1.

Introduction

Optical coherence tomography (OCT) is a noninvasive, noncontact imaging modality that gained great acceptance in ophthalmology as well as in other medical fields due to its high-resolution imaging capability.^{1, 2} Generating cross-sectional images of tissue morphology at high speed with micrometer scale resolution makes OCT a promising tool for measuring time-dependent physiological and pathophysiological processes in a variety of medical experiments. This technique is especially useful for pharmacological and toxicological studies, which require ongoing monitoring of morphological changes. Here, a promising application of time-resolved OCT imaging is the poorly understood initial phase of chemical eye irritation and interaction of the transparent cornea with corrosives.

OCT imaging of the anterior chamber of the eye was first demonstrated in 1994 by Izatt, ³ admitting noninvasive assessment of the corneal thickness and visualization of the corneal epithelium. A slit-lamp-adapted OCT technique was used to study the cornea after laser thermokeratoplasty and phototherapeutic keratectomy.⁴ Abnormal corneal leasons were correlated to slit-lamp biomicroscopy⁵ and to histopathology.⁶ Ultrahigh-resolution OCT was further demonstrated to be capable of imaging corneal structures like Bowman’s and Descement’s membrane as well as the typical stromal structure in an animal model^{7, 8} and in human cornea postmortem.⁹

The ability of OCT to gather nondestructively a time series of a single sample with time constants ranging from a split second to long-term observation over several days was utilized for addressing different questions under investigation. For example, wound healing after laser thermal injuring was examined in skin equivalents using OCT and multiphoton microscopy.¹⁰ OCT was used to monitor tissue freezing during cryosurgery,¹¹ as an in situ imaging technique for tissue engineering¹² and to study the heartbeat of small animals.^{13, 14} In ophthalmology, dynamic OCT measurements were demonstrated to study the corneal response to dehydration stress in vivo in a rabbit animal model¹⁵ and to quantify light backscattering within the cornea during corneal swelling.¹⁶

In chemical eye burns, the validation of the effect of therapeutical intervention by different rinsing solutions in emergency treatment by means of OCT was the objective of the present study. This technique might be helpful to define the action of parameters like temperature, amount, impact force, pH, concentration, redox-potential, and specific reactivity with the ocular tissue that are important for the development of specific treatment procedures.¹⁷ In this article, we evaluate the potential of time-resolved high-resolution OCT imaging to monitor the dynamics of chemical eye burns and the effectiveness of acute intervention procedures in the Ex Vivo Eye Irritation Test (EVEIT). The regions of damaged corneal tissue are characterized with high contrast due to a significant increase in light scattering induced by structural changes within the cornea. Our target is to measure speed and characteristics of penetration and propagation as important parameters defining the damage by chemical injury. The direct access to the damaging diffusion process allows for the development of new therapy strategies with reduced number of tests.

2.

Materials and Methods

2.3.

Measurement

OCT images were composed of up to 1000 A-scans with a lateral step size of $6\mathrm{\mu }\mathrm{m}$ for static tissue examination. Images with a width of $600\mathrm{\mu }\mathrm{m}$ (100 A-scans per image) were taken at a rate of 30 frames per min for time-resolved measurements.

OCT measurements were performed at the center of the cornea. Care was taken to ensure orthogonality of the sample beam to the surface of the probe for the measurements of central corneal thickness and central epithelial thickness. Minor tipping of the probe was used to eliminate the central corneal reflex in the tomograms presented in this study. For comparison, the cornea of each eye was imaged directly before application of the chemicals. Chemical injury was induced by superficial exposure of the cornea to the chemical under investigation. The application of a standardized amount was ensured by spraying $500\mathrm{\mu }\mathrm{l}$ of the chemical by an Eppendorf pipette onto the center of the cornea, which was positioned horizontally. By this means, a continuous film of the fluid is spread over the cornea, thinning out with time. This development is directly monitored in OCT imaging. Penetration of the chemical within the cornea was imaged starting directly after its application. A refractive index of 1.385 was used for the conversion of optical to geometrical path lengths measured by OCT.²¹ It was further assumed that the index of refraction does not significantly change during chemical burn.

Evaluation of the effectiveness of rinsing as first aid treatment was accomplished by comparing OCT images of rinsed and untreated burned eyes at the same time-steps after application of the chemical. Both eyes used in one experiment were obtained from the same animal for direct comparability. One eye was not treated after application of the corrosive. The other eye was rinsed for $15\min$ starting $20\mathrm{s}$ after application of the chemical onto the cornea. An infusion system with a flow rate of $67\mathrm{ml}∕\min$ was used for this purpose. Tomograms of both eyes obtained $16\min$ after exposure to the corrosive were compared in order to evaluate the effect of the rinsing solution.

The chemicals applied were 0.5, 1, and 2 molar $\mathrm{NaOH}$ as well as 1 molar ${\mathrm{H}}_2{\mathrm{SO}}_4$ . For rinsing, we used $1000\mathrm{ml}$ of Previn solution (Prevor, Inc.), which is recommended as an amphoteric therapeutical solution in any type of severe eye burns. The activity of this solution was described earlier.²²

The color scale of the OCT images represents the intensity only and contains no spectral information (color online only). The same logarithmic color scale as given in the inset of Fig. 1b is used for all OCT images, and the same intensity scale was used for all A-scans.

Fig. 1

High-resolution OCT image of an untreated rabbit cornea ex vivo. (a) Overview over the scanned region, $4.5\mathrm{mm}$ in width. (b) Magnification of the central section $(600\mu \mathrm{m}\times 600\mu \mathrm{m})$ of the cornea. The logarithmic color scale used for all OCT images within this study is given in the inset of this figure (color online only). (c) Intensity profile corresponding to (b) given by the average of ten adjacent A-scans.

3.

Results

A characteristic high-resolution image of an untreated rabbit cornea ex vivo is shown in Fig. 1a. The magnification of the central section of the cornea is given in Fig. 1b. Here, the epithelial and endothelial layers, the stroma, and allusively, the stromal fibers are distinguishable. The Descement’s membrane separating the stroma from the endothelium is imaged as low scattering band, which is clearly silhouetted against the higher scattering surrounding layers. Ten adjacent A-scans around the center of the image were averaged to determine layer thicknesses [Fig. 1c]. Here, the boundaries of the different layers are illustrated by peaks in signal intensity with high contrast, whereas the Descement’s membrane is represented by a noticeable drop of the OCT signal.

For direct comparison, OCT images of corneal tissue damage caused by chemical interaction are presented in Fig. 2 . Tomograms taken $10\min$ after topical application of a $1\mathrm{molar}$ solutions of sulfuric acid $\left({\mathrm{H}}_2{\mathrm{SO}}_4\right)$ and a $0.5\mathrm{molar}$ solution of sodium hydroxide $\left(\mathrm{NaOH}\right)$ are shown in Figs. 2a and 2c, respectively. Corresponding diagrams of 10 averaged A-scans are depicted in Figs. 2b and 2d. Tissue damage is delineated with high contrast indicated by a significant increase in OCT signal intensity compared to the image taken prior to the application of the chemical (Fig. 1). The microstructure of the stromal fibers, which is oriented strictly parallel to the corneal surface in unburned eyes, changes to a structure without specific orientation after interaction with the alkaline solution.

Fig. 2

Tomographic images $(600\mu \mathrm{m}\times 600\mu \mathrm{m})$ and corresponding A-scans of corneal tissue damage caused by topical application of corrosives. (a) Section $10\min$ after topical application of a 1 molar solution of $500\mu \mathrm{l}$ ${\mathrm{H}}_2{\mathrm{SO}}_4$ . (b) Intensity profile corresponding to (a) given by the average of ten adjacent A-scans. (c) Section $10\min$ after topical application of a 0.5 molar solution of $500\mu \mathrm{l}$ $\mathrm{NaOH}$ . (d) Intensity profile corresponding to (c) given by the average of ten adjacent A-scans.

For the case of ${\mathrm{H}}_2{\mathrm{SO}}_4$ application, an increase in signal amplitude is restricted to the epithelial layer, while structure and thickness of the stroma appear unchanged. However, penetration of sulfuric acid within the upper part of the stroma was observed for concentrations of $2\mathrm{mol}∕\mathrm{l}$ . With the exception of hydrofluoric acid, the tendency to remain on the ocular surface was also observed for other acids of moderate concentration, e.g., hydrochloric acid (not shown here).

In contrast, after $\mathrm{NaOH}$ application, the inner structure of the epidermal layer appears almost unchanged in OCT imaging, while its thickness increases by a factor of 1.5. Significant increase in light scattering within the stroma was found even for low concentrations of $\mathrm{NaOH}$ . Other alkalis like calcium hydroxide were found to react similarly (not shown here). The easy and fast penetration of alkalis throughout the cornea and the anterior segment is the central issue of the severe clinical manifestation of the alkali-injured eye.

To gain insight into the dynamics of the extensive tissue damage observed for alkali-induced eye burns, OCT image sequences applied after superficial exposure of the cornea to a $1\mathrm{molar}$ and a $2\mathrm{molar}$ solution of NaOH are compared in Fig. 3 . The corresponding cross-sectional image of the cornea before $\mathrm{NaOH}$ application is given in the leftmost image. Time-steps given within the following pictures refer to zero time delay $(t=0)$ , when the $\mathrm{NaOH}$ is applied onto the corneal surface. The liquid film of the applied alkaline solution is imaged in the subsequent tomograms. It is distinguishable from the epithelium by the absence of internal scattering. A decrease of the film thickness of the $\mathrm{NaOH}$ solution in combination with swelling of the epithelium is observed over time. Penetration of the corrosive within the stroma is indicated by an increase in signal amplitude for affected areas. Full penetration of the cornea is completed within $120\mathrm{s}$ for $1\mathrm{molar}$ $\mathrm{NaOH}$ and within $38\mathrm{s}$ for $2\mathrm{molar}$ $\mathrm{NaOH}$ . The temporal evolution of the penetration depth was obtained from averaged A-scans of the OCT image sequence given in Fig. 3. Here it was assumed that the front of penetration is given by the significant drop of scattering intensity within the stroma. The penetration velocity was found to increase with concentration and to exponentially decrease with time of penetration (Fig. 4 ). For $\mathrm{NaOH}$ concentrations of $0.5\mathrm{mol}∕\mathrm{l}$ , the penetration velocity falls down to zero within the cornea and therefore full penetration of the cornea was not observed. No discontinuity of the penetration could be found for the transition from epithelium to stroma.

Fig. 3

OCT image sequences illustrating corneal tissue damage caused by topical application of $500\mu \mathrm{l}$ of differently concentrated $\mathrm{NaOH}$ . (a) 1 molar solution; (b) 2 molar solution.

Fig. 4

Numeric analysis of $\mathrm{NaOH}$ penetration time and velocity within rabbit cornea derived from OCT time series. (a) Plot of penetration depth of $\mathrm{NaOH}$ (1 and 2 molar solutions) versus time after topical application. (b) Plot of the penetration velocity of $\mathrm{NaOH}$ solutions versus time after topical application given by the derivative of plot (a).

The capability of OCT to monitor the effectiveness of rinsing as first aid treatment for chemical eye burns is demonstrated in Fig. 5 . Figure 5a shows an unrinsed burned eye, while the rabbit eye imaged in Fig. 5b was rinsed for $15\min$ starting $20\mathrm{s}$ after application of the corrosive using $1000\mathrm{ml}$ of Previn solution (Prevor, Inc.). Both eyes were imaged $16\min$ after chemical eye burn with $1\mathrm{M}$ $\mathrm{NaOH}$ . There is clear evidence in Fig. 5a that without rinsing, a large scattering amplitude is observed over the complete cross-sectional area of the stroma, which indicates full corneal penetration of the corrosive. In contrast, the lower half of the stroma appears almost unchanged in the OCT image for the rinsed eye. Additionally, after alkali eye burn and subsequent rinsing, the epithelial layer has disappeared.

Fig. 5

Demonstration of the effectiveness of rinsing as a first aid treatment after chemical eye burn. Two rabbit corneas were imaged $16\min$ after topical application of a 1 molar solution of $500\mu \mathrm{l}$ $\mathrm{NaOH}$ . (a) No intervention procedure has been carried out between application of the chemical and OCT imaging. (b) The eye was rinsed with $1000\mathrm{ml}$ Previn for $15\min$ starting $20\mathrm{s}$ after application of the $\mathrm{NaOH}$ solution.

4.

Discussion

In this study, we demonstrate the benefits of high-resolution OCT imaging for providing objective information about the pathophysiological damage caused by chemical eye burns in an ex vivo animal model. A primary source of contrast in OCT is the local scattering cross section. Injury of the eye initiated by a corrosive changes inherently the scattering properties of the tissue. This process can be effectively monitored with high spatial and temporal resolution using OCT imaging. Using high-resolution OCT, additional information about the resistance of potential barriers like the Descement’s membrane as well as microstructural changes as observed for the orientation of the stromal fibers can be extracted.

The inner structure of a healthy rabbit cornea ex vivo is illustrated by high-resolution OCT (Fig. 1). Here, the cellular and lamellar structure is accessible. The cross-sectional information is comparable to slit-lamp microscopy, with improved detection of inner corneal layers including Descemet’s membrane and endothelium. The central epithelial thickness was measured to be $47\mu \mathrm{m}$ on the average, which is in close agreement to the value of $45.8\mu \mathrm{m}$ given by Reiser ⁸ for the same ex vivo animal model. The averaged central corneal thickness (CCT) derived from recorded OCT images was $445\mu \mathrm{m}$ with a standard deviation of $21\mu \mathrm{m}$ . This value is significantly larger than the CCT of approximately $360\mu \mathrm{m}$ measured by pachymetry for in vivo rabbit eyes.²³ Slight overhydration causing an increase of corneal thickness is known to be due to the $4°\mathrm{C}$ storage and consecutive low pumping activity of the endothelium.

The microstructure of the tissue under investigation determines its scattering cross section and therefore the OCT signal amplitude. In reverse, the change of tissue appearance in OCT images caused by interaction with a corrosive can be ascribed to structural changes caused by the chemical. Healthy corneal stroma transmits 99% of the incident light without scattering.²⁴ It is generally accepted that the fibril diameter, the interfibrillar distance, and the lattice-like organization of the collagen fibrils play a crucial role in stromal transparency.²⁵ Any structural changes within the stroma will result in corneal opacification. For chemical eye burns, structural disorders are initiated by direct chemical interaction as well as by water-electrolyte imbalance followed by a change of the hydration state within the stroma.²⁶ These structural changes can be observed by OCT for alkalis [Figs. 2b and 3] as well as for acids that penetrate within the stroma.

In corneal damage caused by acids, the anion involved, i.e., ${\mathrm{SO}}_4^{2-}$ for the case of sulfuric acid, gives rise to protein denaturation and subsequent opacification of the epithelium,²⁶ as is observed in Fig. 2a. The acid-induced coagulation of proteins within the epithelium acts as a barrier for further penetration.²⁶ Therefore, the appearance of the stroma typically remains constant after application of moderately concentrated acids onto the eye [Fig. 2a].

Alkalis penetrate very rapidly into the eye. In the epithelium, the structural dissolution of cell membranes by saponification in alkali can be detected by the low reflection within this layer [see Fig. 2b]. The total loss of integrity of the epithelium is confirmed by OCT imaging after rinsing [Fig. 5b]. Here, the epithelium has vanished as the cellular fragments are washed away by the solvent. Histological examination after corneal $\mathrm{NaOH}$ burn also demonstrate the loss of this cellular layer.²⁷

Alkali-induced structural breakdown of the corneal matrix is a well-known fact shown by the experiments of Pfister, who exposed healthy tissues with proteins collected from freshly burnt corneas.²⁸ Structural damage within the stroma is induced by the cation of the alkali, which reacts with the carboxyl groups of stromal collagen and glycosaminoglycans.²⁶ Such direct chemical interaction does not take place if the cornea is exposed to a neutral hyperosmolar rinsing solution like Previn solution (Prevor, Inc.). OCT imaging during treatment of unburned eyes with this solution shows shrinkage of the corneal thickness as well as an associated minor increase in OCT signal amplitude. The same effect is observed for the lower part of the stroma after rinsing a burned eye [Fig. 5b]. This observation can be contributed to the dehydration of the cornea through the semipermeable corneal epithelium.²⁹ Here, the observed increase in scattering induced by the change in the hydration state of the stroma is marginal compared to that observed after application of a corrosive. This indicates that structural damage by direct chemical interaction with the corrosive induces the observed distinct increase in OCT signal during chemical eye burn. Further evidence to this hypothesis is given by comparing direct measurements of alkali penetration time to the time-resolved OCT data. Topical application of $2\mathrm{molar}$ $\mathrm{NaOH}$ results in a rise of the intracameral pH value after a delay of approximately $50\mathrm{s}$ .²² This measurement of the penetration time throughout the cornea corresponds well to the value of $38\mathrm{s}$ derived from OCT imaging as given in Fig. 4a. As the pH electrode in intracameral pH measurement is not placed directly at the endothelium of the cornea, the observed discrepancy in full corneal penetration time compared to the time derived from OCT measurements can be ascribed to the diffusion time of the corrosive within the aqueous humor. For this reason, the direct measurement of the diffusion time is comparable to the time-resolved OCT experiments introduced in this study. This verifies that the increase in scattering cross section within the stroma is an instantaneous process on the time scale under consideration and therefore is a useful measure to determine the depth of penetration of the corrosive. Using time-resolved OCT imaging to examine chemical eye burns, additional parameters not accessible by established methods, e.g., intracameral pH measurement, can be derived within the present study. Spatial resolution provided by OCT gives access to investigate the depth-dependent penetration velocity [see Fig. 4b]. For $\mathrm{NaOH}$ application, the penetration velocity was observed to decrease with penetration depth. For $2\mathrm{molar}$ $\mathrm{NaOH}$ the diffusion velocity decreases from about $40\mu \mathrm{m}∕\mathrm{s}$ to $5\mu \mathrm{m}∕\mathrm{s}$ during corneal penetration. For a $\mathrm{NaOH}$ concentration of $1\mathrm{mol}∕\mathrm{l}$ , the diffusion velocity is reduced below $1\mu \mathrm{m}∕\mathrm{s}$ at the bottom part of the stroma. Besides the consumption of the reactants, which undergo chemical reactions, dilution of the corrosive within the tissue might play a role for this deceleration. This is consistent with the strong correlation of penetration velocity on the concentration of the applied $\mathrm{NaOH}$ solution [Fig. 4b].

As demonstrated in Figs. 2 and 3, OCT imaging allows for observation of chemical eye burns before full corneal penetration is reached. From the time measured for half cornea penetration, conclusions can be drawn about the time frame that is available for effective intervention procedures like rinsing. Immediate rinsing of the eye is the most important factor in emergency treatment of chemical eye burns. The recommended therapeutic solutions differ with type of chemical reaction and buffer capacity. Here, only few data on the comparative application of rinsing solutions exist.²² OCT is able to supply objective information about the effectiveness of rinsing therapy, e.g., prevention of structural changes and precipitation within the corneal stroma induced by the rinsing solution. The latter negative effect was observed using phosphate buffer as rinsing solution.^{30, 31} Comparison of the appearance of chemical eye burns induced by $1\mathrm{molar}$ $\mathrm{NaOH}$ without rinsing and after rinsing using a commercial antidote for chemical burns demonstrates the potential of this imaging technique for the quantification of first aid treatment conditions (Fig. 5). The status after rinsing can be compared directly to the status at that time-step where rinsing was started [given by the time series data set in Fig. 3a]. When rinsing was started $20\mathrm{s}$ after $\mathrm{NaOH}$ application, the stroma was halfway penetrated by the alkali. In the example introduced here, this status was preserved under the described rinsing conditions by effectively inhibiting further penetration. Furthermore, no structural changes of the intact part of the tissue due to exposure to the rinsing solution was observable using OCT.

In summary, we showed that high-resolution OCT is capable of imaging the dynamics of chemical eye burns in real time. It was demonstrated that OCT is sensitive to the change in the optical properties of the cornea, which can be correlated to changes in the microstructure as a result of chemical damage. Here, the advantage of OCT compared to standard methods like histopathology and measurement of the intracameral pH is the extent of information that can be achieved by time-resolved OCT imaging of corneal penetration. OCT as an instrument for high-contrast imaging of chemical eye burns might lead to further insight into the mechanism of this kind of injuries. Due to the possibility for noninvasive serial measurements, OCT should be employed as a valuable tool in the diagnosis of eye irritation as well as in detection of biological and structural chemical changes, giving objective measures for quantifying the effect of rinsing solutions. This system accomplishes the EVEIT research in chemical product testing, replacing research on living animals, e.g., within the REACH-System (Registration, Evaluation, and Authorization of Chemicals), a new regulatory framework in the European Union.