Pulsed laser-induced thermal damage in whole blood.

An investigation of the effects of laser irradiation with a wavelength of 532 nm and pulse duration of 10 ms on whole blood was performed in vitro. Threshold radiant exposures for coagulation were quantiﬁed and transient radiometric temperatures were measured. The progression of effects with increasing radiant exposure—from evaporation to coagulation-induced light scattering to aggregated coagulum formation to ablation—is described. Results indicate that coagulation and ablation occur at temperatures signiﬁ-cantly in excess of those assumed in previous theoretical studies. An Arrhenius rate process analysis based on hemoglobin data indicates good agreement with experimental results. @ S0148-0731 ~ 00 ! 00902-X #


Introduction
Thermal coagulation of blood constituents may play a significant role in a variety of therapeutic laser applications such as the selective destruction of vessels in the skin ͓1͔, retina ͓2͔, and gastrointestinal tract ͓3͔, as well as tissue welding ͓4͔ and achieving hemostasis ͓5͔.However, there remains a lack of quantitative data regarding optical and thermal effects of blood photocoagulation, particularly for laser pulses on the order of tens of milliseconds or less.
Laser-induced blood vessel necrosis, or selective photothermolysis, typically involves the absorption of light by hemoglobin ͑Hb͒, the diffusion of heat through the blood, vessel wall, and perivascular dermis, with the goal of selectively coagulating endothelial cells and/or blood constituents.In recent studies using optical coherence tomography ͑OCT͒ ͓6͔ and confocal microscopy ͓7͔, pulsed laser irradiation has been shown to produce coagula in the lumen of cutaneous vessels in vivo.As radiant exposure (H o ) levels increase, these coagula increase in size, filling the lumen and causing vessel occlusion, which can lead to necrosis.These results, and the inability of present optical-thermal models to predict them, underscore the need for a greater understanding of the effect of pulsed laser irradiation on blood.
1.1 Background.Blood plasma is comprised primarily of water ͑91 percent͒ and proteins ͑7 percent, 7.3 gm/dL͒ such as albumin ͑4.5 gm/dL͒ and globulin ͑2.5 g/dL͒ ͓8,9͔.The primary protein component of the erythrocytes is hemoglobin ͑14-16 gm/dL of whole blood͒, although a smaller quantity of proteins such as spectrin comprise the cell membranes.
Thermally induced biochemical and morphological effects in blood have been investigated previously.As blood is exposed to heat, the normally biconcave red blood cell assumes a biconvex shape, then forms a spherocyte ͓10͔.Partial membrane fragmentation occurs, resulting in a loss of hemoglobin.Finally, the membrane disintegrates into globules, which aggregate to form a mesh, and the release of hemoglobin is completed ͓11,12͔.The thermal damage to the membrane is attributed to conformational changes in erythrocyte membrane structural proteins ͓13͔.These alterations have been shown to induce an increase in scattering ͑at ϭ632 and 1064 nm͒ and absorption ͑at ϭ586 and 1064 nm͒, although the change in absorption may result from de-oxygenation of hemoglobin, and thus be highly wavelength-dependent ͓10,14,15͔.
Several in vitro studies involved continuous wave ͑CW͒ irradiation of whole blood, blood constituents, or phantoms.Early studies centered on the effect of CW laser radiation on morphological changes in erythrocytes and the release of hemoglobin ͓11,16,17͔.Halldorsson ͓15͔ recorded temperatures with a thermal camera during CW Nd:YAG laser irradiation of whole blood.This study indicated that evaporation played a significant role in surface heat loss and coagulum formation.
Verkruysse et al. ͓14͔ documented changes in transmission and reflection of pulsed dye laser light ͑ϭ586 nm͒, which they attributed to temperature-dependent variations in absorption.Other recent studies documented the development of pulsed laserinduced coagulation in albumen-essentially a water-ovalbumin ͑86-14 percent͒ solution ͓18,19͔.In these investigations, thermal damage produced an increase in light scattering and a coagulum which, for higher radiant exposures, adhered to the optical fiber tip.It should be noted that albumin has been used as a thermally activated adhesive in ''laser-tissue welding'' studies ͓4,20͔.

Predicting Thermal Damage in Blood.
The standard method of mathematically quantifying thermal damage in biological tissue involves using the Arrhenius rate process equation, which describes a unimolecular rate process ͓21-23͔: where ⍀ is the thermal damage coefficient ͑dimensionless͒, C is the concentration of molecules in the native state, t is time ͑s͒, A is the frequency factor ͑1/s͒, E a is the activation energy ͑J/mole͒ required to initiate the chemical transformation ͑protein denatur-ation͒, R is the universal gas constant ͑8.31 J/mole/K͒, and T is temperature ͑K͒.The damage threshold for tissue necrosis is commonly selected as ⍀ϭ1.0 ͑a damage concentration of 63 percent for a unimolecular system͒.When performing an Arrhenius analysis, the threshold is assumed to correspond with an experimental endpoint-typically a visible increase in light scattering, which makes the tissue appear white ͓24͔.It should be noted that this technique may yield differing results depending on the endpoint and detection method used.Rate process coefficients for blood constituents, bulk skin, and arterial tissue from the literature are presented in Table 1, along with the threshold temperatures for coagulation (T th ), for a 10 ms step change in temperature.These data were calculated using the following relation derived from Eq. ͑1͒: The relationship between T th and temperature step duration is also graphed in Fig. 1 for selected tissue property data from Table 1.Previous theoretical models of vascular lesion treatment have generally assumed threshold blood coagulation temperatures of 60°C-80°C.These values are based on studies involving exposure durations on the order of seconds or greater and a study of porcine bulk skin burns ͓25͔, which is commonly cited in the literature ͓1,26,27͔.Therefore, the purpose of this is study to investigate the validity of these threshold temperatures and rate process coefficients through quantification of the thermal and optical effects of pulsed laser-induced irradiation of blood.

Materials and Methods
The experimental portion of this study involved the use of three primary methods: thermal imaging ͑for transient surface tempera-tures͒, optical coherence tomography ͑for coagulum morphology, particularly thickness͒, and laser transmission measurements.All irradiations were performed with a Coherent Versapulse VPW-a clinical KTP ͑frequency doubled Nd:YAG͒ laser with a wavelength of 532 nm.A pulse duration ( p ) of 10 ms and beam diameters of 5 mm (H o ϭ2.5-6.0J/cm 2 ), 4 mm (H o ϭ6.5-7.5 J/cm 2 ), and 3 mm (H o ϭ8.0-13.0J/cm 2 ) were used.Human venous blood from a healthy male donor was withdrawn and placed in vials containing the anti-coagulant ethylenediamine tetraacetic acid ͑EDTA͒.While EDTA prevents the ͑nonthermal͒ blood coagulation process-a cascade of chemical reactions involving clotting factors and fibrin formation that occurs over sev-eral seconds-it has not been noted to have any effect on heatinduced denaturation in blood proteins ͓14,10͔.When necessary, vials were placed in a refrigerator for overnight storage and returned to ambient temperature the following day.All blood was used within 36 hours of withdrawal.Gentle agitation was performed to achieve a homogeneous distribution of blood constituents.For open-cuvette irradiations, blood was mixed in the vial, dispensed to the cuvette, and irradiated within 30 s.For spectrophotometer ͑glass͒ cuvette irradiations, blood was agitated within the closed cuvette immediately before irradiation.
Data have not been corrected for losses due to specular reflection.For the thermal imaging setup, the handpiece was oriented at an incident angle of ϳ22 deg ͑from perpendicular͒ and an open cuvette was used ͑n air ϭ1.0, n blood ϭ1.33͒, thus producing specular reflection of 2.0 percent.For OCT imaging and transmission measurements, the incident angle was 0 deg, and an enclosed cuvette was used, resulting in a specular reflection of 4.3 percent ͑airglass and glass-blood interfaces, where n glass ϭ1.5͒.
2.1 Thermal Imaging and Thermal Damage Calculations.Radiometric temperature measurements were performed using an Inframetrics 600L ͑ϭ3-5 m, band-limited͒ thermal camera.Radiation emitted from a sample ͑or the blackbody͒ was reflected off of a standard first-surface mirror to accommodate a horizontally oriented detector.Calibration curves for several temperature ranges were obtained using a variable temperature blackbody ͑Ealing Electro-Optics, Model BB1001-67͒.When recording high radiant exposure events, the camera was set for wider temperature ranges, thus necessitating the use of calibration curves, which maximized the accuracy at high temperatures at the expense of accuracy at lower temperatures.The spatial resolution of the camera was approximately 0.1 mm.In calculating temperature from the measured emissive power, the emissivity of blood was assumed to be equal to that of water ͑ϭ0.96͒.
Temperature measurements through the center of the beam spot were taken using the camera's line-scan mode.This method provided acquisition of a single line of spatial data at a rate of 8 kHz, and thus a temporal resolution of 125 s.However, the line scan mode also introduced several problems that limited the continuity of measurement: The scale bar and time code, which could not be removed from the video display, blocked several lines of data; fly-back in the camera resulted in a loss of 3.5 ms of data for every field ͑each 16.5 ms͒.Thus, transient temperature distributions at the center of the beam were measured in temporal segments and a continuous data set was reconstructed by averaging data from several discontinuous events.
Quantitative estimation of thermal damage was achieved by evaluating the Arrhenius equation using radiometric temperature data and rate process coefficients for bulk skin ͓25͔ and Hb ͓13͔.Numerical integration of this relation was performed at each 125-s time step using the trapezoidal method.

Optical Coherence Tomography
Imaging.An OCT system used previously to image thermal injury in hamster blood vessels in vivo ͓6͔ was used to determine the macroscopic morphology of coagula.The system consisted of a modified Michelson interferometer that measured the intensity of light backscattered from specific locations within the sample.The optical source was a 1280 nm center wavelength superluminescent diode.A single depth scan was acquired by changing the reference arm length.Scanning mirrors were used to scan the beam laterally across the sample in order to construct a two dimensional crosssectional image.The OCT system had a high dynamic range ͑Ͼ100 dB͒ and a spatial resolution of ϳ20 m, laterally and axially.
Blood was enclosed in a glass spectrophotometer cuvette ͑2 mm sample path length͒, agitated, then irradiated.Immediately after irradiation the sample was moved to the OCT stage and a two-dimensional image was generated.The acquisition time for an image consisting of 200 depth scans was approximately 20 s.

Transmission Measurements.
Transmission measurements were performed to quantify the effect of laser-induced optical changes.The glass spectrophotometer cuvette was filled with blood.A photodetector below the cuvette recorded the intensity of the alignment beam ͑ϭ635 nm͒.A circular aperture 6 mm in diameter was placed between the cuvette and the photodetector to remove highly scattered light.The output from the photodiode was captured by a digital oscilloscope ͑Tektronix TDS 640A͒ and transferred by disk from the oscilloscope to a personal computer.
Although a high-pass filter was used to minimize the effect of the KTP laser and flash lamp, the signal from the alignment beam was overwhelmed by stray light during the laser pulse, making it impossible to obtain transient data during the pulse.Therefore, the intensity of transmitted light at ϭ635 nm was measured before and after the KTP laser pulse.These two values were compared to determine the change in transmission due to coagulation.

Observations During Open-Cuvette Irradiations.
As the radiant exposure was increased from 2.5 J/cm 2 ͑for a pulse duration of 10 ms͒, no visible changes were noted until 3.6 J/cm 2 , when the blood surface appeared to be slightly discolored ͑whit-ened͒.At H o ϭ4.4 J/cm 2 a thin, light-gray disk about 8 mm in diameter, consisting of fine particles ͑coagulated blood compo-nents͒, was produced.The disk became irregular as it dispersed, and appeared to dissolve when the cuvette was agitated.For H o J/cm 2 , this disk became increasingly visible, then began to aggregate at a radiant exposure of 8.0 J/cm 2 .At H o ϭ10 J/cm 2 , the thin ͑ϳ100 m͒ coagulum could be removed from the cuvette with forceps.Although ablation-explosive vaporization identified by an audible acoustic transient ͑popping sound͒ ͓14͔ and fragmentation of the otherwise disk-shaped coagulum-was noted at H o ϭ11 J/cm 2 , it did not occur with every pulse until H o ϭ13 J/cm 2 .Ablation typically involves vapor-pressure-driven bubble expansion and subsequent collapse, both of which can cause material ejection or other forms of mechanical damage.Limited experimental data were collected for radiant exposures greater than 10 J/cm 2 due to the occurrence of ablation events that caused adverse results such as cuvette fracture.
Nonablative evaporation was detected by suspending a glass slide ϳ1 cm above the blood.As the radiant exposure increased, the first noticeable change ͑at H o ϭ2.7 J/cm 2 ͒ was an accumulation of a thin layer of condensation ͑clear liquid droplets assumed to be water͒ on the slide, appearing soon after the laser pulse and disappearing after a few seconds.The apparent thickness of the droplet layer increased steadily with H o .An attempt to quantify blood mass loss due to evaporation, using a scale with sensitivity on the order of tens of micrograms, was unsuccessful due to a low signal to noise ratio.

Temperature Measurements and Damage Predictions.
The temperature reconstructions for a point near the center of the beam are shown in Fig. 2. Each curve represents an average of data from 10-15 irradiations.The standard deviation for points near the temperature peak was Ϯ2°C for H o ϭ3 J/cm 2 , increasing to Ϯ5°C at H o ϭ12 J/cm 2 .Initially, the rate of temperature rise was proportional to the radiant exposure.For lower radiant exposures ͑3-5 J/cm 2 ͒, this rate did not change significantly unless temperatures exceeded 70°C.However, for higher radiant exposures ͑8-12 J/cm 2 ͒, deviation from the initial rate of temperature rise did not appear to be significant until temperatures approached 100°C.
Predictions of transient thermal damage accumulation based on Eq. ͑1͒ and using the temperature data in Fig. 2, and rate process parameters for Hb ͓13͔ and bulk skin ͓25͔ are presented in Figs.3͑a͒ and 3͑b͒, respectively.These figures indicate similar trends, yet Fig. 3͑b͒ indicates much earlier coagulation onset ͑⍀ϭ1͒ and greater damage levels than Fig. 3͑a͒.Figure 4 shows a series of OCT images that document the change in coagulum thickness with increasing radiant exposure.In these images, four layers are visible ͑as labeled in Fig. 4͑d͒, from top to bottom͒: a dark layer of glass, a thin white layer due to the change in index of refraction at the interface between the glass (nϭ1.5) and the blood (n ϭ1.33), a light, speckled region of highly scattering coagulated blood, and a darker speckled region of native blood, which extends to the bottom of each image.
A set of three to five images at each radiant exposure ( p ϭ10 ms) was used to generate the graph of coagulum thickness as a function of radiant exposure presented in Fig. 5.The first mea-surable coagulum occurred at H o ϭ4.4 J/cm 2 .The initial increase in coagulum thickness was followed by a regime of linear growth in coagulum size with radiant exposure.Due to the thickness of the coagula ͑Ͻ200 m͒, the spatial resolution of the OCT system ͑ϳ20 m͒, and the high noise level associated with OCT imaging of scattering media, a high level of precision was difficult to obtain.This variability is indicated in Fig. 5 by the error bars ͑Ϯstandard deviation͒.

Transmission Measurements.
The traces recorded by the oscilloscope indicated a stable pre-irradiation transmission level, followed by a large increase in transmission due to stray light from the KTP laser flash lamp, and a stable post-pulse transmission level.The change in transmission ͑preto post-laser pulse͒ with radiant exposure is presented in Fig. 5.A sample size of six or more was used for each data point in the graph.The data points represent the average of the sample and the error bars representϮthe standard deviation.At radiant exposures of 4.0 J/cm 2 or less, the difference in transmission before and after the laser pulse was insignificant.As H o increased from 4.0 to 5.0 J/cm 2 , a significant decrease in transmission occurred.For radiant exposures greater than 5.0 J/cm 2 , the transmission decrease became approximately linear with radiant exposure.

Discussion
4.1 Threshold Radiant Exposure for Coagulation.In experimental studies it is often assumed that coagulation onset ͑⍀ϭ1͒ corresponds to readily detectable ͑visible͒ tissue alterations.In this study, three different detection methods indicated two different threshold values: visual observation indicated that the threshold radiant exposure (H th ) was 3.6 J/cm 2 , whereas transmission measurements and optical coherence tomography indicated that H th ϭ4.4 J/cm 2 .
Coagulum thickness changed rapidly with H o near the threshold radiant exposure ͑Fig.5͒, whereas significant changes in transmission required slightly higher radiant exposures.The light transmitted through the cuvette was moderately diffuse due to the

Journal of Biomechanical Engineering
APRIL 2000, Vol.122 Õ 199 light scattering properties of native blood ͑ s ϭ460 cm Ϫ1 , g gϭ0.995͒.Coagulum formation may not have significantly increased the diffuse nature of the light propagating through the blood until the coagulum size reached a specific threshold.The scattering parameters of the blood coagulum may have been similar to that of denatured albumin ͑ s ϳ150 cm Ϫ1 , gϭ0.8͒ ͓20͔.If this was true, for a coagulum thickness less than ϳ67 m most photons underwent a single forward-scattering event in the coagulated region, whereas thicker coagula resulted in multiple scattering events which significantly altered photon trajectory.
After the initial sharp decrease in transmission in the H o ϭ5 -6 J/cm 2 range, Fig. 5 shows a more gradual change above 6 J/cm 2 .This change is likely due to the fact that the beam size was decreased from 5 to 4 mm between H o ϭ6 and 7 J/cm 2 .As beam diameter decreased, the probability of light being transmitted through a fixed 6-mm-diameter aperture increased, thus reducing the rate of decrease in transmission with radiant exposure.Another decrease in slope occurs at ϳ9 J/cm 2 where the beam diameter was further reduced to 3 mm.
A previous study by Verkruysse et al. ͓14͔ indicated that for in vitro irradiation of blood with a 0.5 ms laser pulse ͑ϭ586 nm͒, the time at which coagulation onset occurs during the pulse is determined by when a radiant exposure corresponding to H th has been delivered.Although this approach is not strictly consistent with an Arrhenius analysis, the difference between the theoretically determined onset time and that predicted using H th is negligible for many processes involving short laser pulses and high damage accumulation rates ͑Ͼ1000 s Ϫ1 ͒.In the present study, a radiant exposure equivalent to H th ͑assuming a threshold of 4.4 J/cm 2 ͒ was delivered for the 8, 10, and 12 J/cm 2 pulses in 5.5, 4.4, and 3.7 ms, respectively ͑assuming a constant irradiance͒, and resulted in measured temperatures of 89, 90, and 99°C and computed thermal damage ͑⍀͒ levels of 0.1, 0.1, and 10.1 ͑at the respective times͒.These discrepancies indicate that thermal losses occurred during the 10 ms laser pulse, leading to reductions in temperature and thermal damage.Thus, the use of H th for prediction of transient coagulation onset appears to produce only moderately accurate results for the cases studied.

Arrhenius Equation.
Threshold temperatures and radiant exposures are useful for rough approximations of damage onset.However, the Arrhenius equation provides a method of quantifying thermal damage accumulation that is much more accurate and flexible.The cost of these benefits is that calculation of thermal damage at any specific location requires accurate transient temperature data and a knowledge of local tissue rate process coefficients.
The data in Fig. 3 indicate that the accuracy of the Arrhenius relation is highly dependent upon the rate process coefficients used.Threshold radiant exposures of 5.0 and 3.1 J/cm 2 were predicted using rate process coefficients for Hb and bulk skin, respectively.Thus, the results of transmission and OCT experiments indicate better agreement with predictions for Hb than those for bulk skin.However, the visually observed threshold is in better agreement with predictions based on bulk skin coefficients.
Since Lepock et al. ͓13͔ used a different technique ͑calorim-etry͒ to determine the rate process coefficients for Hb, it is likely that the primary difference between the rate process coefficients for Hb and whole blood is the activation energy (E a ).The temperature data corresponding to H th ϭ4.4 J/cm 2 was found to predict ⍀ϭ1 when the rate process parameters found by Lepock et al. were used, with E a reduced slightly to 448 kJ/mole ͑whereas data for H thϭ3.6 J/cm 2 indicated E a ϭ434 kJ/mole͒.This alteration in E a caused the threshold temperatures in Fig. 1 ͑as calculated with Eq. ͑2͒͒ to shift downward while the slope of the curves remained nearly unchanged.The time-temperature relationship seen in this figure indicates that the transmission and OCT-based values are in moderately good agreement with blood constituents such as albumin and Hb.It should also be noted that these materials all have a similar threshold temperature as arterial tissue, given a 10 ms exposure time ͑Table 1͒.This is significant since it indicates that both of the primary tissue components in photothermolysis require similar temperatures to coagulate, and that these temperatures are significantly higher than previously assumed.The data in Fig. 1 corresponding to the visual observation endpoint (H th ϭ3.6 J/cm 2 ) are not in good agreement with any other data set, yet are in better agreement with the bulk skin data than any other material.
Although our results indicate that the Arrhenius equation is useful in predicting post-irradiation coagulation onset for this specific case, true validation will require corroboration of transient temperature and thermal damage data.

Coagulation Threshold Temperatures.
By definition, specification of a threshold temperature (T th ) implies a step change in temperature and a well-defined exposure time ͓23͔.Since the transient thermal processes investigated in this study do not have step temperature profiles, an Arrhenius approach indicates that two temperatures are useful in describing a coagulation event: an onset threshold temperature (T on )-the maximum temperature that occurs for a radiant exposure of H th -and a transient threshold (T tr )-the temperature at the instant of coagulation onset ͑for an event during which coagulation occurs during the laser pulse͒.
Interpolating from the data in Fig. 2, the threshold radiant exposure of 4.4 J/cm 2 corresponds to a maximum radiometric temperature of 90Ϯ3°C (T on ).This result is in reasonable agreement with the threshold temperature for a 10 ms step temperature change in blood constituents such as albumin and Hb ͑Table 1, lines 5 and 8͒, but significantly higher than the 60°C-80°C used to predict vessel necrosis in theoretical models of pulsed laser treatment of vascular lesions ͓1,28,29͔.However, it should be noted that the maximum temperature corresponding to a radiant exposure of 3.6 J/cm 2 was 78.5°C.
For radiant exposures significantly greater than H th ϭ4.4 J/cm 2 , T on was reached during the laser pulse, but since coagulation at T on requires a finite exposure duration, greater temperatures ͑and a greater damage accumulation rate, d⍀/dt͒ were produced before coagulation occurred ͑at T tr ͒.Since transmission and coagulum thickness data were not collected transiently during the laser pulse, it is difficult to identify T tr directly from experimental results.However, using the rate process coefficients indicated in the previous section for H th ϭ4.4 J/cm 2 , coagulation onset for H o ϭ8.0, 10.0, and 12.0 J/cm 2 was predicted to occur at temperatures of 95°C-99°C.The corresponding temperature range determined for H th ϭ3.6 J/cm 2 , was 82°C-86°C.

Thermally Induced Changes in Optical Properties.
The OCT images in Fig. 4 clearly indicate the increased level of light scattering induced by thermal coagulation of blood.These findings are in agreement with previous investigations, which have indicated that thermal damage of blood constituents results in an increase in scattering ͓10,15͔.Although the transient effects of light scattering were not documented in this study, it is likely that the coagulated region increased in thickness both during and after the laser pulse in a manner predictable by rate process kinetics ͓19͔.
Since blood contains various proteins, which have different coagulation thresholds, it is possible that multiple changes in optical properties occurred.Results such as the indication of slight blood discoloration at H o ϭ3.6 J/cm 2 may have been a result of scattering brought on by denaturation of blood constituents which are less thermally stable, such as erythrocyte membrane proteins ͓10͔.However, Arrhenius data ͑Fig.1͒ indicate that membrane proteins have a higher threshold temperature than other blood components for short exposure times.
For radiant exposures above 5.0 J/cm 2 , Fig. 2 indicates that when measured temperatures of 70°C or greater are reached, the slope of the temperature curve decreases, then becomes relatively constant above 100°C.A simple one-dimensional transient thermal model which included convection, conduction and a Beer's 200 Õ Vol.122, APRIL 2000 Transactions of the ASME law-based source term ͑but not evaporative cooling nor changes in optical properties͒ was developed.Simulations indicated a nearly linear temperature rise, without the aforementioned changes at 70°C and 100°C.Therefore, other mechanisms need to be considered.In this temperature regime, heat loss by water vaporization is a significant factor ͑see Section 4.6͒.However, dynamic changes in optical properties may also play a significant role in the transition near 100°C as well.The change in slope is nearly simultaneous with the onset of coagulation as predicted by the Arrhenius relation using the rate process coefficients of Hb ͑Fig.3͒.Coagulation-induced changes produce an increase in scattering, which should increase the rate of temperature rise.Such an increase was not noted, possibly because the increase in scattering was small in comparison to the high absorption coefficient.Dynamic changes in absorption, which may also play a role in the transient changes seen in the temperature distribution, are not well understood.A dynamic decrease in transmission was noted by Verkruysse et al. ͓14͔ during pulsed laser irradiation of blood ͑at ϭ586 nm͒.Temperature-dependent changes in Hb absorption were indicated to be the dominant mechanism; however, dynamic changes in scattering were not accounted for thoroughly.At wavelengths for which there is a large disparity in absorption between Hb and HbO 2 ͑such as 586 or 1064 nm͒, significant transmission changes may result from thermally induced deoxygenation of HbO 2 ͓15͔.Since the difference between a for Hb ͑185.1 cm Ϫ1 ͒ and HbO 2 ͑228.8cm Ϫ1 ͒ is less than 20 percent at ϭ532 nm, it is not likely that this effect played a significant role during laser irradiation ͓30͔.However, there is an order of magnitude difference in absorption coefficient of blood at ϭ635 nm between HbO 2 ͑2.2 cm Ϫ1 ͒ and Hb ͑22 cm Ϫ1 ͒.Therefore, the transmission measurements presented in Fig. 5 may have been affected by changes in absorption as well as scattering.

Coagulum Size and Consistency.
The variation in coagulum thickness with radiant exposure ͑Fig.5͒ shows a steep initial slope followed by a more moderate slope, which continues until ablation.The second phase of the curve indicates a region for which changes in coagulum depth occur in nearly direct proportion to changes in radiant exposure.Similar trends have been noted in a previous study of Holmium:YAG irradiation of albumen ͓18͔.
As shown in Fig. 5, the maximum blood coagulum thickness recorded was 175 m (H o ϭ10 J/cm 2 ).At a depth of 175 m, only 1.8 percent of the initial fluence remains if the blood is highly oxygenated ͑3.9 percent if de-oxygenated͒.Therefore, it is likely that axial heat diffusion caused the coagulated region to be thicker than if the process were thermally confined.It is also possible that convective ͑pressure or temperature-driven͒ flow may have played a role.
The transition from a region of disconnected ͑or weakly bonded͒, denatured proteins to an aggregated coagulum appears to have taken place between radiant exposures of 8.0 and 10.0 J/cm 2 , although it was not detectable by OCT imaging.If a threshold radiant exposure for aggregation of 9.0 J/cm 2 is assumed, the threshold temperature would be ϳ115°C, and the resulting thermal damage, ϳ1000 using the Arrhenius equation and rate coefficients of Hb ͑Table 1, line 8͒.However, accurate prediction will not likely be accurately achieved using the rate process coefficients for hemoglobin.If aggregation can indeed be modeled as a rate process, experimental determination of a novel set of rate process coefficients will be necessary.
The finding of blood coagula in various stages of aggregation corresponding to radiant exposure level is in agreement with previous findings ͓6͔, which indicated that the threshold for embolized coagulum formation was much lower than the radiant exposure required for formation of a flow-stopping coagulum.The mechanism of transition from an unconnected solution of denatured molecules to a strongly bonded coagulum is very significant not only for arresting flow in blood vessels but for other applications such as forming patent anastomoses with a laser, yet it is still not well understood.It is likely that the primary mechanism is either an increase in the concentration of denatured molecules due to additional thermal damage or evaporative water loss, which decreases the intermolecular space, thus facilitating bond formation.In order to achieve a better understanding of this process, it will be necessary to study the molecular composition and microscopic morphology of coagula.

Effect of Phase Change.
Radiometric temperature measurements provided evidence of several significant dynamic mechanisms.In Fig. 2, the linear slope for the lowest radiant exposure indicates a situation in which minimal dynamic changes occur.For higher H o values, maximum temperatures rose above 70°C, vapor production was observed, and the rate of temperature rise decreased.These results indicate that phase change became significant at about 70°C.This is in fair agreement with the literature, which indicates that for CW processes, evaporation becomes significant at about 60°C, and increases exponentially with temperature ͓15,31,32͔.Therefore, one would expect that as temperatures exceeded 100°C, the rate of temperature rise would continue to decrease, whereas in fact, it became relatively constant ͑Fig.2͒.For CW irradiation, Halldorsson ͓15͔ reported a significant increase in the rate of temperature rise above 95°C, which he explained as being due to desiccation of the surface blood.Therefore, an exponential increase in evaporation rate, followed by a reduction in cooling due to desiccation, may help explain the measured temperature distributions.
The threshold radiant exposure for ablation determined in this study was ϳ12 J/cm 2 and the corresponding maximum surface temperature ͑during nonablative irradiation͒ was 145°C.At this temperature water has a saturation pressure of greater than 4 atm ͓33͔.In order to establish whether the maximum surface temperature was also the maximum tissue temperature, it is necessary to consider two possible scenarios proposed in the literature.Evaporative cooling may have acted to moderate the heating rate at temperatures greater than 60°C.As the tissue became desiccated-likely near 100°C-the cooling mechanism was reduced and the rate of temperature rise increased sharply near the surface, leading to ablation ͓23͔.This sharp increase is not seen in Fig. 2, possibly because desiccation caused a reduction in emissivity, resulting in an underestimation of the temperatures in the superficial blood layers.If this description of the ablation process is accurate, then the surface temperature was the maximum tissue temperature.Alternatively, surface heat loss and a layer of highly absorbent, highly scattering coagulum may have produced a subsurface temperature peak which in turn led to phase change and a corresponding subsurface peak in pressure.When internal pressures exceeded the restraining forces ͑e.g., blood surface tension, coagulum tensile strength͒, ablation occurred ͓34͔.If a subsurface temperature peak occurred deeper than 20 m, then the measured temperatures may represent a significant underestimation ͑Fig.2͒.

Conclusions
Pulsed KTP laser-induced thermal damage in whole blood was investigated in vitro.A progression of events occurred with increasing radiant exposure: evaporation, coagulation onset, aggregated coagulum formation, and ablation.Both coagulation and ablation were indicated to occur at higher temperatures than typically assumed by previous theoretical models of laser treatment of cutaneous vascular lesions.Phase change was shown to play a primary role, leading to surface heat loss ͑and possibly desiccation of the coagulum͒ as well as ablation.The Arrhenius equation was indicated to be valid for predicting pulsed laser-induced coagulation of blood, with the rate process coefficients for hemoglobin indicating better agreement with experimental data than the commonly used bulk skin coefficients.
ing for this research was provided in part by grants from the Air Force Office of Scientific Research through MURI from DDR&E ͑F49620-98-1-0480͒, the Office of Naval Research Free Electron Laser Biomedical Science Program ͑N00014-91-J-1564͒, and the Albert W. and Clemmie A. Caster Foundation. A. J. Welch is the Marion E. Forsman Professor of Electrical and Computer Engineering.

Nomenclature
A ϭ frequency factor, 1/s CW ϭ continuous wave E a ϭ activation energy, J/mole Ho ϭ holmium H o ϭ radiant exposure, J/cm 2 H th ϭ threshold radiant exposure, J/cm 2 KTP ϭ potassium-titanyl-phosphate n ϭ index of refraction Nd ϭ neodymium OCT ϭ optical coherence tomography T ϭ temperature, °C or K T th ϭ threshold temperature T on ϭ onset threshold temperature T tr ϭ transient threshold temperature YAG ϭ yttrium-aluminum-garnet ϭ emissivity ϭ wavelength, nm p ϭ pulse duration, ms ⍀ ϭ thermal damage coefficient

Fig. 1
Fig. 1 Threshold temperatures of skin and blood constituents as a function of exposure time, using Eq."2…, the rate process data in Table 1, and assuming step changes in temperature.Symbols are for identification purposes only and do not represent specific data points.

Fig. 2
Fig. 2 Radiometric temperature distributions for a point near the center of the laser beam "Ä532; p Ä10 ms… for six different radiant exposure levels

Fig. 4
Fig. 4 OCT images documenting the morphology of coagulated regions of whole blood at radiant exposures of: "a… 4.4, "b… 6, "c… 8, and "d… 10 JÕcm 2 .As labeled in "d…, four regions are visible "from top to bottom…: glass "black…, glass-blood interface "white…, coagulated blood "bright speckle…, and native blood "darker speckle….The length of size bars corresponds to a distance of approximately 200 m.The thickness of each coagulum is indicated by a pair of thin lines.The left side of image "d… shows the edge of coagulum.Note that the intensity of the OCT signal in "a… is lower in the coagulum "and higher in the native blood… than in the corresponding regions of the other images.

Table 1 Arrhenius rate process coefficients of blood constitu- ents and related biological materials and corresponding threshold temperatures calculated for a 10 ms, constant- temperature exposure. Asterisks indicate E a determined from data in the present study.
th ϭ4.4 J/cm 2 ) Journal of Biomechanical Engineering APRIL 2000, Vol.122 Õ 197