With the recent increase in the number of OCT users and the number of companies manufacturing commercially available systems, there is a need to establish common clinical standards to allow for consistency across devices and comparison between individuals. Standards that fully exploit and summarize all of the data acquired with 3D OCT imaging would improve diagnostic ability and longitudinal follow-up, and may eventually lead to the use of 3D-OCT parameters as endpoints in clinical trials. To do this, scan patterns, scan density and scan area need to be considered, as do the requirements for establishing and employing normative data for 3D datasets. Because many clinicians continue to use TD-OCT and/or have converted from TD-OCT to 3D-OCT, a way to utilize all of the data acquired with TD-OCT would be valuable for the longitudinal follow-up of patients, especially those that are now being followed by 3D imaging.(Kim et al., 2010
The use of OCT for surgical guidance is a recently explored concept in ophthalmology(Dayani et al., 2009
) but has been used by other specialists to guide cochlear implantation,(Just et al., 2009b
; Pau et al., 2007
) biopsy site localization (Just et al., 2009a
), and diagnosis of vocal fold lesions.(Vokes et al., 2008
) Delivery systems such as hand-held and microscopic probes may enable surgical applications of OCT systems, so that a clinician can have access to cross-sectional information when performing ophthalmic surgery - as opposed to only pre- and post-operative structural data.
4.1. Establishment of New Clinical OCT Standards
While only one company produces a commercial ophthalmic TD-OCT system, several commercially available 3D-OCT imaging systems have developed independently, and the numbers of companies entering this arena continues to grow. Although the basic principles of data acquisition are consistent across devices, scan types, the exact retinal location of scans and other software-based differences exist. The most effective scan specifications, with respect to disease discrimination and longitudinal follow-up, have yet to be established.
4.1.1. Scan Patterns
The basic RNFL and macular scan patterns used in TD-OCT are seen in . These scan types were chosen to sample a region of interest in a space- and time-efficient manner. With 3D-OCT imaging, the most common form of sampling is a data cube as detailed in . Most of the commercial devices can sample a data cube centered on the ONH and another centered on the fovea. However, the number of A-scans per B-scan as well as the number of B-scans per volume is not consistent across devices. In addition, the number of scans in a given area and the overall retinal area a scan samples varies with device.
188.8.131.52. Semi-Isotropic and Semi-Isometric Sampling
While imaging for the purpose of confirming the presence of retinal pathologies may only require a limited number of densely sampled cross-sections, 3D sampling is necessary for quantitative measurements of volumes of tissue in order to assess the true extent of retinal damage. The sampling volume and number of A- and B-scans comprising an image vary amongst the commercially available 3D-OCT devices. Some systems sample the ONH and macular regions semi-isotropically and semi-isometrically (i.e., an equal number of A-scans in the horizontal and vertical directions spaced evenly over a sampling area that is equivalent in the x- and y-direction). Sampling evenly (in terms of number of A-scans and the area they comprise) in the horizontal and vertical directions is semi-isotropic and semi-isometric, as a perfectly isotropic and isometric sample would be equal in x-, y-, and z. However, with 3D-OCT imaging, the depth of penetration is limited to 1–2 mm with a light source centered in the 800 nm region because of the attenuation of backscattered light. Other commercial systems sample an uneven area with a varying number of A-scans in the horizontal and vertical direction.
Semi-isotropic and semi-isometric scanning is an appealing technique for obtaining quantitative measurements because it provides an even representation of a volume of tissue that is required for accurately resampling the data volume after acquisition. As a result, spatial integrity of resampled images can be retained after scanning, and uneven interpolation to correct for the unevenly spaced samples will not be necessary.
A second advantage of semi-isotropic and semi-isometric sampling is relevant to registering OCT fundus images to reference fundus photographs. Image matching that relies on features within an OCT fundus image, such as blood vessels, requires an evenly sampled dataset to match features present on a reference fundus image and preserve their spatial integrity.
184.108.40.206. Scan Density
Increasing the density of a scan, or the number of A-scans acquired within a given volume of tissue, has previously been shown to decrease the variability of RNFL thickness measurements obtained using TD-OCT.(Gurses-Ozden et al., 1999
; Paunescu et al., 2004
) However, the more A-scans acquired, the longer it takes to acquire an image. Decreasing the required subject fixation time is important to ensure eye motion artifacts as well as corneal drying-related signal attenuation are minimized. However, when visualizing retinal pathology or other structures of interest that do not necessitate quantitative measurements, denser B-scans can be acquired. This helps improve structural definition and layer boundaries.
220.127.116.11. Scan Area
Current scanning protocols common amongst commercial systems include 3D scanning of the ONH and macular regions separately. These protocols offer the advantage of being comparable to scans acquired with TD-OCT: radial and peripapillary scans can be resampled from within the 3D-OCT volume. However, as 3D-OCT scan speed continues to improve, it may be possible to image larger volumes for a more global view of pathologic conditions. Visualizing the ONH region in the same 3D-OCT volume as the macula may be useful for monitoring changes in the RNFL that occur with glaucomatous damage.
Similarly, depending the structure of interest, a 3D-OCT volume of a smaller scan area may be used to improve visualization of smaller structures. , for example, shows a 3 × 3 mm scan of the ONH consisting of 300 × 300 A-scans. This high-density scan of a smaller volume helps makes possible the visualization of the lamina cribrosa upon C-mode sectioning. Potsaid et al demonstrated high density scanning of a small area with SS-OCT and were able to obtain images of photoreceptors and retinal capillaries of the inner nuclear layer. (Potsaid et al., 2008
) However, even without the rapid scanning capability afforded by SS-OCT, photoreceptors can be visualized with SD-OCT ().
Figure 24 A high-density (3 × 3 mm; 300 × 300 A-scans) 3D-OCT image of the optic nerve can be used to create C-mode sections of the lamina cribrosa. The black line indicates the location of the cross-sectional scan; the three white lines indicate (more ...)
Figure 25 A high-density (800 × 800 μm, 300 × 300 A-scans) 3D-OCT image of the retina can be used to create C-mode sections of photoreceptors. The horizontal red line indicates the location of the cross-sectional scan; the horizontal blue (more ...)
4.2. Normative Databases for 3D Datasets
Comparing individual thickness measurements to those from healthy, normal subjects is a method used in TD-OCT to highlight regions of abnormal thickness. The commercial TD-OCT system has a RNFL thickness normative database, which is comprised of average measurements from the 3.4 mm diameter peripapillary RNFL thickness protocol taken in healthy subjects. The TD-OCT system also has a retinal thickness normative database, generated from the six radial scan protocol. Some of the commercial 3D-OCT imaging systems have incorporated normative databases into their software but, at present, fail to use all of the available 3D-OCT data. For example, on some devices normative RNFL thickness comparisons still rely only on (resampled) 3.4-mm diameter peripapillary information while the majority of the 3D RNFL thickness information is left unused. However, it is critical to go “beyond the circle.” A method for comparing as much data as possible to normative values needs to be established. Directly comparing subject thickness information point-by-point would be inappropriate because of anatomical variations in thickness patterns and blood vessel locations, and ONH size as well as variability of the scanning window location.
One method for summarizing 3D-OCT thickness map data for glaucoma discrimination has recently been introduced.(Ishikawa H, et al. IOVS 2009
;50:ARVO E-Abstract 3328) This method condenses IRC or GCC data into superpixels (adjacent sampling points) and compares the superpixels to normative thickness superpixel data. Summarizing thickness information into superpixels means the analysis is less likely to be skewed by imaging artifacts or algorithm failure than a simple pixel-by-pixel comparison, but all of the 3D information is being used. In addition, a superpixel approach may allow for detection of localized defects that are missed in the sectoral analysis that is currently used in the macular region. This approach has shown a glaucoma discriminating ability at least equal to that of peripapillary RNFL thickness measurements.
4.3. OCT Measurements as Endpoints
Before 3D-OCT measurements can be used as endpoints in clinical trials, it must be shown that 3D-OCT can measure relevant outcomes accurately and precisely, and that those measurements correspond to clinically important outcomes. A 3D assessment may allow increased precision in measurements and better reproducibility because of registration, but this has yet to be established.
In glaucoma, RNFL thickness changes may be used as a clinical endpoint, since it has been established that the disease specifically affects retinal ganglion cells and their axons. Still, it is critical that a connection can be made between the rate of RNFL loss and clinically relevant vision loss. Does loss of RNFL result in decreased functional ability? If clinical intervention can slow the rate of RNFL loss and this predicts or corresponds to loss of clinically relevant visual function, then 3D-OCT RNFL measurements will have profound importance as a clinical endpoint.
In eyes with retinal pathologies, parameters such as macular thickness, extent of PR IS/OS junction repair, volume of cystic spaces, drusen volume, extent of geographic atrophy, and/or inner/outer retinal thickening may be used as an endpoint, as opposed to a subjective parameter such as best-corrected visual acuity. Studies to evaluate the potential of these 3D-OCT parameters as clinical endpoints are underway.
4.4. Compatibility with TD-OCT for Longitudinal Follow-up
Years of longitudinal patient data have been acquired since TD-OCT became commercially available. With the recent introduction of 3D-OCT imaging methods to the clinic, a method to use this longitudinal data is essential, especially for assessing slowly progressing diseases like glaucoma in patients that have been followed for many years.
One novel approach addresses the need for compatibility between devices.(Kim et al., 2010
) This method searches a 3D-OCT dataset for every possible 3.4-mm circular scan within the boundaries of the 3D-OCT volume, and uses the cross correlation between these virtually resampled circular scans and the TD-OCT 3.4-mm scan to automatically find the TD-OCT scan circle location within the 3D-OCT volume. shows an example TD-OCT circular scan whose location has been matched within the 3D-OCT dataset. This approach has the potential to bridge different iterations of the technology by ensuring that follow-up 3D-OCT data are resampled in the same location as previously acquired TD-OCT data. Several studies have shown that measurements from commercial 3D-OCT systems cannot be interchanged with TD-OCT measurements, and that 3D-OCT structural measurements are typically higher than TD-OCT measurements.(Forooghian et al., 2008
; Han and Jaffe, 2009
; Kakinoki et al., 2009
; Leung et al., 2008b
; Sayanagi et al., 2009
) This is likely in part due to automated segmentation algorithm placement of anterior and posterior borders of structures of interest. The commercial TD-OCT system defines the inner segment/outer segment junction as the outer boundary of the retina, but different 3D-OCT devices may detect other structures such as the RPE.(Han and Jaffe, 2009
; Leung et al., 2008b
) Segmentation differences may be contributing to a constant offset attributed to device calibration, but scan location may also be adding variability and could be addressed using this method of backward compatibility.
Figure 26 Summary of method for backward compatibility between TD-OCT and 3D-OCT: scan location matching. (A, B) TD-OCT Fundus video image and 3.4-mm circular cross-sectional OCT B-scan; (C, D) 3D-OCT fundus image and virtually resampled 3.4-mm circular B-scan; (more ...)
4.5. Intraoperative OCT Surgical Guidance
Recently, the use of OCT for surgical guidance has been considered in fields outside of ophthalmology. Pau et al employed an OCT system that was incorporated into an operating microscope to image the cochlea to evaluate its use as a potential guide to surgeons performing cochlear implantation.(Pau et al., 2007
) The authors found the use of an OCT guide is feasible and may help surgeons with the placement of the cochlear implant electrode array. A separate study suggested OCT may be used to evaluate the oval window niche.(Just et al., 2009b
) Vokes et al also developed a surgical microscope to enable non-contact imaging of the vocal folds, which they found may be applicable to the diagnosis of vocal fold lesions, apparent in OCT sections as a disruption of the basement membrane.(Vokes et al., 2008
) Others have developed operating microscopes with integrated OCT systems and used them define a biopsy site so a resection can be precisely planned.(Just et al., 2009a
) In addition, since OCT serves as a form of optical biopsy, some groups are attempting to use optical scattering information to evaluate tissues without excision, (Jung et al., 2007
; Poneros, 2004
; Wang et al., 2006
) using endoscopic OCT systems (Jackle et al., 2000
; Li et al., 2000
; Sivak et al., 2000
; Tearney et al., 1997
). Endoscope-based OCT systems also have the potential to guide laser ablation of tissue.(Boppart et al., 1999
In ophthalmology, OCT has been used postoperatively to evaluate outcomes of macular hole surgery,(Inoue et al., 2009
; Michalewska et al., 2008
; Sano et al., 2009
) patients who had undergone vitrectomy surgery for vitreomacular traction,(Chang et al., 2008a
; Mojana et al., 2008
; Uchino et al., 2001
) and eyes after successful surgery for retinal detachment.(Nakanishi et al., 2009
) Leng et al used a 1310 nm SD-OCT system to visualize corneal incisions in the eye of a patient after phacoemulsification.(Leng et al., 2008
While postoperative OCT is useful for evaluating surgical outcomes, real-time imaging of ocular structures during surgery may provide additional guidance to ophthalmologists in the form of a structural perspective that is not currently available. Dayani et al recently used a handheld OCT system before, during and after full-thickness macular hole, epiretinal membrane and vitreomacular traction surgery.(Dayani et al., 2009
) The authors suggest their intraoperative OCT setup may improve surgical outcomes by confirming the removal of the internal limiting membrane or epiretinal membrane.
OCT integrated into an operating microscope with a “heads-up” display would have high utility in intraocular surgery, and would provide invaluable information for intraoperative decision-making. It could also be used in combination with a robotic surgical approach, either by surgeons actually in the operating room or individuals directing the device remotely. In addition, the OCT Penlight described in section 3.4 may provide a structural image overlay onto tissue and allow real-time monitoring of surgical progress, such as microcatheter insertion into Schlemm’s Canal.
4.6. OCT Delivery Systems
OCT systems used in ophthalmology have traditionally required patients to be seated upright and looking at a fixation target. However, as evidenced by the endoscope and surgical microscope systems described in Section 4.4, OCT delivery does not need to be limited to a slit lamp module. With these approaches to image acquisition, structures of interest can be viewed without the requirement that subjects look at a fixation target.
In addition to endoscope and microscope-based systems, there is also clinical potential for a hand-held OCT delivery system. Scott et al used a hand-held SD-OCT system with a flexible arm to image eyes of infants and adults in a supine position.(Scott et al., 2009
) The hand-held probe contained the fiber optics of the sample arm such that the operator had flexibility in aiming. This type of probe may not only assist with scanning younger and less cooperative patients, but may also be useful for patients lying on stretchers or those unable to be positioned using a slit lamp setup.