1.

Introduction

Retinal photocoagulation was introduced in the 1950s by Meyer-Schwickerath, and its application to seal retinal tears has become a clinical standard.¹ By the 1990s, laser photocoagulation has become the standard of care for various indications according to large clinical trials, most prominently the early treatment of diabetic retinopathy study (ETDRS),^2,3 the branch vein occlusion study group,⁴ and the collaborative central retinal vein occlusion study.⁵

Recently, photocoagulation has been challenged by intravitreal drug administration, but has remained the gold standard of treatment of peripheral ischemic retinal conditions such as diabetic retinopathy and ischemic vein occlusion.⁶ At the macular region, it still represents a first-line or adjunctive therapy in focal diabetic maculopathy⁷ and branch vein occlusion. Major advantages of photocoagulation include availability, low cost, and easy handling, making it the therapy of choice even in cases where intravitreal injections may promise minimally better vision. Recent improvements of photocoagulation aim at minimized tissue destruction,^8,9 improved patient comfort,¹⁰ and increased application performance.^11,12 Pilot studies have demonstrated that photocoagulation has not yet been developed up to its best potential.^13–15 This work targets a method which automatically generates uniform lesions by controlling the dosage for every individual photocoagulation spot, with pre-selectable strength.

For retinal photocoagulation, typically a laser power of 50 to 500 mW is applied for an irradiation time of 20 to 200 ms onto a spot diameter ranging over 50 to 500 μm.^{2–4,16–20} It is mainly performed with laser radiation in the green spectral range because of its high absorption at the retinal pigment epithelium (RPE) and choroid. Owing to heat diffusion the adjacent retinal layers are also thermally damaged. The lesions become visible as whitish/grayish lesions under white light visual ophthalmoscopic examination because of increased light scattering after the onset of denaturation. The visibility of the lesions is used as post irradiation dosage control for successful application. In a typical laser lesion irreversible thermal denaturation of the outer and inner retinal segments and the choroid is found.²¹

The therapeutic idea behind photocoagulation depends on the targeted disease. While artificial scar production to prevent retinal detachment is undisputable, the therapeutic effect following panretinal coagulation in diabetic retinopathy is still debatable. The most popular and accepted theory of Wolbarsht and Landers postulates that by denaturation of the photoreceptor cells in the outer periphery the overall oxygen demand, and therefore the neovascularisation in diabetes is reduced. This leads to a stabilized metabolism in the central area which is spared from coagulation;²² however, the necrosis of the inner neural retina should be avoided in any case.

Varying transmission²³ and RPE/choroidal pigmentation^24,25 of a treated eye account for significant effect variations even at a constant laser power. Hence, laser power needs to be permanently observed for ophthalmoscopical whitening and must be adjusted accordingly. Ophthalmoscopically visible whitening of a lesion takes up to a whole day to fully develop. During a treatment session of several hundred lesions, it is thus not feasible to wait long enough to finally judge every single lesion. Long exposure times of 200 ms grant a wide therapeutical window, but poor axial confinement of a lesion, and the undesired coagulation of the inner retinal layers is likely.²⁶ In order to reduce the thermal damage caused by heat diffusion, newer approaches reduce the irradiation time down to 10 ms;²⁶ however, this approach increases the risk of choroidal rupture and bleeding because fast heat deposition can lead to disruption if the vaporization temperature is exceeded. Thus, the effective therapeutic window, which is defined as the ratio of threshold power for rupture to the threshold power for coagulation, decreases.^26,27 Nevertheless, this approach has already led to reduced pain for the patient, and a reduced treatment time in combination with patterned coagulation.²⁸ In summary, slow occurrence of whitening and the conflicting needs to spare the neuroretina while retaining a wide therapeutic window all result in unavoidable lesion inhomogeneity as well as over- and under-effects of conventional photocoagulation dosage.²⁹ In the worst case, full thickness retinal coagulation causes extended scotoma. Retinal ruptures and choroidal bleedings lead to extensive, unnecessary tissue destruction with a risk of secondary choroidal neovascular membrane formation. This problem has for a long time been recognized³⁰ and first attempts to control the photocoagulation strength by regulating the power of the treatment laser by the onset of light scattering were investigated in the 1980s.³¹

The extent of coagulation depends on the temperature increase and the time of temperature elevation during the irradiation period. It can be described by the thermal damage theory of Arrhenius.³² The temperature rise finally depends on the amount of absorbed light; thus, it not only depends on the laser settings, but also on the retinal/choroidal pigmentation, which can vary by up to a factor of four in humans.³³ Further, the amount of light scattering within the eye globe is unknown, and mainly depends on the patient’s lens condition, which cannot be individually quantified.²³ In case of a treatment time exceeding several seconds, as in transpupillary thermal therapy (TTT), the cooling effect of the perfusion has to be taken into account as well.³⁴

In this project, an optoacoustic (OA) method is used to determine the temperature rise during photocoagulation, with the ultimate goal to control the coagulation process by the temperature course. The ability to measure temperature changes by optoacoustics has first been described for soft tissue heating in vitro,³⁵ and in patients’ eyes during selective laser therapy of the retina.³⁶ This technique was further improved and quantified in the following years.^34,37–39 The temperature threshold required to thermally damage the RPE in the time range of 0.1 to 1 sec was recently determined by Denton et al. to $53\pm 2°\mathrm{C}$.⁴⁰ It can be assumed that the required temperature to damage the adjacent photoreceptors is similar. However, because the photoreceptors only absorb a few percent of the incident light, heat diffusion from the RPE is required to achieve these temperatures. Consequently, the RPE needs to be heated higher. The technique for real-time optoacoustic temperature measurements was investigated on tissue samples for thermal therapy^21,22 as well as fundus tissue for TTT. For the latter, a typical irradiation time of one minute on large spot diameters of 2 to 3 mm was used, and the capability and accuracy of this approach was demonstrated on rabbits; however, it has not been proven in clinic trials yet.³⁴ In order to use this method for photocoagulation with its much shorter irradiation times and smaller spot diameters, the TTT-setup needed to be modified. In this paper, the background and technique of OA temperature monitoring will briefly be reviewed, and the very first measurements on patients will be presented and discussed.

2.

Optoacoustic Pressure Generation

If a collimated laser beam with a top hat beam profile and a pulse duration ${\tau }_p$ smaller than the thermal confinement time (negligible heat diffusion) is applied onto a homogeneous absorber with the absorption coefficient ${\mu }_a$ and negligible light scattering, the energy density $E(z)$ is induced within the absorber. $E(z)$ is proportional to the absorption coefficient ${\mu }_a$ and the radiant exposure $H$, and decreases exponentially according to Beer’s law of absorption in the axial $z$-direction:

The absorbed energy leads to a temperature increase $\mathrm{\Delta }T(\overrightarrow{r},t)$, which can be described with the heat diffusion equation containing $\rho$ as the density, $\kappa$ as the thermal diffusivity, and $C_p$ as the heat capacity at constant pressure of the absorber. For limited cases, the equation can be solved analytically with a model described in detail by Birngruber:³²

For times shorter than the thermal confinement time, Eq. (2) can be integrated, and an instantaneous temperature rise $\mathrm{\Delta }T(z)$ is obtained according to:

If the pulse duration further fulfils the conditions of acoustic confinement (no pressure balance during heating), the temperature rise leads to a pressure increase inside the absorber because in most materials the density decreases with temperature.⁴¹

The pressure rise $P(z)$ is proportional to the energy density $E(z)$ and the temperature-dependent, dimensionless Grüneisen parameter $\mathrm{\Gamma }(T)$, which contains the thermal expansion coefficient $\beta$, the speed of sound $c_s$ and the heat capacity $C_p$. If the laser pulse duration ${\tau }_p$ is longer than the acoustic transit time but shorter than the thermal relaxation time, an average pressure $\overline{P}$ is obtained inside the absorber, which depends on the ratio of pulse duration ${\tau }_p$ to acoustic transit time ${\tau }_{\mathrm{ac}}$.⁴²

The pressure increase leads to an expansion of the absorbing volume proportional to $\mathrm{\Gamma }(T)$ and the absorbed pulse energy $E_{\mathrm{abs}}$ over the beam area $A$. As a consequence, a thermoelastic, bipolar pressure wave is generated. The pressure wave emission and propagation can be described by the photoelastic wave equation with $\mathrm{\Psi }$ as the velocity potential, and $\mathrm{\Delta }E(\overrightarrow{r},t)$ the energy density:⁴¹

The solution of Eq. (6) is an emerging pressure wave that varies in amplitude and phase over the regarded volume in front of the target. The wave depends on the shape and volume of the absorber.

In case of retinal irradiation, the pressure wave needs to propagate through the eye globe before it can be detected at the cornea. The wave amplitude is attenuated during propagation through the eye by absorption, and acoustic impedance mismatches at the lens, iris, and cornea. Its frequency spectrum is also altered owing to the increasing water absorption for frequencies higher than about 10 MHz. Finally upon detection, the shape, dimensions, material, sensitivity, frequency response, and amplification of the transducer influences the conversion of acoustic to electric energy. However, for every individual source at the retina, the detected peak pressure amplitude $p_{\max }$ is proportional to the product of $\mathrm{\Gamma }(T)·E_{\mathrm{a}\mathrm{b}\mathrm{s}}$:

3.

Non-Invasive Real-Time Temperature Determination

3.2.

Retinal Tissue Characterization

In order to determine the tissue-specific parameters, $T_{\max }$ and $T_0$ for retinal tissue, the pressure amplitudes and corresponding temperatures were measured simultaneously on freshly enucleated porcine eyes as shown in Fig. 1.

Fig. 1

Setup to determine the characteristic retinal tissue parameters $T_0$ and $T_{\max }$. Enucleated porcine globes were heated from rear with a cw diode laser while the temperature at the retina was measured with thermocouples. Simultaneously, acoustic pressure transients were excited with nanosecond probe laser pulses. The pressure transients were detected with an ultrasonic pressure sensor integrated in the contact lens.

The eyes were fixed in a special holder which allowed laser heating from rear. A cw diode laser (LISA laser OHG) with a maximum power of 90 W at 940 nm was used for fast tissue heating. The IR-beam was expanded to a top hat diameter of 18 mm to achieve a uniform temperature rise at the central retinal region. Furthermore, thin thermocouples (Newport Electronics GmbH, HYP0-33-1-T-G-60-SMPW-M, tip diameter 200 μm) were pierced in the eye globe close to the center of acoustic probing in order to measure the actual fundus temperature. During heating with a temperature rise time of about $100°\mathrm{C}/\min$, probe laser pulses (CrystaLaser Inc., QG-523-1000: 523 nm, 3.5 μJ, 75 ns, 1 kHz) were applied with a top hat beam diameter of about 100 μm. Pressure transients were measured with an annular piezo ceramic transducer (PZT) with high sensitivity in the MHz frequency range (Medical Laser Center Lübeck GmbH), which was embedded in a contact lens (Mainster OMRA-S Focal/Grid) to be used to apply the laser beam into the eye. The transients were electrically amplified and processed with LabVIEW to calculate the temperature rise from the measured pressure amplitudes. Figure 2 shows the measured increase of the pressure amplitudes in dependence of the specimen temperature. The dashed lines in Fig. 2 are exemplary second order polynomial fits to individual heating curves, the solid line is the fit received by averaging over all single fits. The fits are extrapolated beyond the data range to determine the tissue-specific parameters, $T_0$ and $T_{\max }$. $T_0$ and $T_{\max }$ cannot be measured directly, because the tissue would undergo phase transitions at $T=0$ and above the onset temperature for coagulation, which would change the physical parameter associated with $\mathrm{\Gamma }(T)$ according to Eq. (4).

Fig. 2

Measured pressure amplitudes $p_{\max }$ over probe temperature determined by thermocouples. The data points are fitted with a second order polynomial (dashed curves: exemplarily individual fits, solid black curve: average over all specimens). Lower red curve: strategy to determine temperatures in a clinical setting (explanation within the text).

Figure 2 shows a significant pressure increase of 125% from 10°C to 60°C, and 25% from 37°C to 60°C. With a least square regression fit, $T_0$ and $T_{\max }$ were determined to be $T_0=-17.0\pm 5.0°\mathrm{C}$, and $T_{\max }=93.3\pm 15.6°\mathrm{C}$ as an average value with standard deviation of the parameters obtained from 18 porcine eyes.

3.4.

Average Temperature Rise in the Probed Volume and RPE Peak Temperature

During photocoagulation, probe and treatment radiation are applied through the same fiber, and the temperature is probed over the whole irradiated volume. This process needs to be addressed with respect to temperature monitoring. During tissue heating, a transiently changing spatio-temporal temperature profile builds up at the treatment location. To quantify the temperature evolution, Eq. (2) was solved with the thermal parameters of water. The characteristic parameters of human fundus absorption are shown in Table 1, and were taken from Hammer et al.⁴⁴ These data fit quite well to the few human data available in the literature, e.g. from Gabel et al.²⁴ Selected solutions for a spot diameter of 300 μm and a laser power of 40 mW applied to the retina are shown in a cross section in Fig. 3, and for $\mathrm{\Delta }T(t)$ and over $r$, $z$ and $t$ in Fig. 4. During the onset of irradiation within the first 100 μs, the temperature profile follows the absorption characteristics, which is modified by heat diffusion thereafter. The temperatures scale linearly with laser power for all locations and times. An equilibrium temperature profile is reached within a time frame of about 1 s, depending on the spot size and absorption profile.

Fig. 3

Calculations for different isotherms in the cross section of the chorioretinal junction after an irradiation of 100 ms for a laser power of 40 mW applied onto a spot diameter of 300 μm. The green color demarks the irradiated volume and the fading green indicates the light absorption towards choroidal depth. The temperature in this volume is probed. The red dot denotes the point of highest temperature $T_{\text{peak}}$, which is in the center of the spot inside the RPE.

Fig. 4

Calculated temperature profiles using a laser power of 40 mW onto a spot diameter of 300 μm. (a) Axial temperature profile (negative values: retina, zero: RPE surface, positive values: choroid) for times of 1,10 and 100 ms after the onset of heating. (b) Lateral temperature profile for times of 1,10 and 100 ms after the onset of heating. (c) Temporal temperature rise $\mathrm{\Delta }{T}_{\text{peak}}(t)$ at the center of the RPE.

Table 1

Model parameters for the human fundus as used for the temperature calculations.

	Thickness [μm]	μa [1/cm]	Absorption [%]
RPE	6	1204	51
Bruch’s membrane	4	0	0
Pigmented choroid	400	270	49

The peak temperature $T_{\mathrm{peak}}(t)$ is always achieved in the center of the spot at the level of the RPE. Often this peak temperature is of interest, because here the coagulation is expected to start. In this case, it can be regarded as a threshold temperature for thermal denaturation in case of just barely visible lesions. Since optoacoustics in this approach reveals the average temperature over the whole probed volume, a conversion from this average to the peak temperature is necessary. This can be treated mathematically. Therefore, at first the mean temperature across each depth layer $T_{\text{mean}}(t,z)$ for a certain beam radius $R$ is calculated:

Eq. (10)

$$T_{\text{mean}}(t,z)=\frac{1}{A}·{\int }_0^{2\pi }\mathrm{d}\varphi ·{\int }_0^{R}T(t,r,z)·r·\mathrm{d}r=\frac{2}{R^2}{\int }_0^{R}r·T(t,r,z)·\mathrm{d}r.$$

If we consider that the probe beam is exponentially attenuated towards penetration depth according to Eq. (1) and as sketched in Fig. 3, then the average probed temperature in the irradiated volume $T_{\text{volume}}(t)$ must also be weighted exponentially:

Eq. (11)

$$T_{\text{volume}}(t)={\mu }_a{\int }_0^{\infty }{T}_{\text{mean}}(t,z)·e^{-z·{\mu }_a}·\mathrm{d}z.$$

The average weighted temperature $T_{\text{volume}}(t)$, which is probed, depends on the time and the absorption, and is equal to the optoacoustically determined temperature: $T_{\mathrm{OA}}(t)=T_{\text{volume}}(t)$. The relation between the peak temperature $T_{\text{peak}}(t)$ in the center of the spot at the RPE ($r=0$, $z=0$) and $T_{\mathrm{OA}}(t)$ is given by the time-dependent conversion function $f(t)$:

Eq. (12)

$$f(t)=\frac{T_{\text{peak}}(r=z=0,t)}{T_{\mathrm{OA}}(t)}.$$

Figure 5 shows $f(t)$ for two different radii (50 and 150 μm). As long as the radiation is completely absorbed within the RPE and choroid, and the absorption ratio between RPE and choroid keeps constant, $f(t)$ does not depend on the absorption coefficient and the laser power. This is important for the applicability and reliability of optoacoustic temperature determination because absorption varies inter- and intra-individually for each laser spot.

Fig. 5

Calculated conversion functions $f(t)$ as the ratio of the peak temperature in the center of the RPE, $T_{\text{peak}}(t)$, and the mean weighted temperature of the probed volume $T_{\text{volume}}(t)$, which is equal to the optoacoustically obtained temperature $T_{\mathrm{OA}}(t)$. Calculations for laser spot diameters of 100 and 300 μm are shown.

4.

Clinical Study and First Clinical Results

The study comprised 20 patients with a clinical indication for panretinal and sometimes central photocoagulation. The study was conducted to determine the retinal temperature rise during photocoagulation. Diseases to treat included diabetic retinopathy, retinal vein occlusion, and occlusive retinal vasculitis. Each patient’s body temperature was measured before the laser treatment. Irradiation was performed with defined laser settings inside (100 μm, 100 ms, gentle whitening) or outside the arcades (100 or 300 μm, 20 to 200 ms, power titration around threshold power) of the fundus. Twenty to 40 study lesions were applied and evaluated by fundus photography and OCT (Heidelberg Engineering, Spectralis) after 1 h, and at different times thereafter. Afterwards, the treatment was continued according to clinical routine requirements. All data were processed and evaluated after the treatment was completed. The clinicians had no access to the recorded OA data during treatment. The study had been reviewed and approved by the institutional ethics committee.

A modified photocoagulation laser (Carl Zeiss Meditec AG,Visulas 532 s, $\lambda =532\mathrm{nm}$) was used for treatment. The treatment radiation is superimposed by ns-probe laser pulses applied with a repetition rate of 1 kHz and adjustable pulse energies (4 to 12 μJ). The radiation is transmitted via a 50-μm core diameter fiber (Carl Zeiss Meditec AG, $\mathrm{NA}=0.11$) through a laser slit lamp (Carl Zeiss Meditec AG, SL 130). An equal contact lens and comparable data processing was performed as described in Sec. 3.2.

Prior to each treatment exposure, 20 calibration laser pulses were applied in order to determine the product $S·E_p$ according to Eq. (9). Figures 6 and 7 show exemplary fundus photos of two patients, treated with a 300 or 100 μm laser spot diameter. For selected spots, the peak temperature profiles are plotted in Figs. 6 and 7, and OCT images are shown exemplary. The peak temperature was calculated according to Eq. (12): $T_{\text{peak}}(t)=f(t)·T_{\mathrm{OA}}(t)$ using $f(t)$, $T_0$, and $T_{\max }$ as described above.

Fig. 6

Fundus picture, selected peak temperature curves and OCT images (1 h after irradiation) of patient No. 4. Retinal laser spot diameter: 300 μm, cw laser power: 200 mW, treatment irradiation time: 200 ms, probe pulse energy: 12 μJ.

Fig. 7

Fundus picture, selected peak temperature curves and OCT images (1 h after irradiation) of patient No. 1. Retinal laser spot diameter: 100 μm, laser power: 70 mW, treatment irradiation time: 100 ms, probe pulse energy: 4 μJ.

Figure 6 shows the temperature profiles for 300 μm spot diameter using a treatment power of 200 mW and a probe pulse energy of 12 μJ. The plots show 20 calibration pulses (first 20 ms) which were normalized to body temperature as described above. No temperature rise is observed here, since the calculated radiant exposure for a probe pulse is only $H=17\mathrm{mJ}/{\mathrm{cm}}^2$, which corresponds to a very short and only transient temperature rise of 4.9°C according to Eq. (2). Because only about 30% of the light reaches the target site (see below), the transient temperature peak is most likely only about 1.5°C. Overall, the probe pulses account for an average power of 12 mW, thus 6% of the treatment power, which is their contribution to the retinal heating. After 20 ms, the treatment laser starts and the temperature rises quickly until the treatment beam is ceased after 200 ms. It is evident from the plots that the temperatures increase almost as predicted by heat diffusion calculations depicted in Fig. 4. The maximal peak temperatures after 200 ms are found around 70°C to 85°C at the end of the irradiation period for weak coagulations, while in stronger ones they can exceed 100°C. Spot no. 10 shows a second temperature rise after around 100 ms. This is most likely caused by a slight eye movement (see fundus picture with a slightly elongated coagulation spot). Figure 6 also shows exemplary OCT pictures of spot Nos. 11 and 15. Both pictures indicate that inner and outer retinal layers are affected.

Figure 7 shows exemplary data of a patient treated with a 100 μm spot diameter and an irradiation time of 100 ms. The signal-to-noise ratio is worse than the 300 μm spot applications, since the probe pulse energy was reduced to 4 μJ. Generally, comparable temperatures and temperature profiles for 100 and 300 μm spot are achieved; however, with around 3 times less laser power for the 100 μm spot. This is in good agreement with heat diffusion theory, which predicts almost the same temperature rise when increasing the laser power proportionally with the spot diameter. OCT pictures for two spots demonstrate that the photoreceptor layers are coagulated. The lateral extent of the damage seems to be larger for spot No. 9, which corresponds to a higher peak temperature of around 100°C at the end of the heating period. This spot was identified with a notable eye movement. Further, some more spots were found with a temperature curve not strictly following the expected slope according to the heat diffusion theory, e.g., the linear increase of spot Nos. 9 and 12. This can most likely be attributed to slight eye movements or saccades.

5.

Discussion

A clinical study was conducted to investigate optoacoustic temperature measurements during laser photocoagulation at the retina. The generated pressure transients were processed to determine the average temperature rise during heating and coagulation, and were further evaluated to reveal the peak temperature at the RPE over time.

6.

Conclusion

It has been shown that optoacoustics can serve to measure increasing pressure amplitudes during retinal photocoagulation, which can be used to calculate the temperature during the irradiation process; however, the method has some limitations as discussed above. In this ongoing project it has been shown ex vivo on enucleated porcine eyes,⁵⁰ and in vivo on rabbit eyes^46,51 that these data can be processed in real-time in order to automatically control the extent of laser lesions. Over a wide range of laser powers, equally sized lesion diameters could be obtained in porcine eyes ex vivo, as well as in rabbit eyes in vivo by a real-time automatic laser switch-off algorithm when the appropriate temperature is reached for a certain time.^46,51,52 The first clinical data with automatic feedback controlled photocoagulations are expected in 2012.

In conclusion, all the data look very promising to successfully realize such an automatic dosage control for generating a pre-selected coagulation strength. A temperature-feedback control would also allow to apply sub-threshold laser lesions with a well-defined temperature rise in order to stimulate the neural retina without any irreversible thermal damage.

Scientifically, this method can be used to reiterate the Arrhenius parameters or thermal denaturation of retina.⁴⁶ The results of numerous clinical trials on the efficacy of photocoagulation can significantly be improved when similar coagulation strengths and depths are induced in all treatment sites. Moreover, short term coagulations with 20-ms pulse duration, which have become common since the introduction of patterned laser coagulation, strongly reduce the pain for patients.¹⁰ An automatic dosage control system would merge the advantages of reduced pain due to shallow coagulations with a high safety level like in standard 200-ms exposures. Automatic dosage control can also be applied for 20-ms exposures by a fast temperature-related power control. This is particularly useful in pattern coagulation devices like PASCAL (Optimedica/TopCon) or VISULAS VITE (Carl Zeiss Meditec).