Bubble Formation as Primary InteractionMechanism in Retinal Laser Exposure

With 200-ns Laser PulsesJohann Roider, MD,1* El-Sayed El Hifnawi, PhD,1 and

Reginald Birngruber, PhD2

1Department of Ophthalmology, Medical University of Lubeck, 23568 Lubeck, Germany2Medical Laser Center, 23568 Lubeck, Germany

Background and Objective: Retinal laser photocoagulation isgenerally performed by laser pulses of a few hundred millisec-onds. The tissue interaction mechanism is a pure thermal inter-action mechanism. As pulse duration gets shorter, different,non-thermal interaction mechanisms start to appear. The timedomain for a change of tissue interaction mechanism seems tobe in the ns and ms range. The goal of this study was to charac-terize the tissue interaction mechanism with 200-ns laserpulses, which approximate the thermal relaxation time of singlemelanin granules.Materials and Methods: The retinas of 19 eyes of 10 rabbits wereirradiated by 10 and 500 repetitive laser pulses (wavelength, 532nm; repetition rate, 500 Hz; pulse duration, 200 ns; per pulseenergy, 0–120 mJ; retinal spot size, 100 mm). The effects wereevaluated by fluorescein angiography, ophthalmoscopy and bytheoretical thermal calculations. Light microscopy and trans-mission electron microscopy were additionally performed onlesions irradiated by 500 pulses.Results: Single pulse threshold energies for angiographic vis-ibility were 3.5 mJ (10 pulses) and 2.1 mJ (500 pulses), for oph-thalmoscopic visibility 9.0 mJ (10 pulses) vs. 8.6 mJ (500 pulses).At energy levels above ophthalmoscopic visibility macroscopi-cally visible bubble formation inside the retina could be ob-served. This occurred at energy levels of 35 mJ (10 pulses) vs. 17mJ (500 pulses). Microscopic evaluation of lesions irradiatedwith 500 pulses and energies at the angiographic thresholdshowed a damage primarily to the RPE. Additional outer seg-ment damage of the photoreceptors could be found. A gap be-tween damaged RPE cells and the outer segments could be re-peatedly found as well as damaged RPE cells, which were de-tached from intact Bruch’s membrane. Temperature calculationshows that temperatures above 100°C may exist around singlemelanin granules.Conclusion: The studies suggest that RPE damage may occur bybubble formation around single melanin granules. Lasers Surg.Med. 22:240–248, 1998. © 1998 Wiley-Liss, Inc.

Key words: bubble; laser; melanin; retina; retinal pigment epithelium

INTRODUCTION

Clinical retinal photocoagulation as per-formed, for example, in diabetic eyes is generallyperformed with exposure times of about 100 ms ofa green laser, as recommended by the diabetic

*Correspondence to: Johann Roider, MD, Augenklinik, Klini-kum Der Universitat Regensburg, Franz-Josef-Strauß Allee11, D-93053 Regensburg, Germany.

Accepted 8 January 1988

Lasers in Surgery and Medicine 22:240–248 (1998)

© 1998 Wiley-Liss, Inc.

retinopathy study research group [1]. The histo-logical observations of damaged choriocapillaris,damaged RPE and photoreceptors after retinalexposure are generally explained by heat diffu-sion out of the absorbing layer and subsequentthermal denaturation of adjacent tissue. Themain absorbing layer with a green laser is theretinal pigment epithelium (RPE), where about50% of the laser light gets absorbed [2]. Generalunderstanding is that the interaction mechanismin retinal photocoagulation with pulse durationlonger than 50 ms is a pure thermal mechanism,which can be described by the Arrhenius law [3].The Arrhenius law basically describes the rela-tionship between tissue temperature and damageobserved. The Arrhenius law holds down to thems range [4]. If the laser pulse duration is shorterseveral other effects can be found, which no longercan be explained by a pure thermal interactionmechanism. Experiments with Q-switched singleruby laser pulses and energy densities of 45–65J/cm2 showed disruptive effects with subsequentbleeding [5,6]. If pulse duration of ps and fs areused pure mechanical effects, like damage pro-duced by stress or shock waves, prevail [7,8]. It isunclear at what pulse duration a change in inter-action mechanism from a thermal one to a non-thermal one takes place. It is of special interestwhat kind of interaction mechanism prevails atthe retina, when the laser exposure is in the ns toms range. Recent experiments with repetitive 5-mslaser exposure have shown that the histologicalobservation after retinal laser exposure with re-petitive laser exposures of a green laser (514 nm)could not be explained by a pure thermal mecha-nism and an additional coexisting mechanism hadto be postulated [9]. The order of magnitude ofpulse duration, where a change of mechanismcould start, can be estimated by the thermal re-laxation time of the melanin granules inside theRPE, which are the main absorbers in retinal pho-tocoagulation. Birngruber et al. [4] describes thethermal relaxation time by tr 4 r2/6k, where k isthe thermal conductivity of water (1.5 × 10−3 cm2

s−1) and r the radius. Anderson [10] describe it by4r2/27k. This means that the thermal relaxationtime of a single melanin granule with a size ofabout 0.6 mm [10,11] is on the order of about 100ns. Therefore laser exposure with pulses in thistime domain should show a different interactionmechanism compared to ms or ms laser exposure.Retinal laser exposures in this study were per-formed with 200 ns pulses (FWHM) of a Nd:YAG(532 nm) laser. To study the interaction mecha-

nism repetitive laser exposures with 10 and 500laser pulses were applied to the retina. The ex-perimental parameters for retinal laser exposureswere—beside the pulse duration—nearly identi-cal to the experimental laser parameters per-formed earlier [9,12]. In those earlier experimentsthe parameters were as follows: wavelength 514nm, pulse duration 5 ms, repetition rate 500 Hz,number of pulses applied between one and 500pulses, retinal spot size 110 mm, single pulse en-ergy used 2–10 mJ. Therefore it should be possibleto make statements on the interaction mechanismwith 200 ns laser pulses and to compare it to 5 mswhere the interaction mechanism seems to be aprimarily thermal one.

MATERIALS AND METHODSLaser System

A modified prototype of a frequency-doubledNd:YAG laser (532 nm) (Carl Zeiss GmbH) wasused. The pulse duration (FWHM) was 200 nswith a Gaussian shape. The repetition rate was500 Hz. The number of pulses applied was 10 and500. The maximum energy available was 120 mJ.The spot size of the laser beam, as analyzed by abeam analyzer (Spicron LBA-100A), was about160 mm in air. The laser was focused into the ani-mal’s eye by a standard laser slitlamp and had arectangular shape at the retina, since the fiberwhere the laser was coupled in was imaged to thefundus. A Goldman contact lens was used for allexperiments. Detailed calculations show that themeasured spot size in air transforms to 100 mm ina rabbit fundus using a plan-concave contact lens[13]. Pulse energy was checked before each ex-periment by measuring the average power (Scien-tech MA 100).

Laser Photocoagulation

Nineteen eyes of 10 chinchilla gray rabbitswere used, because the density and location oflight absorbing pigments in the fundus are ratheruniform and similar to that of the human eye. Theanimals were anesthetized with ketamine hydro-chloride (35 mg/kg body weight) and xylazine hy-drochloride (5 mg/kg of body weight). The treat-ment of experimental animals in this study fol-lowed both the principles of laboratory animalcare as well as the national laws.

Before starting the experimental exposuresfor orientation a pattern of six to 12 marker le-sions was performed to the regio macularis by aconventional cw argon laser (100 ms exposure

Bubble Formation in Retinal Laser Exposure 241

time). The necessary power of these marker le-sions varied between 40 and 80 mW. The test le-sions to study were performed between themarker lesions and the topographic relationshipto the marker lesions was recorded. All test le-sions were only performed to the macular region,but not to the periphery. In total 1,170 test lesionswere performed. Fundus evaluation was done im-mediately after exposure by ophthalmoscopy.Fundus evaluation included fluorescein angiogra-phy 2 hours after exposure. Fundus pictures wereadditionally made. Threshold curves for differenteffects were established by plotting a given effectvs. the necessary single pulse energy. Four endpoint criteria were investigated: Visibility by fluo-rescein angiography, visibility by ophthalmoscopyand bubble formation, a new effect which had notbeen expected.

Morphologic Study

Representative lesions were investigated bylight and transmission electron microscopy. Theretinas were fixed in 2.5% glutaraldehyde andpostfixed in Dalton’s osmium fixative, dehydratedin alcohol and embedded in epoxy resin (Epon).Ultrathin sections were stained with uranylac-etate. One-micrometer serial sections were cutuntil the center of the lesions was reached. Sev-enteen different lesions were investigated by lightmicroscopy and seven different lesions were in-vestigated by transmission electronmicroscopy.All eyes were enucleated in vivo 2 hours after reti-nal exposure. Table 1 shows the number of lesionsinvestigated. For comparison to the former 5 mslaser lesions in this 200 ns study only lesions afterexposure to 500 pulses were examined.

Statistical Analysis

For statistical analysis the incidence of dam-age visible by fluorescein angiography, ophthal-moscopy and bubble formation was plotted vs.pulse energy for 10 and 500 pulses in a probit plot.The threshold energy ED50 is the per-pulse en-ergy necessary to achieve 50% probability of de-fined damage.

Thermal Calculations

For thermal calculation a thermal model, de-scribed earlier [14], was initially used. Since theresults of this model showed that the far field ofthe temperature of the single melanin granuleswith 200 ns laser pulses can be neglected a pureanalytical thermal model for a melanin granulewas used. A melanin granule was assumed tohave a cylindrical geometry with a diameter of 0.6mm and a length of 0.6 mm. The analytical solu-tion, described by Vassiliades [15], was used, tocalculate the temperature T:

T~z,t! =S

2rc *0

t 11 − exp 3−a2

4az423erf 1 ~z + l!

2=at82− erf 1−

~z − l!

2=at824 dt8 for t ø t, (1)

where the z is along the axis of symmetry, 2l thelength of the absorbing cylinder and a the radiusof the cylinder. r is the density, c the specific heatand a the thermal conductivity and t the durationof the laser pulse. Thermal properties of waterwere used. A loss of energy from the cornea to theRPE of 7% was assumed. The total energy absorp-tion of a single melanin granule was assumed by63%, which is consistent with the total absorptionof green light of an RPE layer [9].

RESULTS

Statistical Results

Table 2 shows the statistical results ofthreshold energy (ED50) necessary to achieve agiven effect with the Nd:Yag laser (532 nm) (100mm spot size on the retina, 500 Hz, 200 ns).

TABLE 1. Lesions Investigated Histologically (200 ns,500 Hz, 500 pulses, 100 mm on the retina)

Energy per pulseLight

microscopyTransmission

electronmicroscopy

2 mJ 8 23 mJ 5 26 mJ 4 3

TABLE 2. Threshold Energy (ED50) for a Given Effectat 10 and 500 Pulses

10 pulses(200 ns)

500 pulses(200 ns)

Angiographic visibility 3.5 mJ 2.1 mJOphthalmoscopic visibility 9.0 mJ 8.6 mJBubble formation ≈35 mJ ≈17 mJHemorrhage not investigated ≈120 mJ

242 Roider et al.

Ophthalmoscopic Results

With increasing energy the lesions becomevisible by ophthalmoscopy. Initially the lesionsappear gray, and later they become white and ul-timately chalk white. With increasing energy aphenomena started to appear, which we call mac-roscopic bubble formation. Initially tiny, reflect-ing surfaces in middle of the white lesions can bedetected and with more energy they appear as asingle three-dimensional bubble-like appearancein middle of the lesion. The size of the bubble isenergy dependent. It is interesting to note thatafter application of 500 pulses with energies,which are at the threshold of bubble formation of10 pulses, several adjacent bubbles start to ap-pear. Figure 1 illustrates this situation. Thesebubbles were visible more than 1 hour after irra-diation. Bleeding always started from the bubblesin middle of the chalk-white lesions. The hemor-rhage was a choroidal hemorrhage.

Histology

All lesions investigated by histology were ex-posed by 500 pulses and 500 Hz. In order to get anidea on the mechanism with 200 ns all lesionsinvestigated were exposed with energy levels atthe angiographic threshold ranging between 2and 6 mJ (25 mJ/cm2–75 mJ/cm2). All the lesionswere not visible by ophthalmoscopy, only byflourescein angiography.

All the lesions were located primarily to theretinal pigment epithelium (RPE). Melanin gran-ules appear ultrastructurally intact and were api-cally located at their primary site. No damagecould be detected in the choriocapillaris. Bruch’smembrane was intact in all cases. Some damage,depending on the energy used, was visible in thephotoreceptor layer. No damage was visible to the

other inner retinal layers. With 2 mJ energy perpulse the RPE is heavily damaged. The cytoplasmis condensed. Degenerative changes are visible atthe directly adjacent outer segments, showing ho-mogeneous, electronlucent discs. The outer seg-ments are often irregularly orientated both insideas well as at the edge of the lesion. With 3 mJenergy per laser pulse the lesion is also confinedprimarily to the RPE. Regularly a space betweenthe irradiated, damaged RPE and the photorecep-tor outer segments could be found (see Fig. 2).RPE cells can be completely lost, so that photore-ceptor outer segments approach the underlyingmorphologically intact Bruch’s membrane (seeFig. 3). In several cases the damaged RPE wasdetached from intact Bruch’s membrane showinga space between these two layers (see Fig. 4).Outer segments appear repeatedly shortened.Parts of damaged outer segments could be seenbetween morphologically intact OS. Occasionallymitochondrias of the inner segments appear vacu-olized.

The lesions irradiated by 6 mJ are easily de-tected by light microscopy. The space between theRPE and outer segments is easy to detect. Outersegments appear irregularly orientated. It is note-worthy that damaged cell nucleoli could be re-peatedly found, which showed ultrastructurallynormal membrane of the nucleus, while the cellnucleus is condensed and its chromatin isshrunken (see Fig. 5). Damage to the inner seg-ments could be found, showing damaged mito-chondria. Occasionally a pycnotic cell nucleuscould be found inside the outer nuclear layer.

Temperature Calculations

Figure 6 shows the spatial temperature pro-file in the immediate neighborhood of a single

Fig. 1. Schematic drawing of macroscopic visible bubble formation inside a laser lesion after application of 10 and 500 pulsesat a repetition rate of 500 Hz. If 10 pulses with single pulse energies at threshold of bubble formation are applied to the fundusa single bubble can be observed in the middle of the chalk-white laser lesion. However, if 500 pulses of the same per pulseenergy are used several adjacent bubbles can be seen.

Bubble Formation in Retinal Laser Exposure 243

melanin granule at the end of a 200 ns laser pulseand a 5 ms laser pulse using equation 1. At allcalculations the same energy (1.21 · 10−10 J) wasused. The energy used (1.21 · 10−10 J) is the en-ergy which hits the cross section of single melaningranule (0.6 mm diameter) inside a rectangular

laser spot of 100 mm and a total laser energy of 6mJ. A single melanin absorption of 63% was usedand a loss of energy of 7% from the cornea to theRPE is assumed. It can he seen that there is sig-nificant heat conduction out of the granule withe.g. a 5 ms laser pulse, but only little with a 200 nslaser pulse. The adiabatic temperature can be cal-culated to about 170°C. If 200 ns laser pulses areused the maximum temperature increase is about125°C at 6 mJ pulse energy due to some heat con-duction. Using 3 mJ per pulse energy, exposureparameters as experimentally used, the tempera-ture increase can be calculated to 62.5°C. For theabsolute temperatures body temperature of 37°Chas to be added. This means that the absolutetemperature reaches at least 100°C after eachsingle 200 ns laser pulse of 3 mJ energy and 160°Cafter 6 mJ energy.

DISCUSSION

In order to elucidate the mechanism the re-sults have to be compared to results publishedearlier, where also laser exposures have beenmade to the retina [9,12]. The exposures in thoseearlier experiments have been made to the retina

Fig. 2. Light micrograph obtained 2 hours after irradiation(500 pulses, 200 ns, 500 Hz, 3 mJ single pulse energy). Theirradiated RPE is completely damaged (arrows). A space be-tween the irradiated RPE and the outer segments can befound (×550).

Fig. 3. Transmission electron micrograph obtained 2 hoursafter pulsed irradiation (500 pulses, 200 ns, 500 Hz, 3 mJsingle pulse energy). RPE cells have been completely lost, sothat photoreceptor outer segments (O) approach the underly-ing morphologically intact Bruch’s membrane (B). (R: dam-aged RPE) (×3,780).

Fig. 4. Transmission electron micrograph obtained 2 hoursafter pulsed irradiation (500 pulses, 200 ns, 500 Hz, 3 mJsingle pulse energy). The RPE is completely damaged. Mela-nin granules appear normal. The whole damaged RPE (R) iscompletely detached from the morphologically intact Bruch’smembrane (B). (×9,660).

244 Roider et al.

with nearly identical parameters beside the pulseduration. The parameters were as follows: wave-length 514 nm, pulse duration 5 ms, repetitionrate 500 Hz, number of pulses applied betweenone and 500 pulses, retinal spot size 110 mm,single pulse threshold energies for 500 repetitivepulses 1.5 mJ [9,12]. Similarly to the results re-ported here the RPE was heavily damaged whilethere was sparing of the photoreceptors. Theanalysis of these data showed that the primarydamage mechanism is a thermal one with an ad-ditional, unknown mechanism. In those experi-ments no histological clues for a primary thermo-mechanic or acoustic mechanism were found. Theobservations reported here after exposure with re-petitive 200 ns laser pulses show some similari-ties to those 5 ms experiments but many differ-ences.

In both experimental series some additivityof the effects was found. In Figure 7 the energydensity per single pulse is plotted vs. number ofpulses applied in a logarithmic scale for thoseearly ms experiments and the 200 ns experiments.The ED50 values for a angiographic visible lesionwere used. Additionally the ED50 values for vesselocclusion for arterioles and venules for repetitive

laser pulses with 160 ms laser pulses were plottedas well [16]. In those 160 ms experiments arteri-oles and venules of 30 mm diameter of the ham-ster cheek pouch were irradiated with a pulseddye laser (577 nm, 160 ms, 1.2 mm spot size, 0.5Hz) and different number of laser pulses (1, 10and 100) and vessel occlusion was chosen as anendpoint. Threshold values for vessel occlusionwere established. In those 160 ms experiments athermal mechanism was also shown. In all experi-ments with repetitive laser exposures a linear re-lationship can be detected. The relationship canbe described by

E(N)E1

= c ? Nr, (2)

where E1 is the energy necessary of a single pulseand E(N) the single pulse energy, if N pulses areused. However the correlation factor r is different.In the experiments where the mechanism is sup-posed to be a primary thermal one the correlationfactor r is between −0.19 and −0.22, as opposed toa factor r of −0.13 with the 200 ns laser exposures.This difference in correlation factors could be ahint to a different mechanism.

Fig. 5. Transmission electron micrograph obtained 2 hoursafter pulsed irradiation (500 pulses, 200 ns, 500 Hz, 6 mJsingle pulse energy). After irradiation the cell nucleus is con-densed and its chromatin is shrunken. The membrane of thenucleus can still be seen. Melanin granules appear morpho-logically intact. (×8,740).

Fig. 6. Temperature profile inside and around a single mela-nin granule as calculated by using equation 1 after a 5-mslaser pulse and after a 200-ns laser pulse. Each laser pulsecontains the same amount of energy of 1.21 · 10−10 J.

Bubble Formation in Retinal Laser Exposure 245

Macroscopic observations after 200 ns laserexposures showed retinal bubble formation withenergy densities of about 0.45 J/cm2 and 10 pulses0.22 J/cm2 and 500 pulses. The macroscopic vis-ible bubbles were located inside or below the neu-ral retina, as judged by ophthalmoscopy. Macro-scopic bubble formation was also described with aquasi-cw Nd:YAG laser (532 nm) at rabbits and inpatients [17] with energy densities of 0.48 J/cm2.The repetition rate in that laser was 13 kHz andthe pulse duration varied due to bad pulse shapeirregularly between 1 and 10 ms. This energy den-sity is very similar to our results where macro-scopic bubble formation was found with 200 nspulse duration. This could mean that bubble for-mation is a phenomenon associated with repeti-tive short laser pulses.

The energy densities, where angiographicvisible lesion could be achieved with the repetitive200 ns laser pulses, were very similar to the re-petitive 5 ms laser pulse experiments. Thresholdvalues for angiographic visibility with 5 ms laser

pulses were 5.5 mJ for one pulse, 2.6 mJ for 25pulses and 1.5 mJ for 500 pulses [12]. Howeverhistological analysis showed many differences,which could not be found after irradiation withrepetitive 5-ms laser pulses. After 200-ns thresh-old laser exposures in many lesions a gap betweenboth the damaged RPE and the underlyingBruch’s membrane could be found (see Fig. 4).Such effects, like detached RPE cells, had also befound after ps laser exposures [18]. In several le-sions a gap between the RPE and the photorecep-tors could be found (see Fig. 2). The outer seg-ments were orientated irregularly after laser ex-posures and appeared shorter compared to non-irradiated areas. All such effects could never befound after repetitive 5-ms laser exposures. RPEcells can be completely lost, so that photoreceptorouter segments approach the underlying morpho-logically intact Bruch’s membrane (see Fig. 3). In-ner segments showed some damage after irradia-tion with higher laser energy (6 mJ). Compared tothe experiments with the repetitive 5-ms laserpulses the area of damage is larger and the dam-age mechanism seems to be different.

The primary interaction mechanism after ir-radiation with 200-ns laser pulses seems to be anon-thermal mechanism. If one assumes that tis-sue damage takes place inside the RPE cell,maybe a few mm away from the melanin granules,the absolute temperatures are much too low toexplain the histological effects by a thermalmechanism. Generally all authors agree that nothermal effect takes place with laser pulses in thens range. Melanin can oxidize different cell com-ponents like vitamin C [19]. Therefore a photo-chemical mechanism has also to be considered.However reactivity of melanin strongly dependson the intact structure, as shown with experi-ments of 10-ns pulses of a Nd:YAG (532 nm) [20].Therefore a photochemical mechanism seems un-likely, since all melanin granules appear intact inour experiments. Lin and Kelly [8] showed byhigh speed photography shock and stress waveswith 100 ps (1,064 nm) and energy densities of 1J/cm2. Morphological signs for a pure acousticdamage like dilated endoplasmatic reticulumwere not found [21]. Furthermore the retinal en-ergy densities in our experiments are much toolow to create such effects. Energy densities at an-giographic threshold are between 0.025 and 0.075J/cm2.

Temperature calculations showed that no cu-mulative temporal temperature effects due to re-petitive laser pulse exposures exist, because the

Fig. 7. Threshold energy density vs. number of laser pulses atdifferent pulse durations. The threshold energies for 5 ms arederived from Roider et al. [9] and represent the thresholdenergies for angiographic visibility. The threshold energiesfor 160 ms are derived from Roider et al. [16] and representthe threshold values for occlusion of venules and arterioles.The threshold energies for 200 ns represent the current ex-periments for angiographic visibility. Note that the correla-tion factor r, which describes the dependence between energynecessary for multiple pulses vs. a single pulse (see text), is onthe order of −0.2 where a thermal mechanism seems to pre-vail, as opposed to a factor of −0.13 where a pure mechanicalmechanism seems to prevail.

246 Roider et al.

repetition rate is only 500 Hz [14]. Therefore it issufficient to consider the temperature profile af-ter a single laser pulse. The temperature calcula-tions show that absolute temperatures of at least100°C to 160°C may prevail in the energy rangewhere histological examination was performed(3–6 mJ respectively 0.04–0.08 J/cm2). This tem-perature profile is basically built up every 2 ms,because the repetition rate is 500 Hz. Each singlemelanin granule has neighboring melanin gran-ules, which could additionally contribute to thereal temperature profile of the granule of interest,since there is some small heat conduction. A com-parison of the analytical results of the multiplegranule model [14] shows that the real tempera-ture profile of a single melanin granule sur-rounded by neighboring melanin granules with anaverage distance of 1.2 mm in the RPE is nothigher than 30% than the temperature due tosingle absorption of a single granule [14]. The dis-tance of 1.2 mm is the average distance of melaningranules inside the RPE [13]. Therefore it is rea-sonable to consider only the temperature profilearound a single melanin granule. As summary thetemperature calculations show that the tempera-tures inside the melanin granules exceed 100°Cwith 3 and 6 mJ pulse energy (0.04 and 0.08 J/cm2). The absolute temperatures are completelydifferent to those prevailing e.g. after a single 5-ms laser pulse, where significant heat conductionout of a single melanin granule occurs. At energydensities of 0.22 J/cm2, where macroscopic bubbleformation regularly appears with 200 ns, tem-peratures far beyond 100°C prevail. TheoreticallyHansen and Fine [22] thought that steam forma-tion around melanin granules around singlemelanin granules may take place. Gerstman [23]postulated a similar mechanism and tried to cor-relate the size and energy necessary for bubbleformation by analytical analysis. Depending onthe absorption of the melanin granules they pre-dict energy densities between 0.73 J/cm2 and 1.23J/cm2, which may lead to bubble formation. Thesebubbles should have a size of about 10 mm. Thesevalues are close to our experimental values,where macroscopic visible bubble formations havebeen detected. The fact that with 10 pulses only asingle tiny bubble was found, but with 500 pulsesand identical energy several adjacent bubbles ap-pear, is consistent with the theory of bubble for-mation inside the RPE. Gerstman [23] postulatedthat bubble formation can take place even atlower energies, but such situations cannot be ana-lytically calculated. Our experiments show many

signs which are consistent with the assumptionthat even at lower energies bubble formationaround single melanin granules may act as pri-mary interaction mechanism. A space betweenthe RPE and Bruch’s membrane as well a gapbetween the RPE and the outer segments was re-peatedly found. Additionally irregularly orien-tated outer segments were repeatedly found. Alsothe fact that even 2 hours after irradiation the cellnucleus still shows a membrane structure and isnot homogeneously coagulated is not consistentwith a pure thermal coagulation mechanism. Allthese findings could be explained by bubble for-mation around single melanin granules. It may bethat even damaged RPE cells could be pushedapart, as found by TEM (see Fig. 3). It is unclearwhether damage takes place only inside thebubble or outside such a bubble as well. The al-tered and disorientated outer segments could be ahint that damage may take place even outside thebubble. The histological findings of the alteredouter and inner segments could be explained e.g.by compression outside of the bubble.

In summary all findings found in these ex-periments do not support a primary thermal co-agulation mechanism. Both the macroscopic find-ings, the histological findings as well as thethreshold values and the temperature calcula-tions are consistent with the theory of a primarybubble formation interaction mechanism.

REFERENCES

1. The Diabetic Retinopathy Study Research Group. Reportno. 8. Invest Ophthalmol Vis Sci 1981; 88:583.

2. Gabel V-P, Birngruber R, Hillenkamp F. Visible and nearinfrared light absorption in pigment epithelium and cho-roid. In: Shimizu K, Oosterhuis JA, eds. Internat CongrSeries No. 450, XXIII Concilium Ophthalmologicum,Kyoto. Amsterdam, Oxford: Excerpta Medica, pp 658–662.

3. Arrhenius S. Uber die Reaktionsgeschwindigkeit bei derInversion von Rohrzucker durch Sauren. Z Phys Chem1889; 4:226–248.

4. Birngruber R, Hillenkamp F, Gabel V-P. Theoretical in-vestigations of laser thermal retinal injury. Health Phys1985; 48:781–796.

5. Marshall J. Thermal and mechanical mechanisms in la-ser damage to the retina. Invest Ophthalmol 1970; 9:97–115.

6. Marshall J, Mellerio J. Histology of retinal lesions withQ-switched lasers. Exp Eye Res 1968; 7:225–230.

7. Birngruber R, Puliafito CA, Gawande A, Wei-Zhu L,Schoenlein RW, Fujimoto J. Femtosecond laser-tissue in-teractions: Retinal injury studies. IEEE J QuantumElectr 1987; QE-23:1836–1844.

8. Lin CP, Kelly MW. Ultrafast time-resolved imaging of

Bubble Formation in Retinal Laser Exposure 247

stress transient and cavitation from short pulsed laserirradiated melanin particles. Proc SPIE 1995; 2391A:294–299.

9. Roider J, Hillenkamp F, Flotte T, Birngruber R. Micro-photocoagulation: Selective effects in biological tissue us-ing repetitive short laser pulses. Proc Natl Acad Sci USA1993; 90:8643–8647.

10. Anderson RR, Parrish JA. Selective photothermolysis:Precise microsurgery by selective absorption of pulsedradiation. Science 1983; 220:524–527.

11. Feeney L, Grieshaber JH, Hogan MJ. Studies on ‘‘humanocular pigment.’’ In: Rohen JW, ed. ‘‘The Structure of theEye.’’ Stuttgart: Schattauer-Verlag.

12. Roider J, Michaud N, Flotte T, Birngruber R. Response ofthe RPE to selective photocoagulation of the RPE by re-petitive short laser pulses. Arch Ophthalmol 1992; 110:1786–1792.

13. Birngruber R, Gabel V-P, Hillenkamp F. Experimentalstudies of laser thermal retinal injury. Health Phys 1983;44:519–531.

14. Roider J, Birngruber R. Solution of the heat conductionequation. In: Welch AJ, van Gemert M, eds. ‘‘Optical-Thermal Response of Laser Irradiated Tissue.’’ NewYork: Plenum, pp 385–409.

15. Vassiliades A. Ocular damage from laser radiation. In:Wolbarsht ML, ed. Laser Applications in Medicine andBiology. Vol 1. New York, London: Plenum, pp 125–162.

16. Roider J, Schiller M, El-Hifnawi E, Birngruber R. Reti-nale Photokoagulation mit einem frequenzverdoppeltenNd-Yag-Laser (532 nm). Ophthalmologe 1994; 91:777–782.

17. Roider J, Traccoli J, Michaud N, Flotte T, Anderson,Birngruber R. Selektiver Gefaßverschluß durch repetier-ende Laserpulse. Ophthalmologe 1994; 91:274–279.

18. Goldmann AI, Ham WT, Mueller HA. Ocular damagethreshold and mechanisms for ultrashort pulses of bothvisible and infrared laser radiation in the rhesus monkey.Exp Eye Res 1977; 24:45–56.

19. Glickman RD, Lam KW. Oxidation of ascorbic acid as anindicator of photooxidative stress in the eye. PhotochemPhotobiol (Engl) 1992; 55:191–196.

20. Glickman RD, Jaques SL, Schwartz JA, Lam KW, BuhrG. Photochemical reactivity of RPE melanosomes is in-creased after disruption by pulsed laser exposures. InvestOphthalmol Vis Sci 1996; 37 (Suppl.):375.

21. Doukas AG, Flotte TJ. Physical characteristics and bio-logical effects of laser-induced stress waves. UltrasoundMed Biol 1996; 22:151–164.

22. Hansen WP, Fine S. Melanin granule models for pulsedlaser induced retinal injury. Appl Optics 1968; 7:155–159.

23. Gerstman BS, Thompson CR, Jaques SL, Rogers ME. La-ser induced bubble formation in the retina. Lasers SurgMed 1996; 18:10–21.

248 Roider et al.