2.1.

Ex vivo RPE model

RPE explants were used for this study. Freshly enucleated porcine eyes from the local abattoir were dissected. The RPE samples had a size of approximately $1{\mathrm{cm}}^2$ containing retina, choroid, and sclera. A few drops of isotonic $\mathrm{NaCl}$ solution were added. After a few minutes, the neural retina can be easily detached without damage to the underlying RPE cells. The neural retina was carefully removed to expose the RPE layer. After irradiation of the RPE explants, a fluorescence viability assay was performed with diluted Calcein-AM (Molecular Probes, 1:500 in phosphate-buffered saline without $\mathrm{Ca}$ , $\mathrm{Mg}$ , $20\mathrm{\mu }\mathrm{l}$ per RPE sample). Calcein-AM can pass the cell membrane. Living cells metabolize nonfluorescing Calcein-AM to fluorescing calcein (excitation maximum: $490\mathrm{nm}$ ; emission maximum: $520\mathrm{nm}$ ), which is unable to diffuse out of the cell as long as the cell membrane integrity is maintained. After an incubation time of approximately $30\min$ the explant can be analyzed under the fluorescence microscope: Living cells with an intact membrane appear bright.

2.2.

Experimental Setup

The experimental setup for irradiation of RPE explants is shown in Fig. 1. Bubbles can be detected simultaneously by interferometry and optoacoustics. A modified frequency doubled, $Q$ -switched neodymium: yttrium lithium fluoride $(\mathrm{Nd}:\mathrm{YLF})$ laser (Quantronix Inc. Model 527DP-H, $\lambda =527\mathrm{nm}$ ) served for irradiation. The laser pulse duration can be varied in the range from $200\mathrm{ns}$ to $5\mathrm{\mu }\mathrm{s}$ by an active feedback system controlling the intracavity losses during the laser pulse.²³ The pulses are transmitted by a fiber [Ceram Optec, Optron UV-A 105/125A/250, length $200\mathrm{m}$ , numerical aperture $\left(\mathrm{NA}\right)=0.22$ ]. The pulse energy can be adjusted with a variable attenuator. Spatially almost homogeneous irradiation, which is important for measuring threshold radiant exposures, is achieved by locating the optical image of the fiber tip and the RPE explants in the same plane. Because of the long fiber, the spatial modulation of the radiant exposure due to speckle is below 10% at the specimen. The laser spot size on the RPE explants is reduced using an elliptical field aperture. An elliptical spot with diameters of $49\mathrm{\mu }\mathrm{m}$ and $41\mathrm{\mu }\mathrm{m}$ is obtained on the RPE sample. For determination of the laser spot size, a slide micrometer scale was placed in the water-filled cuvette and the laser spot on the reflecting microscope slide was observed by the charge-coupled device (CCD) camera (Fig. 1). A photodiode (Centronics AEPX 65) with a current integrating circuit is used to determine the laser pulse energy, calibrated to an energy meter (Ophir Optronics, Laserstar PE 10).

The RPE tissue samples are mounted in a cuvette with isotonic $\mathrm{NaCl}$ solution on a three-axis translation stage. The explants are irradiated with single laser pulses and radiant exposures up to $700\mathrm{mJ}∕{\mathrm{cm}}^2$ . The laser pulse durations (full width at half maximum) used in the experiments are $350\mathrm{ns}$ and $1.7\mathrm{\mu }\mathrm{s}$ . Typical pulse shapes can be found in a previous publication on bubble nucleation around isolated RPE melanosomes.²⁰

A helium-neon $\left(\mathrm{HeNe}\right)$ laser (Laser Graphics LK 8623, $632.8\mathrm{nm}$ ) is used as the light source for the interferometer. The beam passes a 50/50 beamsplitter. A quartz glass plate serves as a mirror in the reference arm. The probe beam ( $\mathrm{HeNe}$ laser) is coupled into the irradiation setup by a dichroic mirror and is focussed by a lens (focal length $f=10\mathrm{mm}$ ) into the image plane of the fiber tip at the RPE explant. The Gaussian spot diameter on the sample is approximately $12\mathrm{\mu }\mathrm{m}$ . A power of about $1\mathrm{mW}$ is measured at the site of the RPE sample. Both laser spots (pump and probe beam) can be monitored and aligned on a reflecting test object using a CCD camera (Hamamatsu Photonics C5405). The sample arm of the interferometer is chopped with a duty cycle of $30\mathrm{\mu }\mathrm{s}$ : $125\mathrm{ms}$ synchronized to the $\mathrm{Nd}:\mathrm{YLF}$ laser pulses. This technique avoids heating of the RPE explants by the probe laser. The interference fringes are detected by a photodiode with an integrated preamplifier (Hamamatsu Photonics S6468-02), which has a $-3\text{-}\mathrm{dB}$ upper cutoff frequency of $35\mathrm{MHz}$ . The $\mathrm{Nd}:\mathrm{YLF}$ laser light, which is partially reflected from the RPE tissue, is blocked by a bandpass filter with a center wavelength of $632.8\mathrm{nm}$ in front of the photodiode. The interferometer is aligned by means of the quartz glass plate serving as mirror in the reference arm. The plate is mounted on a piezoelectric actuator and can be set to vibration to obtain an interference signal without any movement of reflecting interfaces in the sample arm. This configuration was used to optimize the interference amplitude before RPE irradiation. The piezoelectric actuator was switched off for the interferometric measurements during $\mathrm{Nd}:\mathrm{YLF}$ laser exposition.

Optoacoustic transients are recorded with an ultrasonic broadband transducer (Valpey-Fisher VP-1093, $\mathrm{DC}\text{-}10\mathrm{MHz}$ , Preamplifier Panametrics 5676). The acoustic transients provide information on the bubble incipience and collapse, which can be identified as pressure peaks.²⁴ The optoacoustic data are complementary to the interferometric transients, because a low signal-to-noise ratio is expected at bubble incipience and collapse for the interferometric measurements, when the amount of backscattered light is minimal due to the small bubble size. Therefore, the consistency of the interference data can be confirmed by the acoustic measurements. The distance between the RPE sample and the transducer was approximately $2.5\mathrm{mm}$ . Interferometric and optoacoustic transients are sampled with a digital oscilloscope (Tektronix TDS 540).

We irradiated RPE explants of different eyes for both pulse durations. The number of irradiated cells (including partly irradiated) per laser exposition is 5 to 12. This value was estimated after the viability assay (Sec. 2.1) by a comparison between the laser spot diameter and the size of the stained RPE cells as observed by fluorescence microscopy (e.g., Fig. 2 ). The number of dead cells per laser exposition was obtained by the viability assay described in Sec. 2.1. The ${\mathrm{ED}}_{50}$ radiant exposure for cell death, which indicates death to 50% of the irradiated cells, and the associated probit slopes $S={\mathrm{ED}}_{84}∕{\mathrm{ED}}_{50}$ were determined by standard probit analysis.^{25, 26}

Fig. 2

Left: Fluorescence microscope image taken after irradiation with a pulse duration of ${\mathrm{\tau }}_{laser}=350\mathrm{ns}$ . The laser pulse is emitted at $0\mathrm{\mu s}$ on the time axis. Dead RPE cells appear dark. The white ellipse represents the spot size of the $\mathrm{Nd}:\mathrm{YLF}$ laser beam on the RPE explant. Right: Corresponding interferometric transients at $118\mathrm{mJ}∕{\mathrm{cm}}^2$ [(a) no dead cells], at $125\mathrm{mJ}∕{\mathrm{cm}}^2$ [(b) 4 dead cells], and at $165\mathrm{mJ}∕{\mathrm{cm}}^2$ [(c) 5 dead cells]. The interference signal was ac coupled to the oscilloscope to eliminate the dc component ${\mathrm{I}}_0$ .

3.1.

Threshold for Cell Death

For $350\text{-}\mathrm{ns}$ pulses, 52 laser expositions were applied on 3 different RPE explants. The overall threshold radiant exposure for cell death was ${\mathrm{ED}}_{50}=129\pm 5\mathrm{mJ}∕{\mathrm{cm}}^2$ and the corresponding probit slope $S=1.13$ . A total of 146 laser expositions on 5 RPE explants resulted for $1.7\mathrm{\mu }\mathrm{s}$ pulses in ${\mathrm{ED}}_{50}=180\pm 5\mathrm{mJ}∕{\mathrm{cm}}^2$ and $S=1.14$ . Both ${\mathrm{ED}}_{50}$ radiant exposures are in good agreement with previous measurements reported by Brinkmann, ²³ who used similar laser pulse durations (Table 1 ). Sliney found that typical probit slopes for retinal damage have values between 1.05 and 1.15,²⁶ which is consistent with our results.

Table 1

ED50 radiant exposures for cell death at various pulse durations (λ=527nm) .

Representative interference amplitudes due to the irradiation with a laser pulse duration of $350\mathrm{ns}$ are shown in Fig. 2. The corresponding viability stained RPE cells are also depicted. Similar interferometric transients were obtained for a pulse duration of $1.7\mathrm{\mu }\mathrm{s}$ (data not shown). The three representative interferometer traces in Fig. 2 are recorded at $118\mathrm{mJ}∕{\mathrm{cm}}^2$ (Trace A: no dead cells), at $125\mathrm{mJ}∕{\mathrm{cm}}^2$ (Trace B: 4 dead cells), and $165\mathrm{mJ}∕{\mathrm{cm}}^2$ (Trace C: 5 dead cells). Figure 2 suggests that the threshold radiant exposure for cell death of this specific RPE explant is between $118\mathrm{mJ}∕{\mathrm{cm}}^2$ and $125\mathrm{mJ}∕{\mathrm{cm}}^2$ , which is consistent with the overall threshold value of ${\mathrm{ED}}_{50}=129\pm 5\mathrm{mJ}∕{\mathrm{cm}}^2$ for $350\text{-}\mathrm{ns}$ pulses. Interference fringes caused by random mechanical vibration throughout the experimental setup are in the millisecond domain. These random fringes do not influence the interferometric transients resulting from laser-induced RPE effects, which last only a few microseconds [Figs. 2b, 2c]. At a radiant exposure increase of only 6% in Fig. 2b $(125\mathrm{mJ}∕{\mathrm{cm}}^2)$ compared to Fig. 2a. $(118\mathrm{mJ}∕{\mathrm{cm}}^2)$ , a transient intensity modulation is detected in the interference signal as opposed to sublethal irradiation in Trace A. With increasing radiant exposure, an amplitude modulation with frequencies of more than $10\mathrm{MHz}$ can be observed in the interferometric transients during and after the laser pulse [Fig. 2c].

Since the bubble nucleation temperature at the strongly absorbing RPE melanosomes was determined to be $150°\mathrm{C}$ ,²⁰ a maximum laser-induced temperature increase without bubble formation of

All interference traces were processed for quantification by the LabVIEW environment (LabVIEW 6i, National Instruments). Each transient $x_{IF}\left(t\right)$ was digitally bandpass filtered $(300\mathrm{kHz}$ to $30\mathrm{MHz})$ to eliminate low frequency interferometer fringes due to random vibration of the setup and high frequency electronic noise. Subsequently, the signal energies $E_{IF}$ were calculated according to

Fig. 3

Signal energies $E_{IF}$ of the interferometric transients plotted with respect to the applied radiant exposure [(a) ${\mathrm{\tau }}_{laser}=350\mathrm{ns}$ , (b) ${\mathrm{\tau }}_{laser}=1.7\mathrm{\mu }\mathrm{s}$ ]. The signal energy $E_{IF}$ strongly increases at the occurrence of cell death.

Fig. 4

Signal energies $E_{IF}$ of the interferometric transients plotted over the number of dead cells per laser exposition [(a) ${\mathrm{\tau }}_{laser}=350\mathrm{ns}$ , (b) ${\mathrm{\tau }}_{laser}=1.7\mathrm{\mu }\mathrm{s}$ ]. The chosen threshold of the signal energy $E_{IF}=4.3$ , which serves as criterion for cell death, is plotted as a horizontal line. Data points corresponding to the traces depicted in Fig. 2 are labeled.

3.2.

Bubble Dynamics

In Fig. 5 , representative interferometric and optoacoustic transients resulting from an irradiation four times above the ${\mathrm{ED}}_{50}$ threshold for cell death are depicted. The laser pulse duration was $350\mathrm{ns}$ . According to Eq. 1, every full period in the interference signal [Fig. 5a] corresponds to a change in the optical path difference of $2n_{{\mathrm{H}}_2\mathrm{O}}\mathrm{\Delta }{z}_S=\lambda (\lambda =632.8\mathrm{nm})$ , that is, to a bubble interface displacement of half a probe laser wavelength in water. In order to check for internal interference effects in the sample arm (e.g., between the front and the back interface of a bubble, which can act as a Fabry-Perot interferometer³⁰), the light in the reference arm of the interferometer was blocked, temporarily. In that case, no interference fringes were detected.

Fig. 5

Bubble formation due to a radiant exposure of $494\mathrm{mJ}∕{\mathrm{cm}}^2$ (about $4\times$ above the ${\mathrm{ED}}_{50}$ threshold for cell death). The laser pulse with a duration of ${\mathrm{\tau }}_{laser}=350\mathrm{ns}$ is also plotted. (a) Interferometric transient. (b) Optoacoustic transient corrected by its acoustic transit time to the hydrophone. Note that the interference amplitude in Fig. 5a is 10 times larger than in Fig. 2 displaying the same arbitrary units (a.u.).

At radiant exposures of more than about two times above ${\mathrm{ED}}_{50}$ threshold for cell death shortly after the laser pulse [Fig. 5a: 0 to $4\mathrm{\mu }\mathrm{s}$ ] only irregular interference fringes without a well-defined (instantaneous) frequency were observed [comparable to Fig. 2c]. In this time interval, the motion of a bubble interface cannot be determined from the interferometric transients. Fast flash photography of irradiated RPE cell fragments reveals²⁹ that many microbubbles nucleate around individual melanosomes, which results in a rough reflecting interface. As a consequence, the interferometer signals of all individual bubbles are superimposed with nonstationary phase differences and are spatially averaged on the photodiode.

After this period of time, well-defined interference fringes can be observed [Fig. 5a: $t>4\mathrm{\mu }\mathrm{s}$ ], because a smooth reflecting interface has formed, that is, the microbubbles have coalesced to a larger bubble and its surface is smoothed by the surface tension. This process of bubble coalescence was observed during fast flash photography of RPE cell fragments.²⁹ In this case, the interferometric approach allows monitoring of the coalesced bubble’s dynamics with a high temporal resolution, which is demonstrated in Ref. 28: The initial frequency decrease in the interference trace [Fig. 5a: 4 to $8.2\mathrm{\mu }\mathrm{s}$ ] indicates that the bubble expansion becomes slower. Simultaneously, the interference fringes show increasing amplitudes, because the area from which light can be reflected is increasing with bubble size. The interferometric transient also shows a small increase of the baseline due to this increasing reflecting area. A turning point in the interference trace can be found at $8.2\mathrm{\mu }\mathrm{s}$ after the laser pulse. This point marks the onset of the coalesced bubble’s contraction. The contraction velocity of the bubble increases, which can be deduced from the growing modulation frequency. The vanishing of the interference fringes at $12.5\mathrm{\mu }\mathrm{s}$ [Fig. 5a] coincides with a peak in the transit time corrected pressure signal [Fig. 5b]. Such a pressure peak is typically emitted at the bubble collapse.²⁴ These simultaneous events confirm that the interference fringes originate from the dynamics of the coalesced bubble.

Furthermore, it can be deduced from the point when the laser pulse is applied (and the microbubbles begin to grow, $t\approx 0\mathrm{\mu }\mathrm{s}$ ), the turning point at $t=8.2\mathrm{\mu }\mathrm{s}$ and the bubble collapse at $t=12.5\mathrm{\mu }\mathrm{s}$ in Fig. 5a that the bubble expansion takes longer than the bubble contraction. Well-defined interference fringes were observed for radiant exposures above $250\mathrm{mJ}∕{\mathrm{cm}}^2$ at a laser pulse duration of $350\mathrm{ns}$ [Fig. 5a]. For $n=57$ interference traces, which were recorded for radiant exposures in the range between $250\mathrm{mJ}∕{\mathrm{cm}}^2$ and $550\mathrm{mJ}∕{\mathrm{cm}}^2$ , the mean ratio of the bubble collapse time ${\tau }_c$ to the bubble lifetime ${\tau }_{bubble}$ was ${\tau }_c∕{\tau }_{bubble}=0.32\pm 0.04$ .

Both techniques, optoacoustics and interferometry, were used to determine the bubble lifetimes at radiant exposures near the threshold irradiation for cell death. Figure 6 shows interferometric and transit time corrected optoacoustic traces. The bubble lifetime can be clearly deduced from the pressure peaks occurring at bubble nucleation and bubble collapse in the optoacoustic transient. Such a transient bubble produces high frequency oscillations of about $10\mathrm{MHz}$ in the interference trace. After the bubble has collapsed, which can be seen in the transit time corrected acoustic transient, this region of high frequency oscillations is sometimes followed by low frequency oscillations in the range of $300\mathrm{kHz}$ up to $5\mathrm{MHz}$ [Fig. 6b, Fig. 2c ${\tau }_{bubble}=3\mathrm{\mu }\mathrm{s}$ ; low frequency oscillations: $t>3\mathrm{\mu }\mathrm{s}$ ; Fig. 5a: $t>12.5\mathrm{\mu }\mathrm{s}$ ]. These low frequency interference fringes might be produced by small permanent residual bubbles³¹ or by surface waves of the RPE cell layer originating from the bubble collapse. However, bubble lifetimes, which were extracted consistently from the interferometric and the optoacoustic measurements (e.g., Fig. 5, Fig. 6), are plotted in Fig. 7 as a function of the normalized applied radiant exposure. For bubble lifetimes below 2 to $3\mathrm{\mu }\mathrm{s}$ , no well-defined bubble collapse can be observed in the optoacoustic signal. Only in this case was the bubble lifetime solely determined by interferometry. The transition to well-defined interference traces might indicate that major bubble coalescence occurs at bubble lifetimes of more than 2 to $3\mathrm{\mu }\mathrm{s}$ . The graph in Fig. 7 shows an approximately linear increase of the bubble lifetime with radiant exposure in the investigated range. The ${\mathrm{ED}}_{50}$ thresholds for cell death are different for $350\mathrm{ns}$ and $1.7\mathrm{\mu }\mathrm{s}$ laser pulses (Sec. 3.1). However, after normalizing the radiant exposure to these ${\mathrm{ED}}_{50}$ radiant exposures for both pulse durations, no major differences in the bubble lifetimes between both pulse durations remain.

Fig. 6

Representative interferometric and transit time corrected optoacoustic transients obtained at radiant exposures of $197\mathrm{mJ}∕{\mathrm{cm}}^2$ (a) and $183\mathrm{mJ}∕{\mathrm{cm}}^2$ (b), which are slightly above the ${\mathrm{ED}}_{50}=180\pm 5\mathrm{mJ}∕{\mathrm{cm}}^2$ for cell death (Sec. 3.1). The laser pulse, which is plotted at the bottom of each graph, has a duration of $1.7\mathrm{\mu }\mathrm{s}$ . The bubble lifetimes can be deduced consistently from the pressure peaks in the optoacoustic transient and the high frequency oscillations in the interferometric transient. (b) Sometimes low frequency oscillations can be observed in the interference trace after the bubble collapse $(t>4\mathrm{\mu }\mathrm{s})$ .

Fig. 7

Bubble lifetimes obtained from interferometric and optoacoustic measurements as a function of the applied radiant exposure, which is normalized to the ${\mathrm{ED}}_{50}$ threshold irradiation for cell death (Sec. 3.1).

3.3.

Extent of RPE Damage

The number of dead cells per laser exposition is obtained from the viability assays and is plotted in Fig. 8 with respect to the applied radiant exposure and the bubble lifetime, respectively. An increasing number of dead cells, which describe the lateral damage range in the RPE, can be observed when applying higher radiant exposures. As a simple approximation all measurements were interpolated by a linear fit. This provides a better comparison between the results for different pulse durations. For both pulse durations, $350\mathrm{ns}$ and $1.7\mathrm{\mu }\mathrm{s}$ , more than 12 cells are killed per laser exposition above about 1.5-fold ${\mathrm{ED}}_{50}$ threshold for cell death and for bubble lifetimes longer than 4 to $6\mathrm{\mu }\mathrm{s}$ (Fig. 8). In that case, the cell damage extended beyond the irradiated area. Note that a bubble lifetime of $5\mathrm{\mu }\mathrm{s}$ corresponds according to the Rayleigh formula for inertia limited bubble growth²¹

Fig. 8

(a1) and (b1): Number of dead RPE cells plotted as a function of the radiant exposure. The gray bar in each graph marks the range of the number of irradiated cells (including partly irradiated) per laser exposition ((a) ${\mathrm{\tau }}_{laser}=350\mathrm{ns}$ , (b) ${\mathrm{\tau }}_{laser}=1.7\mathrm{\mu }\mathrm{s}$ ). (a2) and (b2): Dead RPE cells as a function of the bubble lifetime. For a better comparison between both pulse durations, all measurements were interpolated by a first order fit as a simple approximation.

4.1.

Threshold for Cell Death

The detection of microbubbles, as performed here by interferometry, strongly correlates to the occurrence of cell death determined by the viability assay (Fig. 4). This confirms previous experiments, in which microbubbles in the RPE were detected acoustically¹⁷ and reflectometrically.¹⁶ In these studies, microbubble formation was also required for the occurrence of cell death, if the pulse duration was below the range of 20 to $50\mathrm{\mu }\mathrm{s}$ .^{16, 17} These findings support the hypothesis that microbubble formation is the origin of cell death for the pulse durations applied in this experiment.

Because the applied laser pulse durations do not completely fulfill the condition of thermal confinement for the melanosome, an appreciable amount of heat diffuses during irradiation from the strongly absorbing melanosomes to the aqueous surrounding.^{20, 23} As a consequence, a higher radiant exposure is required for longer pulses to reach the nucleation temperature.^{20, 23} The threshold for cell death increases with the laser pulse duration in the investigated range (Table 1).

4.2.

Bubble Dynamics and Damage Range

In contrast to optoacoustics and reflectometry, interferometry can provide information on the motion of the bubble interface. Especially the point in time when the maximum bubble extension occurs [Fig. 5a] can be determined. These data revealed that the expansion of laser-induced bubbles in the RPE, which includes the coalescence of microbubbles, takes longer than its contraction. This temporal asymmetry was also confirmed by time-resolved microscopy during irradiation of isolated RPE cell fragments;²⁹ whereas typically the dynamics of a single laser-induced cavitation bubble in a fluid without solid boundaries are temporally symmetrical.³² The temporally symmetric spherical bubble dynamics go along with a minimal expansion and collapse time.²⁴ If the bubble collapse is disturbed, for example, by a solid boundary near the bubble, the collapse lasts longer than the expansion.²⁴ Because the bubble expansion in the RPE lasts longer than the collapse, the reason for the temporally asymmetrical bubble dynamics might be the coalescence of microbubbles after nucleation, which disturbs the bubble expansion. Because temporal unsymmetrical bubble dynamics were also observed in isolated RPE cell fragments in aqueous suspension,²⁹ it is very unlikely that the elastic-plastic properties of the neighboring tissue are responsible for the temporal asymmetry of the bubble in the RPE explants.

The selective RPE damage was originally thought to be a thermal effect,⁷ and therefore, microsecond laser pulses were preferred for clinical SRT.³³ Recent investigations^{16, 17} and these experiments indicate that bubble formation is responsible for the cell damage. In this context, the question emerged whether thermomechanical damage owing to bubble expansion such as choroidal hemorrhage and disruption of the neuroretina might occur when using shorter laser pulses for SRT. This study shows that the bubble lifetime increases almost linearly with increasing radiant exposure (Fig. 7). As far as bubbles have coalesced and a well-defined collapse can be observed in the optoacoustic transients ( ${\tau }_{bubble}\gtrsim 2$ to $3\mathrm{\mu }\mathrm{s}$ ), it can be assumed that the measured bubble lifetime is correlated to the bubble size.^{27, 28, 29} So, the damage range in the tissue, which is determined by the bubble size, finally depends on the bubble lifetime. Figure 8 also supports this correlation. The maximum bubble size, which still avoids irreversible photoreceptor damage, is unknown. However, a major difference in the bubble lifetime between the two investigated pulse durations $(350\mathrm{ns},1.7\mathrm{\mu }\mathrm{s})$ was not found (Fig. 7). From this point of view, both pulse durations should be equally suitable for SRT assuming that the bubble dynamics are transferable from porcine RPE explants to human RPE in vivo. Animal studies with rabbits by Framme ³⁴ also support the hypothesis that laser pulses of a few hundred nanoseconds are appropriate for SRT. Framme and coworkers defined a safety range for SRT by the ratio of the radiant exposures for 84% probability of fluorescein angiographic to 16% probability of ophthalmoscopic visibility of the laser expositions. Ophthalmoscopic visibility indicates photoreceptor damage; whereas angiographic visibility is an indicator for RPE barrier damage. They found that this safety range is even larger for $200\text{-}\mathrm{ns}$ pulses than for $1.7\text{-}\mathrm{\mu }\mathrm{s}$ pulses.

4.3.

Dosimetry Control for SRT

The RPE damage, which is the primary process of SRT, goes along with the formation of transient laser-induced bubbles. Because the threshold for laser-induced RPE damage varies inter- and intraindividually and because there might be only a small safety range for SRT, a dosimetry control for SRT is required to avoid collateral damage, for example, to the photoreceptors. If microbubble formation can be monitored at the ${\mathrm{ED}}_{50}$ threshold for cell death, it can be used as criterion for RPE damage. We calculated $E_{IF}$ and compared this value to a threshold level of $E_{IF}=4.3$ . This algorithm was capable of discriminating between the occurrence of cell death and sublethal laser exposure in more than 90% of the applied laser expositions when defining a proper threshold value for $E_{IF}$ (Fig. 4). Thus, such an interferometer might serve as an online dosimetry control during SRT. Movements of the patient’s eye during SRT should cause no major complications for the interferometric bubble detection, because these are of the millisecond rather than the microsecond time range. For slit lamp adaptation, the interferometer setup can be implemented as a fiber interferometer, which minimizes the effect of random vibration on the interferometer. The probe laser could serve simultaneously as an aiming laser for SRT. Because the microbubbles around the melanosomes nucleate during the laser pulse,²⁰ even the SRT treatment laser might serve simultaneously as a probe laser.

The detection of bubbles by interferometry should be more sensitive than measuring only the increase of the light, which is backscattered from the bubble interfaces,^{16, 19} because a fast surface motion in the order of the probe laser wavelength without any change of the reflected intensity is sufficient for the occurrence of interference fringes. Additional noncoherent ambient light should not significantly influence the interferometric measurement, because it does not contribute to the interference.

However, interferometric measurements in vivo might be complicated by photoreceptor tissue in front of the RPE, which scatters rather than absorbs the incident light.³⁵ In this case, the choice of the interferometer wavelength can be optimized toward longer wavelengths in order to reduce scattering at the photoreceptors. A decreased signal-to-noise ratio of the detected interferometric transients is expected when the RPE is located out of the focus plane of the probe laser, because less reflected light is collected by the interferometer optics. A coherence length of the interferometer laser, which is in the range of a few millimeters, might also help to increase the reliability the interferometric bubble detection by avoiding unwanted interference effects between other reflecting interfaces than the bubble surface (e.g., at the surface of the cornea). However, a difference in the optical path length between the sample arm and the reference arm of the interferometer, which is shorter than the coherence length of the probe laser, is required.

In contrast to the optoacoustic detection of microbubbles, which has already been evaluated as a dosimetry control for SRT,⁹ interferometry works in a noncontact mode. For this reason it can also be implemented in noncontact treatment modalities for SRT such as in a scanning digital ophthalmoscope. Furthermore, the optoacoustic dosimetry control depends on pulse-to-pulse fluctuations of the optoacoustic tissue response during irradiation with a burst of laser pulses. As opposed to optoacoustics, the interferometric technique should also work when applying single laser pulses as shown in Fig. 4. The spatial resolution of the interferometer is only limited by the spot size of the probe laser. Therefore, this technique should be applicable to small spot sizes, for which the optoacoustic signal-to-noise ratio is not sufficient for the detection of a few or even a single transient mircobubble. The interferometric bubble detection might also be appropriate for SRT performed by a fast scanned continuous wave laser with spot diameters of only $20\mathrm{\mu }\mathrm{m}$ .^{36, 37}