High resolution anterior segment optical coherence tomography of ocular surface lesions: A review and handbook

Abstract

Introduction:

High resolution anterior segment optical coherence tomography (HR-OCT) has revolutionized the way by which clinicians can diagnose and differentiate ocular surface lesions. Careful interpretation of HR-OCT images can provide morphological information about the lesion of interest and help guide the diagnosis.

Areas covered:

This paper reviews the steps to interpreting HR-OCT images of ocular surface lesions and summarizes their characteristic findings.

Expert opinion:

Among the multiple modalities available to image the ocular surface and anterior segment, HR-OCT has emerged as an instrumental tool to obtain “optical biopsies” of various ocular surface lesions. A step-by-step approach to acquiring and interpreting HR-OCT images can allow for accurate in-office characterization and diagnosis of both benign and malignant ocular surface lesions.

Introduction:

The evolution of optical coherence tomography (OCT) has been crucial for the diagnosis of several pathologies in ophthalmology. Particularly, OCT imaging of the posterior segment has allowed identification of many retinal pathologies, such as vitreomacular traction, epiretinal membranes, retinal detachments, macular edema, or macular degeneration. Most ophthalmologists are familiar with the interpretation of these posterior segment images to help guide their clinical practice.

Anterior segment OCT has emerged as a powerful and reliable tool for the diagnosis of various anterior segment pathologies.^1–3 The advancement from time-domain to spectral-domain platforms to obtain images of high and ultra-high axial resolutions has allowed physicians to better visualize with increasing detail the tear film, corneal layers, corneal/scleral junction, anterior chamber angle, and various layer interfaces.^4–12 Anterior segment OCT has shown great utility in the field of glaucoma, allowing for acquisition of highly reproducible iridocorneal angle measurements ^13,14 as compared with ultrasound biomicroscopy (UBM). Similarly, OCT imaging has shown great promise in identifying various corneal pathologies such as dry eye disease,¹⁵ keratoconus,¹⁶ or corneal dystrophies¹⁷ and is a particularly helpful tool to assess for graft attachment intraoperatively and to detect graft detachment post-operatively after endothelial transplantation.^18–20

Notably, high resolution anterior segment (HR-OCT) has proven to be instrumental in its ability to diagnose and differentiate between various ocular surface lesions, especially benign lesions such as pingueculae and pterygia as well as ocular surface tumors, such as ocular surface squamous neoplasia (OSSN), conjunctival melanoma, conjunctival lymphoma, and conjunctival amyloidosis.^6,21–24 The purpose of this paper is to bridge the knowledge gap for the acquisition and interpretation of HR-OCT images, enabling the general ophthalmologist to use “optical biopsies” to diagnose various benign and malignant ocular surface lesions.

Body:

Anterior Segment Optical Coherence Tomography Evolution:

OCT technology was first reported as a method to image internal structures in biologic tissues in 1991.²⁵ OCT of the anterior segment was initially introduced in 1994 using 830 nm wavelengths of light,³ and sectioning capabilities were similar to that of confocal microscopy but with better depth resolution.²⁵ OCTs have now evolved over time to include wavelengths of light ranging from 400 to 2500 nm.^26,27 This technology utilizes a Michelson interferometer which creates an infrared, reference beam of light against which multiple other beams of light are measured as they return from variably reflective tissues of the eye. These reflected light beams are collected by the device to create an interference pattern over the surface of the structure imaged.⁷

Initial platforms were time-domain devices in which a reference mirror moved in a linear fashion over the eye to generate multiple A-scans (scans that measure depth and reflectivity of different tissues). These were then combined to make a cross-sectional image, called a B-scan.¹ With time-domain platforms, axial resolution was limited by the speed of A-scan acquisition, as the reference mirror had to physically move to process serial A-scans.^28,29 These devices were able to achieve excellent penetration of tissues and scan width to encompass all anterior segment structures.¹

The transition to spectral-domain or Fourier-domain devices has permitted faster scanning speeds and higher axial resolution images due to the employment of spectrometers, high-speed cameras and broadband light sources to allow for simultaneous capturing of all A-scans.^26,27 Rapidly obtained images can range from axial resolutions of less than 5 microns (considered ultra-high resolution) to 5 microns (considered high-resolution). Obtained images provide in vivo, cross-sectional views that can display the structural details of the normal ocular surface as well changes in morphology characteristic of various conjunctival and corneal pathologies.^1–3,30

While newer spectral-domain OCT platforms have offered higher axial resolutions, they are limited in their overall depth of penetration of tissues as compared with time-domain OCT platforms as well as UBM. Although depth is essential to evaluate certain anterior segment pathologies such as iris cysts or scleral involvement of conjunctival tumors, among others, higher overall depth power will limit the axial resolution of the final image that is often needed for more superficial ocular surface lesions.³¹

More recently, commercially available swept-source OCTs have emerged that can acquire up to 100,000 A-scans per minute (Table 1). This high-speed technology acquires numerous longitudinal and transverse scans to create 3-dimensional views of the anterior segment and research is ongoing to study the implementation of this technology for improved imaging of various ocular structures.¹²

Table 1:

Summary of currently commercially available anterior segment optical coherence tomography (OCT) machines

Instrument	Company	Measurement Type	Approximate Axial Resolution (microns)	Scanning Speed (A scans per second)
Slit-lamp OCT	Heidelberg	Time-domain	25	2,000
Visante OCT	Carl Zeiss Meditec	Time-domain	18	2,000
Stratus OCT	Carl Zeiss Meditec	Time-domain	10	400
3D OCT 2000	Topcon	Spectral-domain	5	27,000
OCT SLO	Optos	Spectral-domain	<6	27,000
3D OCT-1 Maestro 2	Topcon	Spectral-domain	3-6	50,000
Cirrus OCT	Carl Zeiss Meditec	Spectral-domain	5	68,000
iVue and iFusion	Optovue	Spectral-domain	5	26,000
RTVue-100	Optovue	Spectral-domain	5	26,000
RTVue XR Avanti	Optovue	Spectral-domain	5	70,000
Spectralis OCT	Heidelberg	Spectral-domain	4-7	40,000
Envisu C2300	Leica	Spectral-domain	3	32,000
CASIA SS-1000	Tomey	Swept-source	10	30,000
DRI OCT Triton*	Topcon	Swept-source	8	100,000
ANTERION*	Heidelberg	Swept-source	<10	50,000

^*Not available in the USA

1. How to obtain HR-OCT images:

HR-OCT images can be obtained in a rapid, non-contact, and reproducible fashion in the clinic setting. It is helpful for a clinician to see the patient at the slit lamp first to best identify the locations that need imaging, especially if the lesion is subtle. Trained technicians can obtain images at these identified locations.

Various commercially available anterior segment OCT platforms are now being used to image the anterior segment and are summarized in Table 1.⁸

Here are the steps to successfully obtaining a HR-OCT image:

First, the lesion of interest must be identified by the clinician and shown to the technician. The clinician should create diagram of the lesion and surrounding areas for the technician to help with OCT image acquisition. When obtaining the HR-OCT image, the en-face image (adjacent infrared image that shows where the image cut is located) will aid the technician in localizing the lesion, but in cases of smaller lesions, this will be more difficult for the technician to identify.

Multiple cuts should be obtained through the lesion of interest to help characterize the entirety of the lesion. These scans should be done systematically and spaced about 1 mm or less apart. When imaging the cornea, cuts are usually obtained in a horizontal and vertical fashion, but depending on the location of the lesion, other angles should be used for best results. If a lesion is located at the limbus, images are best obtained when the image cuts are aimed perpendicular to the limbus with radial cuts. If a lesion extends from the conjunctiva onto the cornea, cuts must capture both the corneal and conjunctival portions and can be oriented in various angles to encompass all borders of the lesion. It is also critical to obtain several cuts of the surrounding ocular surface around the lesion of interest as a subclinical disease can be present in adjacent areas. To achieve this in limbal lesions particularly, we recommend aiming each scan radially when scanning at the limbus, as this will cover every clock hour surrounding the lesion. There are locations of lesions such as the palpebral conjunctiva, caruncle and fornices where imaging of the lesion can prove to be more challenging given the need for lid eversion and imaging of thicker surrounding tissues. Imaging these lesions requires more practice and expertise.

A systematic approach to scanning lesions is important when initially imaging a lesion as well as when obtaining images post-treatment (either medical or surgical). When comparing images to determine if a lesion has worsened, improved or changed in nature, it is important to utilize images that were taken in the same location and angle. Especially in patients where a malignant lesion is present in one eye, HR-OCT scans of both eyes should be obtained to ensure there is no subclinical disease in the fellow eye.

In summary, a protocol for systematic scanning of the ocular surface should be established to scan the majority of the ocular surface but focus on the area of pathology. Hopefully, the future will allow for automated en-face scanning of the entire ocular surface, which will obviate the need to localize the lesion prior to scanning and will minimize inadvertently not scanning a subtle area of pathology.

2. How to read HR-OCT images:

A recent study has shown that HR-OCT images can be easily interpreted by the novice as well as experienced clinicians.³² This study showed that with only a brief teaching session of 20 minutes, novices were able to significantly improve their ability to read and interpret HR-OCT images.

HR-OCT can be used to image the ocular surface in a systematic fashion, starting from the outermost tear film to the conjunctiva, individual corneal layers, sclera, angle and finally lenticular structures.

In a normal eye, the layers of the cornea appear on the HR-OCT as follows (Figure 1):

1. The first thin band of hyperreflectivity is the normal tear film.

2. The underlying corneal epithelium is a thin, dark and hyporeflective band with a normal thickness of 50 to 70 microns. Bowman’s layer can be seen as a linear structure with reflectivity similar to the stroma directly underneath the stroma.

3. The underlying stroma is thick with more variable hyperreflectivity than the epithelium and is usually 400 to 500 microns in thickness.

4. Lastly, the endothelium is the thinnest layer, usually seen as a thin, hyperreflective line of 5 microns thickness. The normal Descemet’s membrane is difficult to visualize.

Figure 1:

A. High-resolution optical coherence tomography (HR-OCT) image of a normal cornea. Solid white arrow denotes the thin, hyperreflective tear film. Dashed white arrow denotes normal thin, hyporeflective corneal epithelium. Asterisk denotes corneal stroma. White arrowhead denotes the hyperreflective corneal endothelium.

When viewing the bulbar surface, the structures seen are as follows (Figure 2):

1. The conjunctival epithelium is a thin, hyporeflective band, ranging from usually 50 to 70 microns in thickness.

2. The epithelium overlies a band of bright hyperreflectivity which represents the subepithelial region, namely the substantia propria.

3. Lastly, a thick band of moderately hyperreflective tissue can be seen beneath which is the normal scleral tissue.

Figure 2:

A. HR-OCT image of normal conjunctiva. White arrow denotes normal, hyporeflective conjunctival epithelium. Black arrow denotes the hyperreflective, subepithelial tissue of the substania propria. White arrowhead denotes the well circumscribed band of moderately hyperreflective scleral tissue.

When obtaining a HR-OCT image through an ocular surface lesion, the following steps should be taken:

1. First, the clinician must first identify the band of normal epithelial tissue.

2. Then, the clinician needs to determine if the epithelium in the region of the lesion is normal, thin, or thickened and if the epithelium is hyperreflective or hyporeflective.

3. The measurement tool on OCT platform can easily be used to measure the different layers of the cornea to determine if layers are of the appropriate thickness or abnormal.

4. The location of the lesion in question needs to be identified. Is the lesion epithelial or subepithelial? If the lesion is contained within the epithelium or the epithelium is abnormal in appearance, the lesion is epithelial in nature. If the lesion is located below the epithelial layer, it is subepithelial. The differential of the imaged lesion can then be broadened based upon its epithelial or subepithelial location and reflectivity characteristics (Figure 3).

Figure 3:

Landmarks used to read a HR-OCT image. First the normal epithelium is identified. It is thin and hypoflective (dashed white arrow). Then the inferior border of the epithelium is identified (white arrowheads). The area of abnormal epithelium is identified and is noted to be thickened and hyperreflective (black asterisk). There is an abrupt transition from the normal to abnormal epithelium (solid white arrow). This epithelial lesion is an ocular surface squamous neoplasia (OSSN).

3. Characteristics of lesions on HR-OCT:

3.1. Benign Lesions:

3.1.1. Pterygium

Pterygia clinically present as fleshy wing-shaped growths from the conjunctiva extending into the cornea. Clinically, pterygia and OSSN may present as very obvious lesions, but in some cases, the diagnosis may be unclear. Similarly, a pterygium might harbor an occult neoplasia.²³ HR-OCT can reproducibly differentiate between pterygia and OSSN by identifying statistically significant differences in epithelial thickness and location of the primary lesion (epithelial location for OSSN and subepithelial location for pterygia), and morphological features.^21,23,24 In studies using a custom device of 2 to 3 microns and the commercially available RTVue (Optovue, Fremont, CA), an epithelial thickness of 120 to 140 microns was noted to be highly sensitive and specific for OSSN.^21,24

HR-OCT images of pterygia demonstrate thin or normal thickness epithelium with varying levels of hyperreflectivity. The epithelium over a pterygium is most often a bit hyperreflective secondary to actinic changes. At times, this thin band of hyperreflective epithelium overlying the pterygia can appear to abruptly transition to an area of normal epithelium on the HR-OCT image; however, the epithelium is thin, and is not neoplastic in these cases. To differentiate these cases from OSSN, it is vital to note the thickness of the hyperreflective area, especially in the conjunctival epithelium. In pterygia, the subepithelial layer is dense, hyperreflective, with a fibrillary lesion that is located between the conjunctival epithelium and the sclera over the bulbar surface. When the lesion encroaches on the cornea, a brightly hyperreflective tissue is seen between the corneal epithelium and Bowman’s layer (Figure 4).

Figure 4:

A. Slit lamp photograph of a pterygium on the nasal bulbar conjunctiva of the left eye encroaching the nasal limbus. B. HR-OCT image of the pterygium. The overlying epithelium is slightly hyperreflective likely due to actinic changes (white asterisk). There is no abrupt transition in the epithelium from an area of hyperreflectivity to hyporeflectivity. The subepithelial layer shows dense, hyperreflective, stringy, fibrillary lesion that is between the conjunctival epithelium and the sclera over the bulbar surface (white arrow). In the area of the pterygium encroaching on the cornea, a brightly hyperreflective tissue is seen between the corneal epithelium and Bowman’s layer (dashed white arrow).

In contrast, a pinguecula has similarly thin epithelium with variable hyperreflectivity and a dark subepithelial hyporeflective thickened lesion that does not encroach onto the cornea. (Figure 5). Generally, a pingueculae is a straight forward clinical diagnosis but if atypical, OCT can certainly be employed if there is any concern or question for neoplasia.

Figure 5:

A. Slit lamp photograph of a pinguecula on a bulbar conjunctiva of the right eye. B. HR-OCT image of the pinguecula (black arrow). The epithelium is thin with variable hyperreflectivity, overlying a dark homogenous lesion with mild shadowing (white arrow).

3.1.2. Conjunctival Nevus

Nevi are benign lesions on the ocular surface that are mostly pigmented (51%) but can also be partially pigmented (28%) or amelanotic (21%). The classic feature of nevi that suggests its benign nature is the presence of cysts, found in 57 to 65% of the cases. Even though nevi are benign lesions, there is a 0.7% risk of conversion to melanoma. As such, close follow-up is important.³³

Conjunctival nevi often have normal thickness or slightly thickened epithelium overlying a well-circumscribed subepithelial lesion on HR-OCT. They classically contain cystic spaces which are suggestive of chronicity. In pigmented lesions, the HR-OCT can demonstrate cysts within a nevus which may not be clinically evident, helping the clinician compare to a more malignant conjunctival melanoma (Figure 6). HR-OCT imaging is also helpful in amelanotic nevi in children where the cysts may be difficult to discern clinically but are easily visualized on HR-OCT images (Figure 7). In children specifically, junctional nevi can be seen. Given that junctional nevi are present between the epithelium and subepithelial space, HR-OCT images can sometimes show thickened epithelium with an underlying cystic subepithelial lesion in these cases.

Figure 6:

A. Slit lamp photograph of a pigmented conjunctival nevus. B. HR-OCT of the pigmented nevus shows a well-circumscribed subepithelial lesion that is hyperreflective containing cysts (white arrow).

Figure 7:

A. Slit lamp photograph of an amelanotic, gelatinous appearing lesion in a male child. B. On HR-OCT, white arrow denotes a hyperreflective, subepithelial lesion with the presence of cysts (white arrow), consistent with an amelanotic nevus.

3.1.3. Primary Acquired Melanosis (PAM)

Primary acquired melanosis (PAM) is a flat lesion with diffuse pigment and a “pepper” appearance over the conjunctiva, typically present on the bulbar conjunctiva. However, the palpebral or tarsal conjunctiva as well as caruncle can have pigment as well. A thorough clinical examination with eversion of the upper and lower eyelids is crucial to locate all pigment. PAM carries a 13,50% risk for evolution to conjunctival melanoma in the presence of atypia.^34,35,36 HR-OCT may help differentiate PAM from other types of pigmented lesions, but atypia can only be assessed with a conjunctival biopsy. HR-OCT scans are usually best when obtained horizontally, with multiple cuts encompassing the pigmented area with adjacent normal tissue if available. The clinician should inform the technician on the distribution of the pigment, again with a drawing, especially in cases where tarsal or forniceal pigment is present. Images may be harder to obtain over an everted lid or over the caruncle.

HR-OCT images of PAM will demonstrate a normal thickness epithelium with high reflectivity of the basal epithelium, which corresponds to the precipitation of pigment in that layer. (Figure 8) No discreet contiguous subepithelial mass should be observed as this would be characteristic of a conjunctival melanoma.

Figure 8:

A. Slit lamp photograph illustrating a pigmented lesion on right eye temporally consistent with primary acquired melanosis (PAM). B. HR-OCT shows an epithelial layer of normal thickness with uniform hyperreflectivity of the basal layer epithelium (white arrows).

3.1.4. Conjunctival papilloma

Conjunctival papillomas are acquired benign tumors of the conjunctival epithelium. Clinically, they can appear as sessile or pedunculated lesions with numerous fronds of epithelium surrounding a vascular core; other lesions can exhibit an inverted growth pattern.^37,38

HR-OCT images of conjunctival papillomas typically how thickened, dome-shaped elevations of hyperreflective epithelium. Often, vascular cores may be visualized which emanate from the substantia propria. The transition between normal and abnormal epithelium can be abrupt or gradual. With increased thickness of these lesions, posterior shadowing occurs which precludes visibility of the structural details of these lesions on HR-OCT. (Figure 9.) ³⁸

Figure 9:

A. Slit lamp photograph illustrating a conjunctival papilloma. B. HR-OCT shows dome shaped elevation of thickened hyperreflective epithelium (dashed white arrow) with vascular cores (white arrowheads).

3.2. Malignant Ocular Surface Tumors:

3.2.1. Ocular Surface Squamous Neoplasia (OSSN)

OSSN is an umbrella term that includes a spectrum of conjunctival and corneal lesions ranging from conjunctival intraepithelial neoplasia (CIN) (mild, moderate, severe and carcinoma in situ) to invasive squamous cell carcinoma (SCC). Clinically, OSSN has several manifestations: leukoplakic, gelatinous, papillary, nodular or opalescent lesions that can appear on the conjunctiva and/or cornea.³⁹

HR-OCT has been shown to be a highly sensitive and specific imaging modality for this pathology.^21,24 Changes in the epithelium of this epithelial malignancy can be identified easily and distinguished from other subepithelial lesions. OSSN classically has three distinctive features on anterior segment OCT: 1) a thickened 2) hyper-reflective epithelial layer with 3) an abrupt transition from normal to abnormal epithelium.

HR-OCT images of OSSN lesions can appear differently depending on the location and size of the tumors. In OSSN lesions involving only the corneal tissue, HR-OCT images will show hyperreflective lesions with an abrupt transition from normal to abnormal epithelium; however, these lesions may only be marginally thicker than the surrounding epithelium. When lesions are localized only to the epithelium (non-invasive), a distinct plane between the lesion and underlying tissue may be visible (Figure 10). In contrast, larger conjunctival lesions, while displaying characteristic features, can also have posterior shadowing due to the thickness of the lesion, obscuring visualization of some or all of the deeper margins (Figure 11). When lesions span both the cornea and conjunctiva, it is important to obtain HR-OCT image cuts through both the corneal and conjunctival regions to ensure the entirety of the lesion is captured. Multiple radial cuts should be done in every clock hour to ensure no subclinical disease is present in adjacent tissue.

Figure 10:

A. Slit lamp photography of opalescent corneal lesion consistent with corneal intraepithelial neoplasia (CIN). B. HR-OCT showing thickened hyper-reflective epithelium (white asterisk) throughout the cornea with an area of transition from normal to abnormal corneal epithelium (white arrow).

Figure 11:

A. Slit lamp photograph of a right eye with a gelatinous, papillomatous ocular surface squamous neoplasia (OSSN) on the nasal cornea and bulbar conjunctiva. B. HR-OCT image of the OSSN. First, the band of thin hyporeflective normal epithelium is identified (dashed white arrow). Then, the inferior border of the conjunctival epithelium is identified (white arrowheads). The white arrow denotes the abrupt transition between normal and abnormal epithelium. Adjacent to this transition is a thickened, hyperreflective epithelial lesion consistent with an OSSN (white asterisk). This epithelial lesion is overall contained within the epithelium and there is a distinct plane between the lesion and the underlying substantia propria. However, there is some posterior shadowing along the temporal aspect of the lesion (white cross).

HR-OCT images of conjunctival papillomas can appear similar to images of OSSN lesions. The presence of fibrovascular cores seen in papillomas may help to differentiate from OSSN lesions, but the often dense shadowing seen in conjunctival papillomas may obscure adequate visualization.

3.2.2. Conjunctival Lymphoma

Conjunctival lymphoma has various clinical manifestations including a classic salmon patch lesion but can also appear as subconjunctival masses or nodules, or a chronic follicular conjunctivitis.³⁹

On HR-OCT, conjunctival lymphoma is characterized by a normal layer of epithelium overlying homogenous, dark, hyporeflective subepithelial lesions with smooth, discrete borders. These lesions are typically bordered superiorly and inferiorly by a band of hyperreflective tissue. The lesions contain monomorphic, dot like infiltrates that correspond to the infiltration of monoclonal lymphocytes on histopathology.⁴⁰ (Figure 12) The hyperreflective band seen under the epithelium superior and inferior to the lesion is likely compressed substantia propria on the bulbar conjunctiva flanking the lymphocytic infiltrate.

Figure 12:

A. Slit lamp photograph of a conjunctival lymphoma presenting as a salmon-colored inferior forniceal mass. B. HR-OCT of the conjunctival lymphoma. The white arrows delinate normal hyporeflective epithelium overlying a hyporeflective, homogenous lesion with smooth borders containing monomorphic dot-like infiltrates (white asterisk). The lesion is bordered superiorly and inferiorly by a band of hyperreflective tissue (dashed white arrows). The hyperreflective band is likely compressed substantia propria on the bulbar conjunctiva flanking the lymphocytic infiltrate. The monomorphic dot-like infiltrates correspond with monoclonal lymphocytic infiltration on histopathology.

3.2.3. Conjunctival Amyloidosis

Conjunctival amyloidosis can clinically appear similar to lymphoma, manifesting as pink fleshy lesions on the ocular surface. Amyloidosis can also appear as gelatinous lesions, recurrent subconjunctival hemorrhages or even waxy, yellow lesions.⁴¹

On HR-OCT, images show normal epithelium with underlying heterogeneous, dark lesions with irregular, diffuse borders, as compared with the homogenous and regular appearance of lymphomas. These subepithelial lesions often contain paucicellular deposition and hyperreflective linear infiltrates.⁴⁰ (Figure 13).

Figure 13:

A. Slit lamp photograph of a conjunctival amyloidosis shown as a yellow, gelatinous lesion on the superior bulbar conjunctiva, extending into the superior fornix. B. HR-OCT of conjunctival amyloidoisis. There is a subepithelial, hyporeflective lesion with irregular borders and the presence of hyperreflective linear opacities (white arrow). This heterogeneous infiltrate is consistent with conjunctival amyloidosis.

3.2.4. Conjunctival Malignant Melanoma (CMM)

Conjunctival malignant melanoma (CMM) is classically pigmented but in 20% of the cases it can be amelanotic or with mixed pigmentation in another 20%. CMM is classically elevated, immobile and vascularized. CMM most commonly arises from primary acquired melanosis (74%) but can also arise de novo (19%) and from pre-existing conjunctival nevi (7%).⁴²

HR-OCT images of CMM typically show a highly elevated, hyperreflective, subepithelial lesion. The epithelium is normal or slightly thickened with variable hyperreflectivity of the basal epithelium, which suggests involvement of the epithelium with atypical melanocytes (Figure 14). The main subepithelial mass is moderately hyperreflective often with shadowing due to mass/thickness. This imaging is especially helpful in cases differentiating cases of pigmented OSSN from CMM or conjunctival nevi.

Figure 14:

A. Slit lamp photograph of a pink limbal lesion with areas of pigmentation, consistent with a malignant melanoma of the conjunctiva. B. HR-OCT of conjunctival melanoma. The image shows normal thin epithelium at the corneal border (white arrow) which then transitions to hyperreflective epithelium. There is a large, subepithelial lesion that is mostly, hyperreflective (white asterisk) with significant posterior shadowing (white cross). Histopathology confirmed conjunctival melanoma. C. Ultrasound biomiscropic (UBM) image shows a well demarcated, dome-shaped, slightly hypoechoic mass on the sclera without scleral invasion (white asterisk). Note that the UBM allows good visualization of the posterior tumor border.

4. Comparison of HR-OCT with UBM technology

UBM technology has also been employed as a tool to help diagnose conjunctival tumors.⁴³ Unlike OCT technology which utilizes wavelengths of light and is non-contact UBM uses 50 MHz sound waves via a probe immersed in a water bath on the ocular surface to produce high-resolution images.⁴⁴ With a resolution of approximately 50 um and excellent depth of penetration, UBM is most helpful with intraocular tumors and determining possible invasion of conjunctival lesions. As discussed earlier in this manuscript, HR-OCT, with axial resolutions as high 3 microns, better demonstrates the intrinsic details of ocular surface lesions. Oftentimes, HR-OCT used in conjunction with UBM can allow for optimal visualization and characterization of ocular tumors.⁴⁵

Studies have compared UBM and HR-OCT imaging platforms to determine their usefulness for various ocular surface pathologies. Particularly for conjunctival melanomas, UBM can be more beneficial to help distinguish intraocular invasion of a conjunctival melanoma from extraocular extension of a uveal melanoma.⁴⁶ When imaging various tumors of the anterior segment, UBM was found to provide better resolution of the posterior margin of tumors especially in pigmented conjunctival lesions (Figure 14C). UBM had less posterior shadowing while HR-OCT provided better resolution of the anterior margin and anterior segment anatomy.³⁰ Similarly for non-pigmented iris tumors, UBM was found to provide superior imaging quality when assessing posterior tumor surfaces and tumor thickness (Figure 15).⁴⁷ However, when imaging conjunctival nevi, HR-OCT was found to be superior to UBM to measure the epithelial layer as well as visualize intralesional cysts, which are highly characteristic for this type of lesion.⁴⁸

Figure 15:

A. Slit lamp photograph of a pink, firm, immobile lesion on the inferotemporal bulbar conjunctiva. Histology confirmed conjunctival squamous cell carcinoma (SCCA). B. HR-OCT image of SCCA. Note the finger-like invasion of the thickened hyperreflective epithelium from the surface epithelium to the subepithelial space (white asterisk). Posterior shadowing is present along the temporal aspect of the lesion (white cross). C. UBM image shows a dome-shaped slightly hypoechoic lesion over the sclera, extending from the limbus to the anterior equator, without scleral invasion (white asterisk).

Similar to the literature, we have found in our clinical practice that HR-OCT is capable of providing extremely high resolution and detailed views of the internal structures of ocular surface lesions. UBM is more powerful when imaging larger, thicker and pigmented lesions as well as for assessing for deep margins and intraocular extension. Ultimately, histopathology is needed to confirm the diagnosis, even after UBM and HR-OCT imaging.

Conclusion

The purpose of this paper was to focus on the evolution and applicability of HR-OCT and provide a step-by-step approach to the clinician of how to interpret HR-OCT images of ocular surface lesions. A systematic approach is required when approaching an ocular surface lesion using HR-OCT. Figure 16 delineates our approach to distinguishing epithelial and subepithelial ocular surface lesions based upon distinctive characteristics of the epithelial and subepithelial layers. In conclusion, HR-OCT is a powerful, noninvasive, rapid and reproducible adjunct to the clinical examination for the diagnosis of various benign and neoplastic ocular surface lesions. Its applications are numerous and can enable clinicians to obtain “optical” biopsies in the clinic setting.

Figure 16:

Flow diagram depicting an algorithm to analyze HR-OCT images of common ocular surface lesions.

Expert Opinion:

HR-OCT offers immense value for clinicians to diagnose and differentiate between ocular surface lesions. To incorporate this technology into clinical practice, the physician needs to first invest in an OCT platform (Table 1) that they can include in their office diagnostic imaging suite. Technicians need to be educated on the use of the OCT platform and must learn to systematically image the lesion and area of interest as outlined in this paper. Since imaging of the ocular surface is not yet automated, communication between the physician and imager is essential.

The clinician will need to interpret the HR-OCT images, paying close attention to the location of the lesion (epithelial versus subepithelial), and specific characteristics of the lesion (thickened, hyperreflective or hyporeflective, presence of infiltrates etc) (Figure 15). With time, obtaining this in-office “optical biopsy” of lesions can become a routine part of the clinic flow. While an “optical biopsy” cannot completely replace a true tissue biopsy, HR-OCT images can certainly help narrow the differential diagnosis of a particular lesion, expedite the time to diagnosis and allow for initiation of targeted treatment, all of which provide significant cost and time-saving measures for patients.

Additionally, HR-OCT images can be acquired at sequential follow-up visits to determine response to treatment modalities, to monitor for subclinical recurrence and to serve as visual aids for patients to see their own progress.^7,22,49 In the case of topical chemotherapeutic treatment for OSSN, Abou-Shousha et al showed that HR-OCT can document residual disease and help to prevent premature termination of therapy.⁴⁹ A recent study by Tran et al found that HR-OCT detected sub-clinical OSSN in 17% of patients who were deemed to have complete clinical resolution by slit-lamp examination. These patients were then treated with additional cycles of topical chemotherapy with subsequently no recurrences after 24 months of follow-up.⁵⁰ Both these studies highlight the power of HR-OCT to detect residual disease and prevent premature termination of therapy.

In addition to aiding with the diagnosis of various ocular surface lesions as discussed in this paper, HR-OCT has also shown promise in the diagnosis and treatment of keratoconus,⁵¹ dry eye disease,⁵² peripheral corneal thinning,⁴ corneal dystrophies⁵³ as well as infectious keratitis.^54,55 This technology has also emerged as a tool to assist surgeons intra-operatively during corneal transplantation surgery⁵⁶ to ensure correct orientation and complete attachment of corneal grafts⁵⁷ as well as post-operatively to detect early corneal edema, graft detachments¹⁸, or dehiscence of keratoprostheses.^8,58

Future Directions:

Various modalities of OCT technology (such as HR-OCT, spectral domain OCT, enhanced depth imaging OCT. and swept-source OCT⁵⁹) can be implemented for the diagnosis, treatment planning and monitoring of response of various eyelid and adnexal,⁶⁰ iris, ciliary body as well as choroidal and retinal tumors.⁶¹ Recent work by Karp et al has also highlighted the ability of HR-OCT to accurately map tumor borders of OSSN lesions undergoing excisional surgery; histologic tumor margins confirmed after surgical excision corresponded to HR-OCT predicted margins in all cases. The accurate identification of tumor margins by OCT imaging can aid in reducing incidence of residual disease, recurrence as well as minimize healthy tissue removal. This underscores the need for integration of HR-OCT technology into surgical microscopes to facilitate “real-time” intraoperative imaging of tumor margins.⁶²

Future research with HR-OCT can also be focused on creating artificial intelligence (AI) driven algorithms that can help clinicians interpret the acquired images. AI can flag abnormal findings on an HR-OCT image of an ocular surface lesion, such as a pterygium or OSSN, and can provide clinicians with an index of suspicion for a specific differential diagnosis. These AI algorithms for ocular surface lesions can be similar to what is currently available for the diagnosis of keratoconus with various corneal imaging modalities.^63,64,65,66

Similarly, exploring the ability of swept-source OCT technology to simultaneously acquire numerous longitudinal and transverse scans can potentially create 3-dimensional views that can help clinicians further understand dimensions and depth of ocular surface lesions.⁶⁷ Integration of swept-source OCT technology into operating microscopes could help surgeons identify the peripheral and deep margins of ocular surface tumors when performing a surgical excision.

Finally, OCT angiography is also being studied with the aim of identifying distinctive angiographic characteristics in the diseased as well as normal ocular surface.^68–70 Work by Akagi et al has shown differing vessel patterns in the superficial and deep layers of normal conjunctival epithelium ⁷⁰ while Liu et al have found higher blood vessel densities in the body of OSSN tumors as compared with adjacent subepithelial tissue.⁶⁹ Detecting and characterizing abnormalities in flow patterns and vessel density in normal as well as diseased eyes can help clinicians better understand changes in ocular surface blood flow associated with ocular surface lesions. Implementation of this OCT angiography in the clinic setting can facilitate acquisition of more accurate and detailed optical biopsies of ocular surface lesions.

In the next five to ten years, we look forward to seeing the advances of HR-OCT technology. Optimization of this technology to include angiography, to produce even higher-resolution images and incorporation of these platforms into clinics and operating microscopes can help change the paradigm we use to diagnose and treat patients with ocular surface diseases.

Table 2:

Summary of anterior segment OCT findings for various benign and malignant ocular surface lesions

Pathology	Epithelial characteristics	Subepithelial findings	Additional findings
Pterygium	Normal thickness with variable hyperreflectivity	Dense, hyperreflective, fibrillary, subepithelial lesion on bulbar conjunctiva	When on cornea, a hyperreflective band is often seen between corneal epithelium and Bowman’s layer
Pinguecula	Normal thickness with variable hyperreflectivity	Dark hyporeflective subepithelial lesion, homogenous appearance
Conjunctival nevus	Normal	Well-circumscribed subepithelial lesion	Often contains cysts
Conjunctival papilloma	Thickened and hyperreflective	Vascular cores emanating from substantia propria	Significant shadowing precludes detailed visualization
Ocular surface squamous neoplasia, non-invasive	Thickened, hyperreflective epithelial lesion with abrupt transition from normal to abnormal epithelium	No subepithelial component
Conjunctival lymphoma	Normal	Homogenous, hyporeflective, subepithelial lesion, discrete and smooth borders, stippled monomorphic infiltrates	Often has a hyperreflective band superior and inferior to the lesion
Conjunctival amyloidosis	Normal	Homogenous, hyporeflective, subepithelial lesion, irregular diffuse borders, hyperreflective linear material
Primary acquired melanosis	Normal thickness with hyperreflectivity of basal layer	No subepithelial component
Conjunctival melanoma	Normal to slightly thickened with variable hyperreflectivity of basal layer	Subepithelial, hyperreflective lesion

Article Highlights:

• This article provides a step-by-step approach for the acquisition and interpretation of high-resolution anterior segment optical coherence tomography (HR-OCT) images of benign and malignant ocular surface lesions.

• Ocular surface lesions that have distinctive findings on HR-OCT include pterygia, pinguecula, conjunctival nevi, conjunctival papilloma, ocular surface squamous neoplasia (OSSN), conjunctival lymphoma, conjunctival amyloidosis, primary acquired melanosis and conjunctival melanoma, as described in this paper.

• HR-OCT can allow clinicians to obtain in-office “optical biopsies.” While optical biopsies do not always replace tissue biopsies, they can facilitate characterization and more expedient diagnosis of ocular surface lesions.

• HR-OCT can be utilized to detect sub-clinical disease, especially in cases of OSSN, and can help monitor response to medical and surgical therapy.

• Future research and clinical directions for HR-OCT include intra-operative mapping of tumor margins during surgical excision procedures, creation of artificial intelligence algorithms for the rapid interpretation of images and use of OCT angiography to identify angiographic characteristics of various ocular surface lesions.