Retinal safety evaluation of two-photon laser scanning in rats.

Safe use of retinal imaging with two-photon excitation in human eyes is crucial, as the effects of ultrashort pulsed lasers on the retina are relatively unknown. At the time of the study, the laser safety standards were inadequate due to the lack of biological data. This article addresses the feasibility of two-photon retinal imaging with respect to laser safety. In this study, rat retinas were evaluated at various laser exposure levels and with different laser parameters to determine the effects of laser-induced optical damage. The results were experimentally verified using confocal reflectance imaging, two-photon fluorescein angiography, immunohistochemistry, and correlated to the IEC 60825-1 laser safety standard.

evaluated the retinal safety of two-photon laser scanning ophthalmoscopes [19,20]. Therefore, experimental studies on laser safety with respect to exposure levels and stationary vs scanning laser beams would help in a better understanding of the biological effects of the pulsed lasers and support refining the MPE for the safety standards.
We evaluated the effects of a pulsed laser (780 nm, 270 fs) on rat retinas for different power levels and exposure times. Although there are structural differences between rats and humans, this pilot study was carried out to determine the laser-tissue interactions in rodents since understanding the structural characteristics of the eye in different species is a key point. Also, rodents are widely used in ophthalmic research as retinal disease models which allow for studying disease mechanism and corresponding treatments.

Animal preparation
Twelve brown Norway rats and twelve albino rats (Crl: CD (SD) IGS) with jugular vein catheterization were used for the experiment. Four brown Norway rats (pigmented) and four albino rats (non-pigmented) were exposed to the scanning laser beam, and eight brown Norway rats and eight albino rats to the stationary laser beam. Experiments were performed on one eye at a time and the fellow eye served as a control. The experimental rats were anesthetized using isoflurane 1-3% inhalant, and in addition, a drop of topical proparacaine anesthesia (proparacaine hydrochloride ophthalmic solution USP, 0.5% sterile) and a pupil dilator (0.5% tropicamide ophthalmic solution, USP) were applied to the eyes. Furthermore, sterile saline drops were applied to the experimental eyes every 30 seconds to keep the cornea moisturized. The Institutional Animal Care and Use Committee (IACUC) at Loma Linda University approved the animal experiments, and the experiments were carried out in accordance with the Association for Research in Vision and Ophthalmology (ARVO) statement for the use of animals in ophthalmic and vision research.

Two-photon laser scanning ophthalmoscope
The two-photon prototype has been described in detail elsewhere [21]. In brief, a compact femtosecond laser with 50 MHz repetition rate, 270 fs pulse width, 780 nm central wavelength, with tunable output power up to 500 mW was used as a light source. The confocal reflectance imaging and two-photon imaging can be performed using the same instrument in the high-resolution mode and high-speed mode with 30° x 30°, 20° x 20° and 15° x 15° transversal field of view. The scanning frequency per frame in the high-resolution mode is 5 Hz/7 Hz/9 Hz at 30°/20°/15° scan angle, whereas, in high-speed mode, the scan frequency is 9 Hz/12.5 Hz/16 Hz, respectively. The digital image readout for the highresolution mode is 1536 × 1536, 1024 × 1024, 768 × 768; and for the high-speed mode 768 × 768, 512 × 512, and 384 × 384. In real time, the signal-to-noise ratio of the images was increased by a frame averaging. For this study, the standard objective lens (focal length, f = 30 mm) with an additional objective lens (f = 40/+25 diopters) from Heidelberg Engineering GmbH was used to collect the two-photon signals. The additional objective lens reduces the beam diameter by 70%, adapting the system to the shorter axial length of the rat eyes. The images were acquired in high-resolution mode with 30° scan angle for fundus overview and 20° and 15° scan angles for the detailed view of the retina.

Study design
Group 1: The retina was exposed to a scanning raster beam for two different time scales 100 seconds and 300 seconds at 160 mW for 30° and 15° scan angles. In total, four Brown Norway rats and four albino rats were exposed to the scanning laser beam as shown in Table  1.

Group 2:
The retina was exposed to a stationary beam (spot size 5 µm) for different time scales, i.e., from 60 seconds to 600 seconds at 7 mW, 13 mW, 80 mW, and 160 mW laser power ( Table 2). The threshold measurements for minimal visible lesions were determined using the stationary laser beam.
Prior to the experiment, a confocal reflectance image was acquired at a lower power level to determine the retinal location. The laser power levels were measured at the cornea since the power measurement at the retina has limitations. On rare incidents, the targeted eye drifts under the influence of anesthesia. At such instances, the exposure to the laser was blocked until the eye returned to its original position. 100, 300 2 + ˟ One eye exposed to single exposure time and another eye exposed to multiple exposure times. For example, two rat eyes were used for 7 mW laser beam exposure, of which one eye was exposed to single exposure time (300 seconds) and another eye was exposed with multiple exposure times (60, 100, 300 and 600 seconds). + Retina was exposed to only one single exposure time. For example, two rat eyes were used for stationary laser beam exposure at 80 mW laser power. One rat was exposed to 80 mW for an exposure time of 100 seconds, and another rat to 80 mW for 300 seconds.

Two-photon fluorescein angiography
Fluorescein (10% AK-FLUOR, Akron) was administered through the jugular vein catheter for two-photon fluorescein angiography. The administered dose was in correlation with the animal's body weight. The two-photon fluorescein angiography in animal models has been described in detail elsewhere [21]. In brief, the fluorescein was injected to visualize the retinal vasculature, and in the two-photon mode, the flow of fluorescein through the blood vessels was captured. The two-photon fluorescein angiography was performed to assess the potential retinal damage, and to evaluate the efficacy of utilizing the two-photon fluorescein angiography in determining the light-induced damage.

Histology
The experimental rats were followed up for four days with confocal reflectance and twophoton fluorescein angiography and then euthanized humanely for histological analysis. The control and the experimental eye globes were enucleated and fixed in Davidson's solution for 24 hours, and then embedded in paraffin. The retinal cross-sections of size 5 µm were cut from the paraffin-embedded eyes and transferred to the microscopic slides. The retinal sections were analyzed immediately under the light microscope, and after careful evaluation, some retinal sections were selected for hematoxylin and eosin (H&E) staining. In parallel to the H&E stained retinal sections, further sections from the same region were selected for the DeadEnd Fluorometric TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) assay. The TUNEL assay was performed according to the manufacturer's instructions (Promega Corporation, USA), and the retinal sections were mounted with VECTASHIELD + DAPI to allow for staining the nuclei. The TUNEL assay detects and quantifies the apoptotic cell death by measuring the nuclear DNA fragmentation, a vital biochemical hallmark of apoptosis in many cell types [22]. The TUNEL assay is an established method to detect and quantify the apoptotic cells due to light damage [23][24][25].

Microscopy
The light microscopic (Aperio scan scope, Leica Biosystems) images of the H&E stained sections were captured to analyze the structural changes in the retina. The fluorescence microscopic (Olympus Corporation) images of the TUNEL stained retinal sections were captured to detect the apoptotic cells due to the laser exposure laser. Under a fluorescence microscope, the retinal sections were analyzed with 10 × , 20 × and 40 × objective lenses using a standard fluorescein filter at 520 nm to view the green fluorescence of fluorescein and at 490 nm for blue DAPI. Figure 1 shows the confocal reflectance and two-photon fluorescein angiography images of brown Norway and albino rat exposed to the scanning laser beam exposure. The exposure settings were: scan angle 30°, exposure duration 300 seconds, laser power 160 mW. The confocal reflectance image [ Fig. 1(a) and 1(d)] right after the laser exposure did not show any retinal abnormalities. The real-time two-photon fluorescein angiography from early phase [ Fig. 1(b) and 1(e)] to late phase [ Fig. 1(c) and 1(f)] showed neither hypofluorescence nor hyperfluorescence. Retinal capillaries were visible at the early phase of the fluorescein influx, and the fluorescein diminished slowly. The late phase demonstrated a graded elimination of dye from the retinal vasculature. The same results were observed for a scanning laser beam exposure at a 15° scan angle, 160 mW laser power exposures for 100 seconds and 300 seconds. Furthermore, the immunohistochemistry (TUNEL assay) analysis from the exposed retinal sections showed no cellular damages. The retinal sections of the control and the experimental eye were analyzed using the fluorescence microscope, and neither significant changes in cell loss nor disruption in the retina or retinal swelling were noticed. Figure 2 shows the fluorescence images of the control and the experimental retina of brown Norway and albino rats. The green fluorescence (arrow) noticed in Fig. 2(d) is an artifact of histology sections. And this was verified by one of the authors and a researcher from the department of pathology and human anatomy, Loma Linda University.  Figure 3 show the brown No to 13 mW for seconds (arrow pulsed laser (7 laser beam av retina was cap exposure to th . Scanning laser b r in brown Norwa e laser-exposed ret -7 to 10 minutes 2. Fluorescence mi ons exposed to the he experimental ey he experimental ey ed images of DAP l cells. mages of s exposed ) and 600 ultrashort he parked age of the nduced by (b). With a longer exposure time, more radiation has been absorbed and hence there was an increase in the size of the lesion. After the exposure, two-photon fluorescein angiography was performed, and the early phase to late phase angiograms was recorded. High-contrast retinal microvasculature's were clearly seen in the early phase two-photon fluorescein angiography [ Fig. 3(c), yellow box], and at the late phase [ Fig. 3(d), arrow] the hyperfluorescence was noticed on the laser-exposed retinal area. The irregularity in the nerve fiber layer and the absence of capillaries in the exposed area were noticed on day 4 [ Fig. 3(e), 3(f)]. In Fig. 3(e), the confocal reflectance image of the retina shows high reflectivity and irregular nerve fiber layer at the laser-exposed area (yellow box), whereas the retina on the other side of the optic nerve (unexposed retina) did not show any structural changes or irregularities. Also, the early two-photon fluorescein angiography on day 4 [ Fig. 3(f), yellow box] showed the absence of capillaries in the laser-exposed retina. confocal reflectance image before exposure, (b) confocal reflectance image after exposure to the laser at 13 mW for 60 seconds (arrow 1), 100 seconds (arrow 2), 300 seconds (arrow 3) and 600 seconds (arrow 4). The early phase (c) and late phase (d) two-photon fluorescein angiography, the hyperfluorescence (d, arrow) seen in the laser-exposed area. (e) Confocal reflectance image of the laser-exposed retina on day 4. Early phase (f) two-photon fluorescein angiography of the exposed retina on day 4

Brown
In addition, the H&E staining and TUNEL assays were performed on the retinal sections to evaluate the potential thermal damage to the retina. The retinal cross-sections were evaluated at different depths. The retinal sections close to the periphery of the eye globe did not show any structural damages on the retina [See Fig. 4(a), 4(d), 4(g)], whereas, the retinal sections closer to the exposed area showed disruption of the outer nuclear layer and the pigment epithelium [See Fig. 4

Albino
Conversely, i exposure at 7 seconds, no d exposure for 6(b), 6(e)] an albino rat w exposure. Th compared to t on day 4, neit in the brown N The TUNE any apoptotic was noticed. F on the contro could be due t even in contr including env housing, and m animals ensur rats n albino rats, 7 mW, 13 mW damages on th 300 seconds, nd the two-pho ere acquired he hyperfluore the brown Nor ther change in Norway rat [Se  Fig. 7(a) and 7 n damage the R nduced damage s, and aging. the health and tigations [26]. e stationary la ower exposure 160 mW las l reflectance [F 6(c), 6(d), 6( mages due to l in the expo follow-up exa pillaries were n r power. Confocal image after laser angiography. The flectance and early albino rat's retinal power. and 7(b)] did disruption in t uclear layer (O 7(b), yellow b RPE and photo es can be mult Animal envi wellbeing of la aser beam e for 100 er power Fig. 6(a)  med on the ra revious study 0 mW laser po at 13 mW for n to determine L assay were p ture recovery. wn in Fig. 8. T nd the follow-u ) and 8(b), a Also, the refle ina exposed to xposure at 13 m image analys 8(c)]. wo-photon ed to the e in these rats took approximately 45-60 minutes. The fluorescence microscopic analysis of these retinal sections reported no apoptosis or structural changes in the retina even after multiple exposures. The fluorescence images of the brown Norway rat's retinas exposed to the scanning laser beam is shown in Fig. 9(a) and 9(b). Since no apoptosis was noticed in the exposed retinal sections, a positive control test was performed on retinal sections according to the manufacturer's instructions [22]. The apoptosis was clearly noticed [ Fig. 9(c)] on the positive control retinal sections. This shows that the scanning laser beam even after multiple exposures produces no considerable damages.

Pigmented vs. non-pigmented retinal toxicity to laser exposures
A scanning laser beam returned no retinal toxicity in both animal models was verified by confocal reflectance imaging, two-photon fluorescein angiography, and TUNEL assay. However, for stationary laser beam exposure, the retinal lesions were noticed at 13 mW for 60 seconds in brown Norway rats; and only after exposure to 160 mW laser power for 300 seconds in albino rats. The number of TUNEL positive cells was significantly higher in brown Norway rats compared to albino rats. Retinal swelling, a disorganized outer segment with no visible RPE and nuclear layer degeneration were noticed in brown Norway retinal sections. In contrast, the albinos did not show any retinal damages at low laser power and only minimal damages at longer exposure to high laser power. The albino rats had a higher laser safety threshold than the pigmented brown Norway rats due to lack of melanin in the RPE layer. Absorption serves fundamentally in determining the potential toxicity of light on the retina.
Also, no potential thermal damages noticed on the brown Norway rat retina exposed to the stationary laser beam of 7 mW, while, the stationary laser beam exposure at 80 mW and 160 mW induced a retinal burn, pigmentary change, retinal detachment, and hemorrhage. These results infer the morphological changes related to tissue radiation interaction. In brown Norway rats, this was primarily due to the absorption of radiant energy by melanin within the RPE and the choroidal melanocytes. But in albino rats, the laser-induced effects occurred because of multiple scattering, together with absorption within hemoglobin and possibly also within tissue water [27].

Laser safety studies in animal models
Rodents have been predominantly used in ophthalmic research to study the disease progression and effect of therapies. Also, the rodents (rats 0.43) have larger numerical apertures compared to human eyes (0.20), and this possibly could resolve smaller retinal features [28]. Two-photon retinal imaging has been studied in different animal models including rodents, however, for the laser safety study rodents may not be the appropriate animal models due to several factors. The laser safety standards determine the MPE values by the means of experimental studies in non-human primates (NHP) since NHP eyes are closer to human eyes. Also, the retinal damage depends on the transmission of the laser radiation (ocular transmission in rats is 0.9 or greater at 780 nm), absorption, and the diameter of the laser spot at the retina. Multiphoton absorption by ultrashort pulsed laser needs to be ensured for safe retinal imaging, and in rats, it might not be possible to evaluate the UVA since the rat's cornea transmits UV radiation and supports its vision [29].
Laser safety guidelines specify that the diameter of the irradiance profile (D) at the retina determines the damaging potential of energy incident on the retina [30]. The retinal irradiance diameter can be calculated as e D = f α , where α is the source angle and e f is the effective focal length of the eye (rats, e f = 3.37 mm , humans, e f =17 mm ). The retinal irradiance diameter for the small source ( .5 mrad α = 1 ) in rats (5 µm) is 3 times lower than the human eyes (25 µm). Therefore, the energy required to create the damage must be lower in rats than humans since the irradiance diameter is smaller in rats. However, in this study, the visible retinal lesions were noticed in Brown Norway rats only at 13 mW laser power which is 2.3fold higher than the safety standard (see Appendix). This could be due to the influence of the retinal-laser spot diameter, and differences in animal species. Also, rats have higher refractive power and higher-order aberrations, and these factors influences the spot diameter. Therefore, accurate analysis of the potential retinal damage in different animal species needs to be carefully evaluated and correlated for safe two-photon imaging in humans.

Future directions
From our previous study, the power threshold required for the two-photon autofluorescence imaging with our prototype was determined to be 160 mW [21] and is two orders of magnitude higher than the safety standards. Although no potential thermal damages noticed due to the scanning laser beam, further improvement in light delivery is needed for the twophoton ophthalmoscope in clinical applications. Implementing adaptive optics to the current two-photon prototype would improve light delivery by a factor of four. Also, adaptive optics has emerged as an empowering technology for retinal imaging enabling diffraction-limited and holds potential for non-invasive detection and diagnoses of eye diseases. However, the cost and complexity of adaptive optics ophthalmoscopes with a limited field of view currently impede its clinical use [31]. The high-magnification objective (HMO) lens (Heidelberg Engineering GmbH, Germany) with phase plates has the potential to establish adaptive optics in a clinical application by simplifying its incorporation in the prevailing systems [32,33]. The combination of HMO with the phase plates could resolve ocular microstructures without the use of complex adaptive optics system for an 8-degree field of view. Furthermore, a femtosecond laser with a reduced pulse width would improve the two-photon efficiency, since the shortest pulse width coupled with group velocity dispersion compensation would greatly increase the two-photon excitation induced fluorescence [1]. Thus, employing a shorter pulse width laser (~55 fs) with adaptive optics to the prototype will offer highresolution retinal imaging sustaining the laser safety standards.

Conclusion
This study concludes that the use of two-photon scanning laser ophthalmoscope for in-vivo retinal imaging is safe in rats. No potential thermal damages observed due to the scanning laser beam at both high laser power and multiple exposures. The high contrast two-photon fluorescein angiography images were effective in evaluating the light-induced retinal damage. Furthermore, the immunohistochemistry analysis supported the study by analyzing the damages at a cellular level. Since no potential thermal damages were noticed even at high laser power, the physiological and biological process of the retina in rodents can be studied in vivo using the two-photon ophthalmoscope. However, for implementation of this technique in clinics, the appropriate light safety standards must be well established. The retinal toxicity evaluation at different operating wavelengths, and for shorter pulse widths (typically < 100 fs) in NHP could establish a broader sense of these safety standards. Shorter pulse widths have a direct relationship to the generated two-photon fluorescence, and therefore it is vital to establish the correlation of the shorter pulse widths to retinal phototoxicity. For repetitively pulsed lasers the following conditions should be tested to determine the MPE.

Appendix
Condition 1: The exposure from any single pulse shall not exceed the single-pulse MPE. Thus, the radiant exposure for the time period 270 fs is: Comparing the three MPE's, the condition 2 yields the most restrictive and therefore the single pulse MPE for the two-photon prototype must be 562 µW. The MPE in laser safety guidelines is 10 times lower than the damage threshold.

A.2 Laser safety analyses for an extended source
Here the total illuminated field considered as an extended source. The repetition rate f PRF of the scan field equals the frame rate; f PRF = 5 Hz and 9 Hz for 30° and 15° scan angle. The angular subtense of the scan field is 525 × 525 mrad and 262.5 × 262.5 mrad for 30° and 15° scan angle respectively. Duration of the field illumination, f t = 192 ms and 96 ms for 30° and 15° scan angle. Since, mrad α > 100 , 5 C =1. Therefore, condition 3 will result in the same as condition 1. The thermal limits for the two-photon prototype can be interpolated from condition 2 and it is 21 mW and 16 mW for 100 and 300 s. For intentional long-term viewing, the output power must be 5 mW for the study prototype. The laser safety guidelines developed the standard for the human eye with a focal length of 17 mm and the pupil size of 7 mm. Therefore, the MPE for the rat's retinal imaging can be obtained by scaling the obtained MPE using the square of the ratio of effective focal length of the rat and human eyes [34].