Ординатура / Офтальмология / Учебные материалы / Uveitis Text and Imaging Text and Imaging Text and Imaging 2009

Figure 44: OCT line scan showing multiple pockets of serous retinal detachments with increased foveal thickness

Figure 45: Repeat OCT line scan after 48 hours of receiving intravenous methylprednisolone showing reduction in retinal thickness over next 48 hours

Figure 46: OCT scan done at one week shows complete resolution of subretinal fluid with normal foveal contour

OCT scan done at one week shows complete resolution of subretinal fluid with normal foveal contour (Figure 46).

Retinal thickness change map shows reduction in retinal thickness in microns (Figure 47).

Figure 47: Retinal thickness change map shows reduction in retinal thickness in microns

KEY POINTS

1.OCT provides high-resolution, cross-sectional tomographic images of the retina and RPE-choriocapillaris.

2.It uses light to detect relative time delay taken by light rays to reflect from various optical interfaces by the use of low-coherence interferometry.

3.The procedure is non-contact, non-invasive, quick, easy to perform with no discomfort to the patient.

4.Clear optical media is required to obtain a good quality scan.

5.It helps in depicting the anatomical location of the lesion, monitoring natural course of the disease, diagnosing and quantifying macular oedema, early diagnosis of CNVM, differentiating CNVM from healed choroiditis scar, and monitoring response to an intervention.

6.OCT can be repeated as often as required.

REFERENCES

1.Huang D, Swanson EA, Lin CP, et al. Optical coherence tomography. Science 1991; 254:1178-81.

2.Swanson EA, Izatt JA, Hee MR et al. In vivo imaging by optical coherence tomography. Opt Lett 1993; 18:1864-66.

3.Puliafito CA, Hee MR, Schuman JS et al. Optical coherence tomography of ocular diseases. Thorofare, NJ: SLACK, 1996.

4.Hee MR, Izatt JA, Swanson EA, et al. Optical coherence tomography of the human retina. Arch Ophthalmol 1995; 113: 325-32.

5.Fujimoto JG, DeSilvestri S, Ippen EP, et al. Femtosecond optical ranging in biological systems. Opt Lett 1986; 11: 150-52.

6.Youngquist RC, Carr S, Davies DE. Optical coherence domain reflectometry: a new optical evaluation technique. Opt Lett 1987; 12:158-60.

7.Gilgen HH, Novak RP, Salathe RP, et al. Submillimeter optical reflectometry. IEEE J Lightwave Technol 1989; 7: 1225-33.

8.Fercher AF, Mengedoht K, Werner W. Eye-length measurement by interferometery with partially coherent light. Opt Lett 1988;13:1867-69.

9.Fercher AF, Hitzenberger C, Juchem M. Measurement of intraocular optical distances using partially coherent laser light. J Mod Opt 1991; 38:1327-33.

10.Martidis A, Lane RG, Puliafito CA. Optical Coherence Tomography. In : Ciulla TA, Regillo CD, Harris A (Eds). Retina and Optic nerve imaging. Lippincott Williams and Wilkins 2003:51-58.

11.Gupta V, Gupta A, Dogra MR. Optical coherence Tomography of macular diseases. UK: Taylor and Francis, 2004:1-13.

12.Antcliff RJ, Stanford MR, Chauhan DS et al. Comparison between optical coherence tomography and fundus fluorescein angiography for the detection of cystoid macular edema in patients with uveitis. Ophthalmology 2000; 107: 593-99.

13.Markomichelakis NN, Halkiadakis I, Pantelia E, et al. Patterns of macular edema in patients with uveitis: qualitative and quantitative assessment using optical coherence Tomography. Ophthalmology 2004;111:946-53.

14.Ciardella AP, Prall FR, Borodoker N, Cunningham ET Jr. Imaging techniques for posterior uveitis. Curr Opin Ophthalmol 2004;15:519-30.

15.Browning DJ, Fraser CM. Optical coherence Tomography to detect macular edema in the presence of asteroid hylosis. Am J Ophthalmol 2004;137:959-61.

16.Antcliff RJ, Spalton DJ, Stanford MR, et al. Intravitreal triamcinolone for uveitic cystoid macular edema: an optical coherence tomographic study. Ophthalmology 2001;108:765-72.

17.Maruyama Y, Kishi S. Tomographic features of serous retinal detachment in Vogt-Koyanagi-Harada syndrome. Ophthalmic Surg Lasers Imaging 2004;35:239-42.

18.Zolf R, Glacet-Bernard A, Benhamou N, et al. Imaging analysis with optical coherence tomography: relevance for submacular surgery in high myopia and in multifocal choroiditis. Retina 2002;22:192-201.

INTRODUCTION TO OCT/SLO

Soon after Optical Coherence Tomography (OCT) was introduced in ophthalmology (see Part A of this chapter), Podoleanu and co-workers (University of Kent, Canterbury, UK) pioneered the development of a different approach to OCT imaging.1 In contrast to conventional OCT imaging which is a compilation of scans along the Z (depth) axis, this method involves fast scanning in the X,Y-plane. It also combines high resolution tomographic OCT images with the surface imaging capability of the scanning laser ophthalmoscope, produced simultaneously (OCT/SLO).2-6 Using this scanning technique, longitudinal OCT scans are produced, similar to those described in the previous chapter. It can also provide two-dimensional images in an orientation familiar to the ophthalmologist. However, these coronal scans are initially difficult to interpret, given the general unfamiliarity with high resolution images in coronal sections.6 This does allow for three-dimensional visualisation of the area under investigation, either mentally or by software rendering.

Coronal scanning is capable of compensating for some of the limitations of conventional OCT (Carl Zeiss Meditec, USA). For one, it is difficult to know the precise location and extent of any longitudinal OCT scan. Secondly, there is no proper recording of eye movement during acquisition. Both make it very hard to appreciate the extent of any pathological change within the area that is presumably scanned. The designers of the conventional OCT tried to overcome this problem by providing the operator with an infrared fundus video image displaying the scan line. However, there is no way of knowing the scan’s exact location as the images are acquired sequentially and not simultaneously. The conventional OCT also ignores any lateral eye movement occurring during the two seconds required to register the 512 A-scans that form a single line. This assumption is difficult to accept, as the speed of saccadic movements can exceed 900° per second,7 while the lateral resolution of the system is around 20 microns, i.e. equivalent to one fifteenth of a degree. Hence, if a saccadic eye move-

ment occurs during image acquisition, the scan’s first section can be imaging a different retinal area than the second. Fixation problems are more common with macular pathology, potentially magnifying the degrees of microsaccadic eye movements as is evident during clinical examination. Axial movement (from breathing, change in muscular tension of the head while seated in the chin rest and pulsatile blood flow in the retina and choroid) can also lead to acquisition shifts. Within the conventional OCT post-acquisition processing using a technique called auto-correlation provides some degree of compensation.8

The strength of the OCT/SLO device lies in its ability to show the location and the extent of any pathology in the central retinal area, which may provide anatomic clues to understanding pathophysiology.6 The fundus image provided by the confocal channel is of high quality, and therefore, better able to indicate the location of the longitudinal scan line. Pixel-to-pixel correspondence between the OCT and confocal images insures that any axial or lateral movement during image acquisition will be recorded and can be visualized on the confocal image as a line shift or blurring. Recently the system has been commercialized by Ophthalmic Technology Inc. (Toronto, Canada). The OCT/SLO is also referred to in the literature as OCT-Ophthalmoscope, en-face OCT or transversal OCT.

PRINCIPLE OF THE OCT/SLO SYSTEM

A schematic diagram of the OCT/SLO is shown in Figure 1.9,10 Major differences with conventional OCT

Figure 1: Diagram of OCT/SLO setup. SLD = super luminescent diode; M = mirror; OCT = optical coherence tomography; PC = personal computer

includes its coronal scanning approach as well as its simultaneously acquired confocal image. Two mirrors are used in the sample arm to acquire the OCT images in the X,Y-plane, which resembles the set-up of a confocal scanning laser ophthalmoscope (SLO).

Similar to the conventional OCT, the system uses a super luminescent diode with a central wavelength of 820 nm and a bandwidth of 20 nm.2,6 The light beam is split, directing one part to the patient’s eye (sample arm) and the other part to the reference arm (mirror). The returning light beams from both the patient’s eye and the reference arm are collected through an interferometer to produce the OCT signal. A fraction of the light returning from the patient’s eye is directed through a pinhole to another detector to produce a confocal signal. The images are produced simultaneously at the same scanning rate, and the images in the confocal and the OCT channels are thus in strict pixel-to-pixel correspondence.9-12

Depth (Z-axis) resolution is around 10 μm and transversal (X-Y plane) resolution is about 15 μm in the OCT channel.11 The confocal channel provides a depth resolution of ~3 mm.9 The focus in the confocal channel is not adjusted when changing the depth value in the OCT channel, and therefore the confocal image looks the same in all the pairs collected in the stack. However, due to eye movements, shifts in the confocal images are observed and the variations from frame to frame can be used later to correct the stack.13

THE OCT/SLO SCANNING REGIMEN: ACQUIRING OCT C- AND B-SCANS

As explained in the first part of this chapter, the longitudinal OCT scans (referred to as OCT B-scans) in the conventional OCT are built up from successive in-depth reflectivity profiles acquired along the Z-axis in the X, Z- or Y, Z-plane (A-scans).8,14,15 The OCT/ SLO produces coronal OCT scans (C-scans) in the X,Y- plane at a fixed Z-coordinate using en-face flying-spot T-scan lines, as shown in Figure 2.4 By changing the Z-coordinate in the OCT channel, OCT C-scans are taken at different depths in the retina. Aside from coronal scans, the OCT/SLO also produces longitudinal B-scans by making T-scans along a fixed axis in the X,Y-plane and continuously changing the Z-coordinate, i.e. depth (see Figure 2). The OCT C- scans are acquired at a rate of 2 frames per second and

Figure 2: Scanning protocols for conventional and en-face OCT scanning

the OCT B-scans are acquired in one second. The scan depth along the Z-axis is set at 1.125 mm, but is adjustable up to more than 2.0 mm. It is also possible to increase the acquisition rate, but both at the cost of image resolution. In the current system, each scan covers an area of 24 by 24 degrees, roughly equivalent to an area of 8 by 8 mm. Both the confocal image and the coronal OCT image are displayed together as a pair, and each part is displayed as an array of 512 × 512 pixels. The images are visualised on a personal computer in grey scale, but can also be displayed in false color.

THE OCT/SLO HARDWARE

The set-up of the commercialised version of the OCT/ SLO is similar to the conventional OCT where the operator is facing the patient as well as the computer screen during acquisition. The patient module consists of an adjustable chin rest. The device is mounted on a motorised table, and consists of the OCT/SLO system with a joystick to move the machine vertically and horizontally, a PC, a flat screen monitor and a foot pedal to operate the acquisition program (Figure 3). The acquisition program is pre-installed, and starts up automatically when the system is turned on. It is not necessary to dilate the patients’ pupil before scanning the patient, although it does make it easier to acquire a scan, especially in patients with very small pupils.

Figure 3: Commercial model [Courtesy of Ophthalmic Technolgies Inc.]

THE OCT/SLO IN CLINICAL USE

Compared to the Stratus OCT, this system asks a little more effort from both the operator and the patient. Operating the machine is fairly similar to the Stratus OCT. However, the operator does need to understand the combined coronal OCT and confocal images in order to take full advantage of this scanning technique. The confocal image supplies the operator with a high quality fundus image which may reveal affected areas by a change in surface reflectivity. When present, these areas require further attention as otherwise, important observations may be missed.

Scanning a patient takes a little longer than with the conventional OCT, although the scanning rate does not differ much when comparing single scan acquisition rates. With the OCT/SLO all captured images are saved, while in the Stratus OCT only one at a time is saved. A single topographic stack of the OCT/SLO is acquired in 2 seconds, which is similar to the Stratus OCT’s fast macular thickness protocol which uses 1.9 seconds. Although the software in the OCT/SLO system offers various scan modalities, we chose to only highlight those we think are suitable in scanning the majority of patients. The best option is to use “Freeform acquisition”. This program allows the operator to switch back and forth from OCT C-scann-

ing to OCT B-scanning as he/she pleases. The images are initially captured as C-scans, and by moving the device gently back and forth the operator can sweep through the macular or ONH area and review the pathology in real-time. By clicking the foot pedal the operator is able to switch to the OCT B-scan mode. The scan line for the OCT B-scan can be placed at any given position along the horizontal or vertical axis by dragging the centre of the line using the mouse. It is also possible to acquire radial lines by dragging the edge of the line to the preferred angle. During OCT B- scanning the coronal confocal image is refreshed every 2 seconds so that the position of the scan line can well estimated. Up to 100 successive scans can be saved. Alternatively, a “Detailed Stack” can be made that captures 50 consecutive OCT C-scans over an operatorselected fixed distance. The default distance for this detailed stack is 1.125 mm, each frame being roughly 20 micron apart.

To evaluate retinal thickness a “Retinal Topographic Stack” can be made. This will allow the acquisition of either macular or optic nerve head topography maps. It acquires a stack of 100 consecutive OCT C-scans over a fixed distance and an area of 15 by 15 degrees in 2 seconds. The operator is given the option to indicate whether the macula (1.125 mm), thickened retina (1.5 mm) or the optic nerve head (2.0 mm) will be scanned, but for each option the scan depth of the stack is adjustable from 1.0 up to 6.0 mm. The intended scan area can be changed by the operator by dragging the square to the desired position on the fundus.

The current software allows for post-acquisition analysis by providing the option to localise areas of interest by either encircling the area or marking a specific point (Figure 4). Whether the outline is made on the OCT or the confocal image, the corresponding area will be outlined in the fellow image. On OCT B-scans, a calliper function allows the distance between two vertical or horizontal lines to be measured (Figure 5). Retinal thickness can also be measured in the topographic stack. This is displayed as a two-dimensional square showing either contour lines indicating various height levels, or as a grid giving 16 averaged thickness measurements in areas of each X by X mm (Figure 6). The retinal thickness is defined as the thickness between the vitreoretinal interface and the RPE (i.e. the

Figure 4: Outline and indication function in OCT C-scan. The green encircled area represents the outline function, indicating an area of retinal vessels, the green cross in foveal pit represents the single spot indication function. (centered red line is indicator for B-scan position)

Figure 5: Vertical (top) and horizontal (bottom) caliper function for OCT B-scan

second hyper-reflective layer at the outer retina). The confocal image provides a good quality reference image for aligning consecutive topographic stacks and monitoring change over time.

EVALUATION OF THE NORMAL FUNDUS

To the uninitiated, individual OCT C-scans can be difficult to interpret. But, as with all diagnostic imaging techniques, it is a matter of pattern recognition and repetition. OCT/SLO imaging is a dynamic investigation as is, e.g. fluoroscopy in radiology. When acquiring OCT C-scans, abnormal or unusual areas are

Figure 6: Topographic map in a healthy subject

often seen. The operator should then proceed to capture longitudinal OCT B-scans in those areas of interest. Based on his familiarity with OCT scans, the operator can then differentiate healthy and normal looking OCT C-scans from diseased and abnormal looking scans.6 Therefore, the normal anatomy as

visualised by the OCT C-scans must first be discussed, before disease related OCT images are introduced.

The images shown in Figure 7 are part of a stack of OCT C-scans from the right macular area of a healthy subject. At different depths, the various retinal layers become visible, beginning at the vitreoretinal interface and ending in the choroid. Due to the curvature of the eye, the edge of the scan images the various retinal layers arranged inside out, with the inner and outer retinal layers visible like “onion rings”. At the level of the ganglion and inner plexiform layers, the retinal vessels are easily recognised. They initially appear white in the more superficial layer (when selected by coherence in the OCT C-scan), and appear dark (shadow) when OCT C-scans are taken at deeper layers of the retina. As the foveal pit is the deepest part of the retina, the optically translucent vitreous remains

Figure 8: OCT B-scan of a healthy subject. ph = posterior hyaloid; ilm = inner limiting membrane; rnfl = retinal nerve fibre layer; gcl = ganglion cell layer; I/OPL = inner/outer plexiform layer; I/ONL = inner/outer nuclear layer; IS/OS = inner/outer segments photoreceptor; RPE = retinal pigment epithelium; * = choroidal vessels

visible as a small central dark circle, while in the surrounding macular region the highly reflective OCT signal of the retinal nerve fibre layer (RNFL) and less reflective OCT signals of the inner and outer plexiform and nuclear layers become apparent. The second highly reflective OCT signal represents the intersection of the inner/outer segments of the photoreceptors and the hyper-reflective layer following this is thought to be the interface with the retinal pigment epithelium (RPE) and choriocapillaris. The OCT B-scan of the same person taken through the central fovea has a more familiar appearance (Figure 8).

As mentioned earlier, the confocal part of the OCT C-scan is able to demonstrate possible movement in both the coronal and longitudinal cuts. The most obvious movement artifact is blinking, which is easily recognised in any OCT device (Figure 9). Rapid (saccadic) eye movements during coronal scanning are seen as an unusual “tear or rip” in the confocal image (Figure 10). Axial movements caused by breathing or change in muscular tension can be seen as distortions in the coronal OCT scan, with complete loss of the circular outline of the globe (Figure 11). In longitudinal scans the confocal image functions as a motion-mode image, as successive rows of pixels represent repeated scans along the same line in space at different times. Any lateral movement during scan acquisition causes a displacement in these lines that corresponds to a displacement in the OCT B-scan. These displacements can be marked on the confocal image to reveal the corresponding movement in the B-scan (Figure 12). Axial movement in the OCT B-scan can be seen as

Figure 9: Blink artifact in OCT C-scan (top) and OCT B-scan (bottom) (centred red line in OCT C-scan indicated position of OCT B-scan)

Figure 10: Lateral movement artifact in OCT C-scan. The red box indicates area of artifact, which is visible as a small “tear” in both the confocal and the OCT part (centred red line is indicator for B-scan position)

Figure 11: Axial movement artifact in OCT C-scan caused by breathing or change in facial muscle tension showing a peculiar wavy pattern

Figure 12: Yellow outline of movement artifact in OCT B-scan

Figure 13: Axial movement artifact in OCT B-scan caused by breathing or change in facial muscle tension

partial or complete elongations in certain layers of the retina, which can be either subtle and deceiving or very evident (Figure 13). Awareness of these movements artifacts in the OCT C-scan images enables full use of all captured images to reveal objective information about changes within the retina.

EVALUATION OF UVEITIC PATIENTS: CYSTOID MACULAR OEDEMA

The appearance of cystoid macular oedema (CME) in longitudinal OCT scans is well known (Figure 14A). On OCT C-scans, this can also be easily recognised. The OCT C-scan frames in Figure 14 show two successive frames in a patient with evident CME.

Since the foveal contour has changed from a depression in the macular surface to a fluid filled mount towering above the retinal surface, its crest is now seen before any retinal layers are visualised in the C-scan. The area affected by cystoid changes is clearly visualised and even the separate cystoid spaces can be distinguished. The OCT C-scans also allow for recognition of more subtle (cystoid) changes, as shown in this case of Birdshot retinopathy.16 The retinal layers

Figure 14: OCT/SLO images of a patient with Cystoid Macular Oedema (CME) (A) OCT B-scan shows that the CME is located in the inner and outer nuclear layers and shows a small serous detachment (arrow). (B,C) Consecutive OCT C-scans showing the extent of the CME (arrowheads)

Figure 15: OCT C-scan of a patient with Birdshot retinopathy. The OCT C-scan taken in preretinal, shows small areas of retinal thickening and small cystic changes in temporal area (arrows)

at the border of the C-scan frame (Figure 15) are irregularly shaped, indicative of subtle retinal thickening.

OCT/SLO TOPOGRAPHY: VALUE IN FOLLOW-UP

The only objective way to evaluate retinal thickness is by measuring it in the OCT scans. Based on a stack of 100 consecutive fast OCT C-scans, the OCT/SLO provides a 3D rendering of an area of interest for example in the macula (Figure 16). Software-inferred OCT B-scans are generated by scrolling through this

Figure 16: Topographic 3D rendering in patient with CME

3D block. Although these OCT B-scan show a less detailed image than the regular OCT B-scans, the outer borders of the retina and fovea are clearly visible. They are particularly useful in identifying possible misalignment in a stack. A stack is topped with the confocal image, and overlain with a false-colour scheme indicating the thickness. Figure 16 shows the topographic images of the patient with CME in Figure 14. Figure 17 shows images taken at the next visit after treatment. The regular OCT C- and B-scan images (Figure 17A) still show evident CME, and it is difficult to assess the thickness. The topographic images (Figure 17C) however, show an evident decrease in retinal

Figure 17: Patient with Cystoid Macular Oedema (CME) presented in Figure 14 after treatment (A) OCT B-scan shows that the CME, located in the inner and outer nuclear layers, has decreased and that the small serous detachment (arrow) has subsided. (B,C) Consecutive OCT C-scans showing the extent of the CME (arrowheads)

Figure 18: Monitoring of patients with topographic change analysis (Top) Topographic images of the patient with cystoid macular edema presented in Figures 14 and 17. (Centre) Alignment of the topographic maps on the confocal images. (Bottom) Subtraction of the two topographic maps result in a thickness change mapping (left) and 3D rendering hereof (right)

thickness, clearly visible in the colour profile. Topographic comparison between two different visits is also possible as shown in Figure 18. Such comparisons are generated by first selecting both stacks, and aligning them on a larger confocal image, using the intrinsic confocal image of the stack. This allows for comparison of the exact same areas. After both images are satisfactory aligned, a comparison is given as a subtraction image. On one side this shows a 2D image using a different false-color scheme to indicate the loss or gain in thickness, and on the other side a 3D contour plot showing the change in retinal thickness is generated.

OVERLAY WITH CONVENTIONAL ANGIOGRAPHY

The high quality fundus image that is provided by the confocal channel can be used as a reliable reference image to make software-inferred overlay images from more conventional diagnostic techniques as fluorescein angiography (FA) onto the OCT part of the C-scans.