Optical Coherence Tomography (OCT)
James Strong, CRA, OCT-C
Penn State Hershey Eye Center
Optical Coherence Tomography (OCT) is the most valuable advance in retinal diagnostic imaging since the introduction of fluorescein angiography in 1959. OCT is a non-invasive imaging technique relying on low coherence interferometry to generate in vivo, cross-sectional imagery of ocular tissues. Originally developed in 1991 as a tool for imaging the retina, OCT technology has continually evolved and expanded within ophthalmology as well as other medical specialties. Specialized anterior segment OCT machines became available in 2005 and the introduction of Spectral (Fourier) Domain OCT (SD-OCT, FD-OCT) technology now provides greater tissue resolving power, significantly higher scan density, and faster data acquisition than original Time Domain OCT.
Cross-sectional visualization is an extremely powerful tool in the identification and assessment of retina abnormalities. The high resolving power (10um - Time Domain, 5um – Spectral Domain) provides excellent detail for evaluating the vitreo-retinal interface, neurosensory retinal morphology, and the RPE-choroid complex. The ability to perform volumetric and retinal thickness analysis also provides a quantitative and repeatable method to evaluate surgical and pharmacological interventions.
Individual high resolution line scans are a simple way to identify overt as well as very subtle retinal interface pathologies, such as a persistently adherent posterior hyaloid, fine epiretinal membranes, and vitreomacular traction. In a procedure that is easily tolerated by most patients, well-placed line scans can differentiate between pseudo holes, lamellar holes and full thickness macular holes with a high degree of confidence. Line scans can also confirm the presence of retinal edema from various causes. When combined with serial thickness map or volume analysis, these different data sets provide a detailed picture of disease progression or therapeutic response.
OCT is also quite useful in the assessment of subretinal fluid, neurosensory detachments, pigment-epithelial detachments, and choroidal neovascular membranes. OCT confirmation of persistent subretinal fluid can influence the treatment plan when considering intravitreal injection therapy. RPE irregularities associated with both wet and dry AMD can be monitored using line scans. With experience, OCT imaging may allow differentiation between wet and dry AMD eliminating the need for more invasive testing such as fundus photography and fluorescein angiography; while in other cases OCT is a valuable adjunct to these modalities.
Utilization of OCT imaging as a pre and post surgical assessment tool can provide invaluable information in the surgical management of macular holes and retinal detachments. OCT can provide visualization of surgical outcomes, confirming reattachment and normal contour. Immediate post-surgical imaging can sometimes be challenging due to ocular turbidity that can result in significantly reduced OCT signal strength; but images adequate for subjective, if not quantitative, interpretation can usually be obtained. Pre-surgical scanning, especially in cases of poor ocular media, can often reveal pathologies that could complicate surgery, such as the presence of undetected macular hole, CNV, edema or VMT. The OCT's scanning beam technology allows successful imaging even through a small pupil or tiny peripheral opening in a dense cataract that would otherwise confound thorough ophthalmoscopic examination.
Optical Coherence Tomography generates cross sectional images by analyzing the time delay and magnitude change of low coherence light as it is backscattered by ocular tissues. An infrared scanning beam is split into a sample arm (directed toward the subject) and a reference arm (directed toward a mirror). As the sample beam returns to the instrument it is correlated with the reference arm in order to determine distance and signal change via photodetector measurement. The resulting change in signal amplitude allows tissue differentiation by analysis of the reflective properties, which are matched to a false color scale. As the scanning beam moves across tissue, the sequential longitudinal signals, or A-scans, can be reassembled into a transverse scan yielding cross-sectional images, or B-scans, of the subject. The scans can then be analyzed in a variety of ways providing both empirical measurements (e.g. RNFL or retinal thickness/volume) and qualitative morphological information.
While OCT imaging provides a unique perspective for evaluating any number of retinal conditions, clinicians should be aware that scans and analysis are not without fault. Poor ocular media, patient compliance and even saccadic movement can introduce image artifacts that can masquerade as pathology. Even more problematic are the occasional software algorithm failures or the inappropriate application of certain analysis algorithms by undertrained operators. Certain image artifacts and analysis errors can be identified by characteristic patterns; however there are instances when it may be necessary to review the raw scan data to determine the nature of inconsistent results.
Spectral (Fourier) Domain OCT
Spectral or Fourier Domain OCT (SD-OCT) is the newest technological variant of this evolving modality. SD-OCT provides nearly a 100 fold increase in the amount of data captured as well as a significant increase in axial resolving power. In comparison, time domain functions at a rate of 400 A-scans per second, while SD instruments can perform tens of thousands of A-scans per second. This dramatic increase in scanning speed allows for greater data acquisition with lower likelihood of motion artifacts and a much finer raster pattern of B-scans. SD axial resolution currently ranges from 3-7 microns, an improvement over the 8-10 microns of time domain OCT. Transverse resolving power is a function of both instrument and ocular optics, and therefore has not improved with this generation of OCTs. Ongoing research using adaptive optics to correct existing optical aberrations in OCT has demonstrated the potential for lateral resolution to the cellular level.
Cross-sectional visualization of retinal pathology is improved with SD-OCT. "High” resolution B-scans are comprised of 4,096 A-scans, compared to time domain's maximum of 512 A-scans. With this level of detail, the photoreceptors, RPE, and choriocapillaris can be distinguished as distinct layers. Subtle disruptions within various retinal layers can be pinpointed and it is now possible to recognize lesions which were well beyond the resolution limit of traditional OCT. In cases of severe morphological remodeling, SD-OCT provides a view that allows greater certainty in confirming a suspected etiology. The increased signal "sensitivity” also allows a cleaner view of vitreous body and interface echoes.
Rapid scan acquisition further increases the diagnostic accuracy of SD-OCT by eliminating the use of alignment algorithms to correct patient movement in "lengthy” TD scans. This speed has facilitated scanning patterns which can cover up to a 6mm x 6mm area of retina with B-scans spaced so tightly that the data can be presented as a three-dimensional cube. Cubed data can be rotated, flipped, and viewed from any perspective. These dense scan patterns now also provide continuous C-scan information. Time domain OCT volumetric maps attempt to generate this z-plane tomographical information, but the speed-limited scan patterns necessitate extensive extrapolation of data for areas of retina that falls between B-scans. Another benefit of the dense raster patterns of SD-OCT is consistent lesion comparison over time. Tightly spaced B-scans insure that all but the smallest of lesions are imaged, so location and size can be accurately followed over time.
Three-dimensional visualizations are some of the most significant advances in this generation of OCT evolution. Once cube data has been obtained, it can be rendered in a number of ways. Sophisticated algorithms have been created to identify the major layers of the retina based on interface echo characteristics so that various segments of the retina can be "peeled” away allowing for z-plane viewing for more than just the ILM surface. Manual controls can also be utilized to rotate the cube and pull back any number of individual or combined X and Y scans so that morphology can be examined in the exact context of the retinal 3D matrix.
While the fundamental principle and application of SD-OCT remains the same as traditional Time Domain OCT, it remains uncertain whether this increased level of morphological information will provide greater understanding of the relationship between retinal disease processes and their physiological implications.
SD-OCT relies on three technical modifications to generate increased resolution and high acquisition rates. By utilizing super-luminescent diode (SLD) technology that was not available in the 1990's, SD-OCT utilizes a wider bandwidth light source than TD-OCT, which allows increased axial resolving power. Pulsed-light femtosecond lasers, with even wider bandwidth, can achieve 1-2 micron resolution; however their current cost is prohibitive, which limits them to use to dedicated research instruments.
The primary speed gating factor for TD imaging was the use of a movable mirror in the reference arm scanning path. By eliminating this mechanical limitation and including a spectrometer with a diffraction grating in place of the traditional photodetector, SD-OCT is no longer time-bound. This modification combined with the application of a Fourier transform applied to the reference/sample interferogram allows for the huge increase in scanning and analysis speed. It should be noted however, that faster A-scan rates can lead to some loss of image quality which can be somewhat ameliorated using over-sampling techniques.