thermalconductivity (and possibly diffusivity) is lowerthan that of dermis [23]. As a result, the effect ofthe surface boundary is more confined than if ho-mogeneous tissue properties were assumed. Thiseffect is apparent in Figure 5, where the hetero-geneous tissue case resulted in lower surface tem-peratures than the homogeneous (dermis thermalproperties) case and higher temperatures at loca-tions deeper than the dermoepidermal junction.Low thermal conductivity also causes heat gener-ated in the epidermis during irradiation to re-main in that region, an effect that was exacerbat-ed by the lack of heat loss at the surface after theprecooling phase, often resulting in extensivedamage in or near the epidermis.

Although the crossover points that occurnear the dermoepidermal junction in Figure 5seem to be a result of an energy balance, this isnot strictly true. In fact, the lower surface tem-peratures in the heterogeneous case produced aweaker temperature differential and reduced con-vective losses at the surface as well as the totalquantity of energy removed. Because the most su-perficial regions of the epidermis have the leastwater content, removal of the stratum corneummay decrease the resistance to heat flow and im-

Fig. 5. Comparison of temperature distributions immediatelyafter precooling phase for heterogeneous (thick lines) and ho-mogeneous (thin lines) thermal properties. Cryogen spurts of20 ms (black lines) and 60 ms (lines) were simulated. Dermo-epidermal junction is at a depth of 0.05 mm.

152 Pfefer et al.

prove the ability of cryogen to remove heat fromepidermal and superficial dermal regions.

Although CSC reduces temperaturesthroughout the epidermis, end of precoolingphase temperature distributions are highly non-uniform across this layer (Fig. 5). For short laserpulses, the post-irradiation temperature distribu-tion across the epidermis maintains this variation(Fig. 6). Because the basal layer is the deepestepidermal layer (epidermal rete pegs may extendinto the dermis to a depth of 100 mm) and has thehighest melanin content, the subsurface tempera-tures during CSC/laser treatment may be evengreater than predicted. A more realistic, multi-layer representation of epidermal morphology, toaccount for variations in both optical and thermalproperties, may yield more accurate predictions[17,32].

Transient surface temperatures (Fig. 10) in-dicated a greater thermal relaxation rate for theuncooled case (21°C over 100 ms) than for Dcsc 460 ms (8°C over 100 ms). This finding was becausethe post-irradiation temperature difference be-tween the epidermis and superficial dermis wasreduced when the surface was cooled (Fig. 7).Such a result is in apparent disagreement withprevious experimental data, which indicated thatfor cooled tissue, the surface temperature in-creased dramatically during the laser pulse (tp 40.45 ms), and returned to initial (baseline) tem-perature within milliseconds [6]. The discrepancyin results may be explained by recent studies inwhich it was demonstrated that for cooling dura-tions of 50–100 ms, the duration of the coolingoccurred was tens to hundreds of millisecondslonger than the spray duration [9,12] because of aresidual cryogen/ice layer on the tissue surface.Numerical studies have simulated CSC pulses asbeing uniform in intensity over time, and stoppedcooling at laser onset [15,17] or continued it un-changed during and after the laser pulse [11]. Inthe future, boundary conditions used to simulateCSC may need to be altered to account for re-sidual cooling, which tapers off after the end of acryogen spurt.

Several previous studies have incorporatedradiometric surface temperature measurementduring CSC. In one study, surface temperatureswere shown to be reduced by 30–40°C for spurtdurations between 5 and 80 ms [14]. A more re-cent investigation indicated a temperature reduc-tion of 25–40°C for pulse durations of 20–100 ms[16]. Our model indicated post-irradiation surfacetemperature reductions of 24°C for Dcsc 4 20 ms

Fig. 8. Temperature distributions immediately after the endof the laser pulse for tp 4 10 ms (Ho 4 13.00 J/cm2) and Dcsc4 0, 20, 60 ms, and continuous cooling simulation (CSC be-gan 60 ms before 10 ms laser pulse and continued throughheating and relaxation phases).

Fig. 6. Temperature distributions immediately after the endof the laser pulse for tp 4 0.5 ms (Ho 4 3.45 J/cm2) and Dcsc4 0, 20 and 60 ms. The letters E, D, and B indicate thelocation of epidermis, dermis and blood regions, respectively.

Fig. 7. Temperature distributions immediately after the endof the laser pulse for tp 4 2 ms (Ho 4 6.05 J/cm2) and Dcsc 40, 20, and 60 ms.

Cryogen Spray Cooling During PWS Treatment 153

and 27°C for Dcsc 4 60 ms. The discrepancy be-tween these data sets may be caused by variationsin temperature and thickness of the ice/cryogenlayer on the skin surface. However, radiometricdata published recently by Torres et al. [12] indi-cated better agreement with our results — a sur-face temperature reduction of 28°C at the end of a40 ms CSC spurt. In a recent in vivo study of CSCin hamster skin, temperature reductions of5–10°C were recorded for a cooling pulse of 100ms with a thermocouple located 625 mm below thetissue surface [33]. These results indicate a muchdeeper cooling effect than that seen in the presentsimulations, possibly due to significant anatomic

and physiological differences between human andhamster skin or the finite thickness of the skinpreparation.

Previous theoretical studies have predictedthe temperature response versus skin depth inresponse to CSC. One study indicated that a 20-ms cooling pulse caused a decrease of 10°C at adepth of ∼60 mm, whereas a 60 ms pulse resultedin a 10°C reduction to depths greater than 100 mm[16]. Another study indicated that post-irradiation temperatures might decrease bynearly 10°C down to a depth of 200 mm, for typicalclinical laser parameters (tp 4 0.5 ms, l 4 585nm) and CSC durations of 80–100 ms [17]. Both ofthese studies are in reasonable agreement our re-sults, although differences in laser parametersmake accurate comparisons difficult.

Optimization of Treatment Parameters

Our simulations indicated that the thresholdradiant exposure for epidermal injury at tp 4 2ms was approximately 6 J/cm2. This rate is inreasonable agreement with the finding of a previ-ous theoretical study that, in patients with type IIskin (intermediate melanin content), no coolingwas required for epidermal preservation for a ra-diant exposure (Ho) of 5 J/cm2, whereas for higherradiant exposures, a longer spurt duration wasrequired to avoid coagulation [17]. Within the lim-ited parameter space investigated in the presentstudy, the optimal pulse duration for this PWSwas found to be 2 ms and the optimal CSC spurtduration was 60 ms. This combination of param-eters should provide vascular and perivascular co-agulation to a depth of approximately 300 mm,without epidermal coagulation. The authors arenot aware of any previous study that has theoret-ically or experimentally evaluated PWS treat-ment by using this combination of laser and cool-ing parameters; however, laser pulse durations onthe order of 2 ms have been identified as optimalfor noncooled treatment in several previous stud-ies [3,19,34–36].

Our results indicate that tp not only has asignificant influence on vessel coagulation, but onthe efficacy of CSC for epidermal sparing as well.To produce a specific maximum temperature in asmall blood vessel (such as the 20- to 30-mm di-ameter vessels typical of the PWS used in thisstudy), an increase in tp must typically be accom-panied by an increase in radiant exposure. How-ever, due to the air-tissue (adiabatic) boundarycondition and low thermal conductivity in the epi-dermis, epidermal temperatures are approxi-

Fig. 9. Distribution of native (light gray) and coagulated(dark gray) tissue after the end of the relaxation phase forcontinuous cooling simulation.

Fig. 10. Transient temperature distributions at the epider-mal surface for tp 4 2 ms with no cooling (dashed line); andtp 4 2 ms with Dcsc 4 60 ms (gray solid line).

154 Pfefer et al.

mately proportional to radiant exposure. The im-pact of CSC is reduced in the epidermis as laserpulse duration increases because the cooling ef-fect becomes more distributed within the tissue(Figs. 6–8), causing a reduction in the extent ofcoagulation in deeper regions (Fig. 4c,f,i). Thus,safe CSC/laser treatments may be more difficultto achieve when pulse durations of 10 ms orgreater are used. However, it must be noted thatthe apparent superiority of shorter laser pulsetreatment may be reduced or eliminated whentreatment involves irradiation of blood vesselswith more typical PWS diameters of 50–120 mm.

Figure 8 also documents the results of asimulation involving continuous cooling (includ-ing 60 ms of pre-irradiation cooling), which wasused to investigate a possible methodology forminimizing the excessive temperatures and co-agulation seen for tp 4 10 ms. Continuous coolingproduced a skin surface temperature at least 55°Cless than the other CSC cases at tp 4 10 ms.However, by a depth of about 70 mm, the differ-ence in temperature between Dcsc 4 60 ms andthe continuous case was minimal (4°C). Thus,only part of the most superficial epidermal re-gions remained native. Although these resultsseem to indicate that continuous or postcoolingwould have little effect on tissue damage, it isworthwhile to consider that for longer pulse du-rations, larger tissue volumes reached tempera-tures at or exceeding the threshold for coagula-tion, which led to longer thermal relaxation times.As indicated by Figure 2, if the exposure durationincreases, then the threshold temperature be-comes reduced and coagulation may occur. There-fore, post-irradiation cooling may be most impor-tant for longer laser pulse durations.

The use of a maximum temperature endpoint (150°C) in this study was problematic inthat it produced a 10-ms pulse duration case inwhich extensive coagulation was predicted. Thisoutcome did not represent the best possible treat-ment result. This result is interesting becauseprevious studies that have used threshold tem-perature-based end points have typically not pre-sented information on collateral damage [3,28,29]. In future studies, it may be more appropriateto consider coagulation of a specified percentageof vessel (or vessel wall) regions as the end point,based on the Arrhenius equation.

Our results indicate that to more accuratelyestimate the thermal mechanisms in the epider-mis, it may be important to account for the varia-tion in absorption across its layers. Because a sig-

nificant portion of epidermal melanin is containedin deeper layers, which can be 50–100 mm deepand vary from patient to patient [3], it may beworthwhile to investigate a variety of epidermalmorphologies. Computational investigation ofseveral combinations of precooling and postcool-ing durations and delay periods may also be use-ful in identifying optimal treatment parametersfor a specific PWS anatomy.

CONCLUSIONS

A 3-D optical-thermal numerical model hasbeen implemented to study the effect of CSC onsimulated laser treatment of PWS by using tissuemorphology specified from a true PWS biopsy. Re-sults were obtained for laser pulse durations of0.5, 2, and 10 ms (l 4 585 nm) in conjunction with20- and 60-ms-long CSC (precooling) spurts aswell as noncooled irradiations. The use of continu-ous cooling during irradiation was also simulated.

The most important findings of this studyinclude (1) The effect of CSC cooling was confinedto the most superficial 100–200 mm, particularlyfor shorter laser pulses; (2) As the cryogen spurtduration was increased from 20 ms to 60 ms, thepre-irradiation temperature reduction at the der-moepidermal junction increased from 7°C to 12°C;(3) Superficial skin temperatures were signifi-cantly effected by the low thermal conductivity ofthe epidermis; (4) Of the treatment parametersinvestigated in this study, the optimal laser pulselength and CSC spurt duration (for this specificPWS morphology) were 2 ms and 60 ms, respec-tively; and, (5) CSC was most effective for epider-mal temperature reduction during shorter laserpulses, but more necessary for longer laser pulsedurations.

When planning CSC/laser treatment ofPWS, tissue morphology and laser pulse durationshould be taken into account. The results of thisstudy indicate that our novel modeling techniquehas significant promise as a tool for improvingunderstanding of laser-tissue interactions inhighly nonhomogeneous tissue and optimizing la-ser and cooling parameters on an individual pa-tient basis.

ACKNOWLEDGMENT

A.J. Welch is the Marion E. Forsman Profes-sor of Electrical and Computer Engineering.

Cryogen Spray Cooling During PWS Treatment 155

REFERENCES

1. Anderson RR, Parrish JA. Selective photothermolysis:precise microsurgery by selective absorption of pulsed ra-diation. Science 1983;220:524–527.

2. Anderson RR. Laser-tissue interactions. In: GoldmanMP, Fitzpatrick RE, editors. Cutaneous laser surgery:the art and science of selective photothermolysis. St.Louis: Magby; 1994.

3. van Gemert MJC, Welch AJ, Pickering JW, Tan OT. La-ser treatment of port wine stains. In: Welch AJ, van Ge-mert MJC, editors. Optical-thermal response of laser-irradiated tissue. New York: Plenum Press; 1995.

4. Gilchrest BA, Rosen S, Noel JM. Chilling port wine stainsimproves the response to argon laser therapy. Plast Re-constr Surg 1982;69:278–283.

5. Welch AJ, Motamedi M, Gonzalez A. Evaluation of cool-ing techniques for the protection of the epidermis duringNd-YAG laser irradiation of the skin. In: Joffe SN, editor.Neodymium-YAG laser in medicine and surgery. NewYork: Elsevier; 1983.

6. Nelson JS, Milner TE, Anvari B, Tanenbaum BS, KimelS, Svaasand LO, Jacques SL. Dynamic epidermal coolingduring pulsed laser treatment of port-wine stain. ArchDermatol 1995;131:695–700.

7. Hohenleutner U, Walther T, Wenig M, Baumler W, Land-thaler M. Leg telangiectasia treatment with a 1.5 mspulsed dye laser, ice cube cooling of the skin and 595 vs600 nm: preliminary results. Lasers Surg Med 1998;23:72–78.

8. Chess C, Chess Q. Cool laser optics treatment of largetelangiectasia of the lower extremities. J Dermatol SurgOncol 1993;19:74–80.

9. Exley J, Dickinson M, King T, Charlton A, Falder S, Ke-nealy J. Comparison of cooling criteria with a cryogenspray and water/air spray. Proc SPIE 1999;3601:130–140.

10. Strauss JS, Kligman AM. Observations on dermabrasionand the anatomy of the acne pit. Arch Dermatol 1956;74:397–404.

11. Nelson JS, Milner TE, Anvari B, Tanenbaum BS,Svaasand LO, Kimel S. Dynamic epidermal cooling inconjunction with laser-induced photothermolysis of portwine stain blood vessels. Lasers Surg Med 1996;19:224–229.

12. Torres JH, Anvari B, Tanenbaum BS, Milner TE, Yu JC,Nelson JS. Internal temperature measurements in re-sponse to cryogen spray cooling of a skin phantom. ProcSPIE 1999;3590:11–19.

13. Incropera FP, DeWitt DP. Fundamentals of heat andmass transfer. 4th ed. New York: John Wiley & Sons;1996.

14. Anvari B, Milner TE, Tanenbaum BS, Kimel S, SvaasandLO, Nelson JS. Selective cooling of biological tissues: ap-plication for thermally mediated therapeutic procedures.Phys Med Biol 1995;40:241–252.

15. Anvari B, Milner TE, Tanenbaum BS, Kimel S, SvaasandLO, Nelson JS. Dynamic epidermal cooling in conjunctionwith laser treatment of port-wine stains: theoretical andpreliminary clinical evaluations. Lasers Med Sci 1995;10:105–112.

16. Anvari B, Milner TE, Tanenbaum BS, Nelson JS. A com-

parative study of human skin thermal response to sap-phire contact and cryogen spray cooling. IEEE TransBiomed Eng 1998;45:934–941.

17. Anvari B, Milner TE, Tanenbaum BS, Kimel S, SvaasandLO, Nelson JS. A theoretical study of the thermal re-sponse of skin to cryogen spray cooling and pulsed laserirradiation: implications for treatment of port wine stainbirthmarks. Phys Med Biol 1995;40:1451–1465.

18. Sturesson C, Andersson-Engels S. Mathematical model-ling of dynamic cooling and pre-heating used to increasethe depth of selective damage to blood vessels in lasertreatment of port wine stains. Phys Med Biol 1996;41:413–428.

19. Pfefer TJ, Barton JK, Smithies DJ, Milner TE, Nelson JS,van Gemert MJC, Welch AJ. Modeling laser treatment ofport wine stains with a computer-reconstructed biopsy.Lasers Surg Med 1999;24:151–166.

20. Smithies DJ, van Gemert MJC, Hansen MK, Milner TE,Nelson JS. Three-dimensional reconstruction of portwine stain vascular anatomy from serial histological sec-tions. Phys Med Biol 1997;42:1843–1847.

21. Pfefer TJ, Barton JK, Chan EK, Ducros MG, Sorg BS,Milner TE, Nelson JS, Welch AJ. A three dimensionalmodular adaptable grid numerical model for light propa-gation during laser irradiation of skin tissue. IEEE J SelTop Quantum Electron 1996;2:934–942.

22. Barton JK, Pfefer TJ, Welch AJ, Smithies DJ, Nelson JS,van Gemert MJC. Optical Monte Carlo modeling of a trueport wine stain anatomy. Optics Express 1998; 2:391–396.

23. Duck FA. Thermal properties of tissue. In: Physical prop-erties of tissue. London: Academic Press; 1990.

24. Moritz AR, Henriques FC Jr. Studies in thermal injury II.The relative importance of time and surface temperaturein the causation of cutaneous burns. Am J Pathol 1947;23:695–720.

25. Pearce J, Thomsen S. Rate process analysis of thermaldamage. In: Welch AJ, van Gemert MJC, editors. Optical-thermal response of laser-irradiated tissue. New York:Plenum Press; 1995.

26. Lepock JR, Frey HE, Bayne H, Markus J. Relationship ofhyperthermia-induced hemolysis of human erythrocytesto the thermal denaturation of membrane proteins. Bio-chim Biophys Acta 1989;980:191–201.

27. Weaver JA, Stoll AM. Mathematical model of skin ex-posed to thermal radiation. Aerosp Med 1969; 40:24.

28. de Boer JF, Lucassen GW, Verkruysse W, van GemertMJC. Thermolysis of port-wine-stain blood vessels: diam-eter of a damaged blood vessel depends on the laser pulselength. Lasers Med Sci 1996;11:177–180.

29. Kimel S, Svaasand LO, Hammer-Wilson M, Schnell MJ,Milner TE, Nelson JS, Berns MW. Differential vascularresponse to laser photothermolysis. J Invest Dermatol1994;103:693–700.

30. LeCarpentier G, Motamedi M, McMath L, Rastegar S,Welch A. Continuous wave laser ablation of tissue: analy-sis of thermal and mechanical events. IEEE TransBiomed Eng 1993;40(2):188–199.

31. Pfefer TJ, Choi B, Vargas G, McNally KM, Welch AJ.Mechanisms of pulsed laser-induced thermal coagulationof whole blood in vitro. Proc SPIE 1999; 3590:20–31.

32. Miller ID, Veitch AR. Optical modelling of light distribu-

156 Pfefer et al.

tions in skin tissue following laser irradiation. LasersSurg Med 1993;13:565–571.

33. Vargas G, Barton JK, Choi B, Izatt JA, Welch AJ. Assess-ing vessel damage with color Doppler optical coherencetomography. Proc SPIE 1999;3598:185–194.

34. Anderson RR, Parrish JA. Microvasculature can be selec-tively damaged using dye lasers: a basic theory and ex-perimental evidence in human skin. Lasers Surg Med1981;1:263–276.

35. Smithies DJ, Butler PH. Modelling the distribution oflaser light in port-wine stains with the Monte Carlomethod. Phys Med Biol 1995;40:701–731.

36. Pickering JW, Butler PH, Ring BJ, Walker EP. Com-puted temperature distributions around ectatic capillar-

ies exposed to yellow (578 nm) laser light. Phys Med Biol1989;34:1247–1258.

37. McNally KM, Optical and thermal studies of laser soldertissue repair in vitro. Ph.D. Thesis, School of Mathemat-ics, Physics, Computing and Electronics. 1998, Macqua-rie University: Sydney, Australia.

38. Agah R, Pearce JA, Welch AJ, Motamedi M. Rate processmodel for arterial tissue thermal damage: implicationson vessel photocoagulation. Lasers Surg Med 1994;15:176–184.

39. Henriques FC, Jr. Studies in thermal injury V. The pre-dictability and significance of thermally induced rate pro-cesses leading to irreversible epidermal injury. ArchPathol 1947;43:489–502.

Cryogen Spray Cooling During PWS Treatment 157