7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
1/36
Combinations of techniques in imaging the retina with high resolution
Adrian Gh. Podoleanu a,, Richard B. Rosen b
aApplied Optics Group, University of Kent, Canterbury, UKb New York Eye and Ear Infirmary, NY, USA
a b s t r a c t
Developments in optical coherence tomography (OCT) have expanded its clinical applications for high-resolution imaging of the retina, as a standalonediagnostic and in combination with other optical imaging modalities. This review presents currently explored combinations of OCT technology with a
variety of complementary imaging modalities along with augmentational technologies such as adaptive optics (AO) and tracking. Some emphasis is on
the combination of OCT technology with scanning laser ophthalmoscopy (SLO) as well as on using OCT to produce an SLO-like image. Different OCT
modalities such as time domain and spectral domain are discussed in terms of their performance and suitability for imaging the retina. Each modality
admits several implementations, such as flying spot or using an area or line illumination. Flying spot has taken two principle forms, en-face and
longitudinal OCT. The review presents the advantages and disadvantages of different possible combinations of OCT and SLO with AO, evaluating criteria
in choosing the best OCT method to fit a specific combination of techniques. Some of these combinations of techniques evolved from bench systems into
the clinic, their merit can be judged on images showing different pathologies of the retina. Other potential combinations of techniques are still in their
infancy, in which case the discussion will be limited to their technical principles. The potential of any combined implementation to provide clinical
relevant data is described by three parameters, which take into account the number of voxels acquired in unit time, the minimum time required to
produce or infer an en-face OCT image (or an SLO-like image) and the number of different types of information provided. The current clinically used
technologies as well as those under research are comparatively evaluated based on these three parameters. As the technology has matured over the
years, their evolution is discussed as well with their potential for further improvements.
& 2008 Elsevier Ltd. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
2. High-resolution imaging technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
2.1. Optical coherence tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
2.2. Scanning laser ophthalmoscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
2.3. Adaptive optics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
2.4. Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
3. Different scanning procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
3.1. The one letter terminology of scanning, A, B, C, T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
3.1.1. Longitudinal OCT or A-scan-based B-scan. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
3.1.2. T-scan-based B-scan. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
3.1.3. C-scan images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
4. Different OCT imaging methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
4.1. Time domain optical coherence tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
4.2. Spectral domain optical coherence tomography. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
ARTICLE IN PRESS
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/prer
Progress in Retinal and Eye Research
1350-9462/$ - see front matter& 2008 Elsevier Ltd. All rights reserved.doi:10.1016/j.preteyeres.2008.03.002
Abbreviations: 3D, three-dimensional; AMD, age-related macular degeneration; AO, adaptive optics; APD, avalanche photodiode; cSLO, confocal scanning laser
ophthalmoscopy; dB, decibel; D-OCA, Doppler-optical coherence angiography; ERG, electroretinography; FA, fluorescein angiography; FD, Fourier domain; FF, full field; FFT,
fast Fourier transformation; I, imaging content units, or number of channels simultaneously working in a combined configuration; ICG, indocyanine-green; IS, inner
segments; LF, line field; L-SLO, line-scanning laser ophthalmoscopy; mfERG, multifocal electroretinography; Mv/s, megavoxels per second; NA, numerical aperture; n.a.,
non-applicable; OCA, optical coherence angiography; OCT/SLO, combined OCT and scanning laser ophthalmoscopy; OCT, optical coherence tomography; ODT, optical
Doppler tomography; OPD, optical path difference; OS, outer segment; PED, pigment epithelium detachment; RNFL, retinal nerve fiber layer; RPE, retinal pigment
epithelium; S-OCA, scattering-optical coherence angiography; SD, spectral domain; SLD, superluminescent diode; SLO, scanning laser ophthalmoscopy; SS, swept source;
TD, time domain; mm, micron; ms, millisecond; X, Y, Z, rectangular axes, with X- and Y-oriented lateral to the retina and Z along the depth. Corresponding author. Tel.: +441227823272; fax: +441227827558.
E-mail addresses: [email protected] (A.Gh. Podoleanu), [email protected] (R.B. Rosen).
Progress in Retinal and Eye Research 27 (2008) 464 499
http://www.sciencedirect.com/science/journal/jprrhttp://www.elsevier.com/locate/prerhttp://localhost/var/www/apps/conversion/tmp/scratch_4/dx.doi.org/10.1016/j.preteyeres.2008.03.002mailto:[email protected]:[email protected]:[email protected]:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_4/dx.doi.org/10.1016/j.preteyeres.2008.03.002http://www.elsevier.com/locate/prerhttp://www.sciencedirect.com/science/journal/jprr7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
2/36
4.2.1. Fourier domain optical coherence tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
4.2.2. Swept source optical coherence tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
4.3. Full field or en-face non-scanning systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
4.4. Line-field-SD-OCT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
4.5. Multiplexing in OCT. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
4.5.1. Multiplexing in A-scan-based OCT imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
4.5.2. Multiplexing in T-scan-based OCT imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
4.6. Comparative assessment of the OCT methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
5. Depth of focus range and dynamic focus in OCT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4736. Combining OCT with SLO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
6.1. Simultaneous OCT/SLO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
6.1.1. Recognizable patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
6.2. No splitter OCT/SLO configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
6.3. Sequential generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
6.3.1. Sequential OCT C-scan/SLO C-scan, sequential OCT B-scan/SLO B-scan. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
6.3.2. OCT/line-SLO (sequential B-scan OCT with C-scan SLO). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
6.4. Quasi-simultaneous operation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
6.5. Generation of an SLO-like image using OCT. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480
6.5.1. Real-time generation by using a low coherent source with sufficiently long coherence length. . . . . . . . . . . . . . . . . . . . . . . . 480
6.5.2. Generation of an SLO-like image from OCT stacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
7. Combination of high-resolution modalities with fluorescence imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
7.1. SLO/fluorescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
7.2. OCT/fluorescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
7.3. OCT/SLO/fluorescence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
7.4. No-dye fluorescence-based OCT imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 486
8. Combinations of high-resolution imaging procedures with AO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
8.1. SLO+AO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487
8.1.1. SLO/fluorescence+AO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
8.2. OCT+AO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
8.2.1. Trade-off between depth resolution in OCT and level of correction using AO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
8.2.2. Incompatibility between A-scan-based OCT and AO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489
8.2.3. Small size imaging using TD-en-face OCT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489
8.2.4. Full field TD-OCT+AO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 490
8.2.5. Longitudinal TD-OCT+AO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 490
8.2.6. FD-OCT+AO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 490
8.2.7. FD-OCT+AO equipped with two mirrors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 490
8.3. OCT+SLO+AO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 490
8.3.1. Simultaneous OCT/SLO+AO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 490
8.3.2. Sequential OCT/SLO+AO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
9. Combination of high-resolution imaging technologies with tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4929.1. SLO+AO+tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492
9.2. TD-OCT+lateral tracking. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
9.3. TD-OCT+AO+axial tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
10. Combination of high-resolution imaging technologies with other techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
10.1. Combinations with physiology methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
10.1.1. Multifocal ERG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
10.1.2. Combinations with optophysiology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
10.2. Combinations with polarization information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
10.3. Combinations with flow information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
10.4. Combinations with spectroscopy imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
10.5. Polarization/flow/optophysiology/spectroscopy imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
11. Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
11.1. Potential of existing technologies and trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
11.2. Synergies provided by the combination of techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
1. Introduction
Today, ocular imaging technology has reached high heights of
sophistication, building on the tremendous progress in the
last 5 years. However, none of the current imaging methods
available fulfill all the ideal requirements of the ophthalmologist
faced with the need for rapid and accurate diagnosis. This
has led to exploration of combinations of imaging and assistive
techniques by groups attempting to solve these deficiencies.
The goals driving the combination of different imaging
technologies are diverse, including the need for precise targetingand real-time focusing of the en-face optical coherence tomo-
graphy (OCT) (which lead to the addition of a scanning laser
ophthalmoscopy (SLO) channel to the OCT channel), the need for
correlation of retinal blood flow with changes in morphology
(such as in the combination of OCT with fluorescence imaging) or
the need for enhancing the imaging performance (such as the
addition of adaptive optics (AO) and tracking to SLO or OCT or
combined OCT/SLO).
Expansion of a familiar perspective found in one type of
instrument may stimulate interest in combining it with an
additional modality, which shares the same viewpoint. For
example, en-face imaging (C-scan) has the advantage thatophthalmologists are more familiar with the interpretation of
ARTICLE IN PRESS
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499 465
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
3/36
transversal images since they are of similar orientation as those
found in ophthalmoscopes, fundus cameras and SLOs. This was an
important factor that stimulated research into combining the OCT
with SLO technology.
This review will focus primarily on combinations of techniques
where the core technology is OCT. Initially, OCT technology
advanced towards enhancing the acquisition rate along a line in
depth in the image. Nowadays, OCT imaging is found into an ever-growing collection of combinations, which pair OCT with
techniques such as SLO, flow imaging, polarization, multifocal
electroretinography (mfERG), optophysiology, oximetry, micro-
perimetry, etc.
A major goal of current research is scanning a target volume of
the retina as fast as possible. Significant progress has been
achieved in terms of line-scanning rate, which increased from tens
of Hz in the first OCT implementation (Huang et al., 1991) to tens
of kHz (Nassif et al., 2004) to using the channelled spectrum or
the Fourier domain OCT (FDOCT), and to hundreds of kHz (Huber
et al., 2007) using the swept source OCT (SS-OCT) method (see
Table 1). However, a fast line-scanning rate may not be sufficient
to guarantee superiority in respect of all performances required by
an accurate diagnosis.For instance, imaging methods that are recognized as very fast
in the modern OCT imaging today, such as spectral domain OCT
(SD-OCT), operate under fixed focus, limiting the accuracy of
three-dimensional (3D) acquisition. If the same sensitivity is
required within the whole 3D volume, then repetition of
acquisitions under several different positions of the focus could
lead to longer acquisition times for high-density, high-resolution
volumes than the time required by slower line-scanning methods,
which allow focus change. Therefore, specific imaging require-
ments may take precedence over the raw line-scanning rate in
order to respond to the need of good sensitivity and sufficient
sampling data.
In the discussion, which follows, different imaging modalities
and specific implementations are compared in their performance
taking into consideration three parameters:
1. Mv/s: The number of pixels along three rectangular directions
(two lateral and one in depth) acquired in the unit of time.
Sometimes, in order to shorten the overall scanning time
required to capture a given retina volume, a coarse sampling
size is chosen for one of the axes, leading to enlargement of the
pixel size along that particular direction, trade-off best
described by the parameter Mv/s.
2. Tenface: The time to produce a two-dimensional (2D) OCT
image with the SLO orientation.
3. Imaging content units (I): The number of different types of
information provided in a system by a specific configuration.
As combination of techniques compound different types ofinformation (OCT, SLO, fluorescence), the performance Mv/s is
multiplied by I to obtain an overall performance of a given
combination of techniques, as IMv. For instance for a combined
system incorporating OCT, SLO and fluorescence channels
which can deliver images simultaneously, I 3.
The combination of techniques is evolving in two principal
directions: combination of channels providing multiple informa-
tion and combination of imaging techniques with assisting
technologies, such as AO and tracking. Both lines of development
will be presented here.
The quest for faster and more complete acquisition of
information from the eye demands evaluation of several trade-
offs in the performance of the technologies combined. Suchdifferent demands and trade-offs will be discussed, presenting the
problems raised by the hardware combination as well as the
challenges in developing synergistic interpretations of composite
images collected from several imaging channels.
2. High-resolution imaging technologies
2.1. Optical coherence tomography
OCT is a non-invasive high-resolution imaging modality, which
employs non-ionizing optical radiation. OCT derives from low-
coherence interferometry. This is an absolute measurement
technique that was developed for high-resolution ranging and
characterization of optoelectronic components (Al-Chalabi et al.,
1983, Youngquist et al., 1987). The first application of the low-
coherence interferometry in the biomedical optics field was for
the measurement of the eye length (Fercher et al., 1988). Adding
lateral scanning to a low-coherence interferometer, allows depth-
resolved acquisition of 3D information from the volume of
biologic material (Huang et al, 1991). The concept was initially
employed in heterodyne scanning microscopy (Sawatari, 1973).
OCT has the potential of achieving high depth resolution, which is
determined by the coherence length of the source. This is the
length over which a process or a wave maintains strict phase
relations; an ideal laser source for instance, emits light with more
than a few km coherence length while the coherence length of
light emitted by a tungsten lamp could be as short as 1 mm. More
intense optical sources, suitable for use in scanning the eye are
now available with coherence lengths below 1 mm (Drexler, 2004).
Using sources with extremely short coherence length, submicron
depth resolution is achievable even when the microscope
objective is far away from the investigated target, feature not
achievable with confocal microscopy. This is one of the most
important advantage of OCT, which explains the high level of
interest for OCT in ophthalmology. OCT delivers fast, non-contact
images of the cornea, lens and the retina with depth resolutionsbetter than 3mm (Drexler, 2004).
2.2. Scanning laser ophthalmoscopy
Confocal imaging was the first high-resolution imaging
technology applied to the eye (Webb, 1990). The influence of
scattered light from outside the focus point within the target is
suppressed by a pinhole in front of the photodetector and
conjugate to the focal plane (Elsner et al., 1996). 3D imaging
(Masters, 1998) is performed by acquiring en-face images (C-
scans) at different positions of the focusing element, each position
corresponding to a different depth. Increasing the beam diameter
of the beam sent to a lens leads to a better confinement of thelight in the focus of the lens and therefore to better transversal
and depth resolution. The key figure in following these changes is
the numerical aperture (NA), a quantity proportional with the
beam diameter and inversely proportional with the focal length of
the lens used in imaging. The transversal resolution varies inverse
proportional to NA while the depth resolution varies inverse
proportional to the square of the NA. Therefore, to improve the
resolutions, it will be desirable to work with a large eye opening.
However, in practice, increasing the eye pupil only leads to
cumulative addition of aberrations. Therefore, with or without the
pupil dilated, an SLO will provide approximately 15mm transversal
resolution and larger than 300mm axial resolution (Bartsch and
Freeman,1994; Woon et al., 1992). Similarly, a relatively low NA of
the anterior chamber limits the achievable resolution in imagingthe eye lens.
ARTICLE IN PRESS
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499466
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
4/36
Table 1
Evolution of the OCT technology in imaging the retina in terms of number of Megavoxel/s and Tenface
Method employed Line rate (kHz) Frame rate (Hz) Pixels in the B-scan
(Nx, Nz)
Pixels in
the
C-scan
(Nx, Ny)
Tenface (s) (time to produce an
OCT C-scan image)
M
f
1 TD-OCT longitudinal OCT, first report on OCT,
in-vitro (Huang et al., 1991)
0.0008 0.0053a 150118b n.a. Not contemplated 0
2 TD-OCT longitudinal OCT, in-vivo (Swanson
et al., 1993)
0.042 0.42 100285c n.a. 6095d 0
3 TD-OCT longitudinal OCT with fast axial
scanning (Rollins et al., 1998)e8 16 (32 possible) 250250 n.a. 16 (8)c 1
4 En-face TD-OCT (Podoleanu et al, 1998b) 0.6 1.18 196100 196196 0.85 0
5 En-face TD-OCT (van Velthoven et al., 2006) 1 2 (8) 512250 512512 0.5 0
6 En-face TD-OCT using a resonant galvo-scanner
(Hitzenberger et al., 2003)
4 53.3 Not contemplated,
but possible
256128 0.019 1
7 FD-OCT (Wojtkowski et al., 2004) 16 6.7 (or 31) 30001024
(or 5121024)
n.a. 38.2c(or 8.25) 2
8 FD-OCT (Nassif et al., 2004) 29 29 (real-time
display 10)
1000320 n.a. 8.83c 9
9 FD-OCT Gotzinger et al., 2005) 20 20 1000292f n.a. 12.8d 5
s
10 FD-OCT/Line-field SLO (Iftimia et al., 2006) Not specified 15 1024512 n.a. 17c 7
1
11 Line-field FD-OCT (Nakamura et al., 2007) 51.5 (single
frame: 823.2)
201 128g108h n.a. 1.27 2
12 SS-OCT at 850nm (Lim et al., 2006) 43.2 84 512140i n.a. 3j 6
13 SS-OCT at 1050nm (Huber et al., 2007) 236 461 512512 n.a. 0.56c 1
14 Fastest SS-OCT (Moon and Kim, 2006) 5000 105 5069 n.a. 0.0026d 3
a Although potential to 5 Hz was also mentioned.b Evaluated using the values quoted in the paper of 2 mm depth range and 17mm depth resolution in air.c Evaluated using the values quoted in the paper of 3 mm depth range and 10.5mm depth resolution in air.d
Not contemplated and evaluated for 256 frames using the frame rate for B-scan imaging.e Although images from the anterior chamber of a murine eye are presented only and not from the retina, the fast scanning delay line method presents sufficient sensitivit
comparison of technologies.f Evaluated from 1.75mm depth range with 6mm resolution quoted.g Dividing the line of 2.1mm with the indicated value of transversal resolution, of 16.4mm.h Evaluated from 0.8 mm depth range and 7.4mm resolution quoted.i Evaluated from 1.4mm depth range and 10mm resolution in tissue.j Evaluated for 256 B-scan frames.
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
5/36
2.3. Adaptive optics
AO uses a wavefront sensor, which instructs a corrector,
usually a deformable mirror, to alter the wavefront in order to
compensate for the aberrations in the eye and in the instrument
(Liang et al., 1997; Dreher et al., 1989; Roorda et al., 2002).
However, if aberrations are compensated using AO, the transversal
resolution of a confocal-SLO (cSLO) through a 6 mm pupil could beimproved to less then 3 mm (Zhang and Roorda, 2006) and the
axial resolution could be reduced to 40mm (Venkateswaran et al.,
2004). For an 8 mm pupil size, resolution is even more improved
(Miller et al., 2003). These levels of resolution pose compatibility
considerations when added to other imaging modalities, such as
OCT. After AO correction is applied, a fixed focus scanning
approach may adversely affect the signal strength of points not
far from the focus, due to the shrinkage of the confocal profile
produced by the correction of aberrations.
2.4. Tracking
Tracking is an assisting technology to high-resolution imaging,which has seen impressive evolution in the last 5 years. Tracking
is essential when information to be extracted needs a stationary
target. This is often the case in high-resolution or small size
imaging and especially when the signal returned from the retina
is weak. With weak signals, information from a single frame
may be insufficient, necessitating some form of aligned multiple
image collection and averaging. Tracking also enables psychophy-
sical and neurophysiological techniques, which require accurate
eye motion compensation, such as micro-perimetry and laser
surgery. Tracking is a complex technology which raises several
issues including: (i) selection of the reference region used for
tracking, either retina or cornea or both; (ii) wavelength of
operation; (iii) integration of the tracking system into the imaging
system, requiring modification of the interface optics tohandle both tracking and imaging beams; and (iv) adjustment of
power safety levels to compensate for at least two beams
launched into the eye.
3. Different scanning procedures
To obtain 3D information about the retina, any imaging system
is equipped with three scanning means, one to scan the object in
depth and two others to scan the object transversally. Depending
on the order these scanners are operated and on the scanning
direction associated with the line displayed in the raster of the
final image delivered, different possibilities exist. One-dimen-sional (1D) and 2D scans are known. 1D scans are labeled as: A-
and T-scans, while 2D scans are labeled as B- and C-scans and this
terminology will be explained below. A- and T-scans are 1D
reflectivity profiles while B and C are 2D reflectivity maps or
images. In terms of the strength, the brightness in the SLO image
is proportional to the reflectivity while in the OCT with the square
root of the reflectivity. While in the SLO, the display is generally
linear, in OCT, especially for B-scans, the display often represents
the logarithm of the reflectivity.
OCT systems, using CCD cameras or arrays of sensors or arrays
of emitters eliminate the need of scanning the beam. However, the
terminology below applies in such cases as well, where the ray
scanning has been replaced by different forms of electronic
scanning. The scanning terminology is illustrated in Fig. 1 and theutilization of the three scanners described in Fig. 2.
3.1. The one letter terminology of scanning, A, B, C, T
A-scan: represents a reflectivity profile in depth. This scanning
technology is used clinically for determining the eye length. In
principle, an A-scan can be provided by cSLO (Bartsch and
Freeman, 1994) as well. The depth scanning requires the axial
movement of a lens to alter the focus, as the lens is heavy, the
scanning cannot be fast. Electrically adjustable lenses may provide
a solution for fast axial scanning in cSLO, however, there is no
motivation for development in that direction in view of the much
better resolution obtained using OCT.
T-scans: represent the reflectivity (SLO) or square root of
reflectivity (OCT) obtained by scanning the beam transversally
across the target.
B-scan: represents a cross-section image, a (lateraldepth)
map. This could be obtained by grouping T-scans together fromdifferent depth values or A-scans together for different lateral
ARTICLE IN PRESS
B-scan
C-scan
A-scan
Y-Z
X
T-scan
Fig. 1. Relative orientation of the axial scan (A-scan), transverse scan (T-scan),
longitudinal slice (B-scan) and en-face or transverse slice (C-scan).
(iii)
C-SCAN IMAGES AND
3D SCANNING BASED
ON C-SCANS
(CONVENTIONAL
cSLO OPERATION
(ii)
B-SCAN IMAGE
GENERATED FROM T-SCANS
(METHOD COMPATIBLE WITH
C-SCANNING)
(i)
B-SCAN IMAGE GENERATED
FROM A-SCANS
(CONVENTIONAL)
LONGITUDINAL OCT
SCANNING)
-Z
XY
-Z
SLOW
SLOW
SLOWFAST
X
-Z
XSLOWEST
FAST
FAST
A-scans
T-scans
(iv)
3D SCANNING BASED ON
B-SCAN SLICES AT
DIFFERENT POSITIONS Y
CURRENTLY USED BYSD-OCT SYSTEMSZ
SLOWFAST
Y
X
SLOWEST
Fig. 2. Different modes of operation of the three scanners in a 3D imaging system.
Lateral scanning along the X- and Y-axes are implemented using an XY or 2D
transverse scanner in both cSLO and OCT systems. The scanning in depth, along the
axis Zdiffers, implemented using focus change in cSLO and OPD change in OCT. (i)
B-scan image generated from A-scans (conventional longitudinal OCT scanning).
(ii) B-scan image generated from T-scans (method compatible with C-scanning).
(iii) C-scan images and 3D scanning based on C-scans (conventional cSLO
operation. (iv) 3D scanning based on B-scan slices at different positions y
currently used by SD-OCT systems.
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499468
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
6/36
positions, both in use in the OCT practice. Historically, the first
OCT image was a B-scan image of the retina (Huang et al., 1991)
made from A-scans, using flying spot longitudinal OCT technology
(see below).
Although possible in the cSLO, B-scans are not used in the
imaging of patients. cSLO can provide only T-scan-based cSLO
B-scan images. (An A-scan-based B-scan would require fast focus
change; see the comment above in connection to the A-scans incSLO). However, such a B-scan image is inferred in the practice of
glaucoma imaging in a post-acquisition process after collection of
C-scan images (Mikelberg et al., 1995).
B-scan images, analogous to ultrasound B-scan are generated
by collecting many A-scans (Fig. 1) for different and adjacent
transverse positions, as shown in Fig. 2(i). The lines in the raster
generated correspond to A-scans, i.e. the lines are oriented along
the depth coordinate. The transverse scanner (operating along X
or Y, or along the polar angle y in polar coordinates in Fig. 1 right,
with X shown in Fig. 2(i)) advances at a slower pace to build a B-
scan image. The majority of OCT reports in literature refer to this
mode of operation.
3.1.1. Longitudinal OCT or A-scan-based B-scanDevelopment of the longitudinal OCT based on A-scans was
facilitated by a technical advantage: in time domain OCT (TD-
OCT), when moving the mirror in the reference path of the
interferometer, not only is the depth scanned, but a carrier is also
generated (Huang et al., 1991; Swanson et al., 1993). The
reflectivity information is superposed on a carrier signal, having
a frequency equal to the Doppler shift produced by the long-
itudinal scanner itself (moving along the axis of the system, Z, to
explore the retina in depth). In longitudinal OCT, the axial scanner
is the fastest and its movement is synchronous with displaying
the pixels along the line in the raster, while the lateral scanning
determines the frame rate. Longitudinal OCT is performed as
explained here in TD and is also provided by SD-OCT methods (see
Section 4.2.)
3.1.2. T-scan-based B-scan
In this case, the transverse scanners (or scanner) determine(s)
the fast lines in the image (Podoleanu et al., 1996, 1998a,b). These
image lines represent T-scans (Fig.1). A T-scan can be produced by
controlling either the transverse scanner to scan along theX-coordinate, or the Y-scanner to scan along the Y-coordinate
with the other two scanners fixed, or controlling both transverse
scanners, along the polar angle y, with the axial scanner fixed. The
example in Fig. 2(ii) illustrates the generation of a T-scan-based
B-scan, where the X-scanner produces the T-scans and the axial
scanner advances slower in depth, along the Z-coordinate. This
procedure has a net advantage in comparison with the A-scan-
based B-scan procedure as it allows production of OCT transverse(or 2D en-face) images for a fixed reference path, images called
C-scans. In this way, the system can be easily switched from B to
C-scan, procedure incompatible with A-scan-based OCT imaging
mentioned above.
3.1.3. C-scan images
A C-scan represents a raster image, with the same orientation
as a TV image or image provided by microscopy, a (lateral lat-
eral) scanned map. Historically, the C-scan was the native
orientation for fundus cameras, SLOs and cSLOs. C-scans are
provided by the flying spot en-face OCT and the full-field (FF) OCT.
They can also be inferred post-acquisition in longitudinal OCT,
either TD or SD. C-scans are made from many T-scans along either
ofX, Y, r or y coordinates, repeated for different values of the othertransverse coordinate, Y, X, y or r, respectively in the transverse
plane (with the most used case, oriented along the horizontal axis,X). The repetition of T-scans along the other transverse coordinate
is performed at a slower rate than that of the T-scans (Fig. 2(iii)),
which determines the frame rate. In this way, a complete raster is
generated. For 3D imaging, different transversal slices can be
collected at different depths Z, either by advancing the optical
path difference in the OCT in steps after each complete transverse
(XY) or (r,y) scan, or continuously at a much slower speed than theframe rate. The depth scanning is the slowest in this case. In cSLO,
the focus is changed to select a C-can from a different depth
position and this is the typical procedure for 3D cSLO imaging
(Masters, 1998).
It is more difficult to generate en-face OCT images than
longitudinal OCT images as the reference mirror is fixed and no
carrier is produced. Therefore, in order to generate T-scans and
T-scan-based OCT images, a phase modulator is needed in order to
create a carrier for the image bandwidth (Hitzenberger et al.,
2003). This complicates the design and introduces dispersion.
Research has shown that the X or Y-scanning device itself
introduces a path modulation (Podoleanu et al., 1996, 1998a, b),
which plays a similar role to the path modulation created by the
longitudinal scanner employed to produce A-scans or A-scan-based B-scans.
4. Different OCT imaging methods
There are two main OCT methods, TD-OCT and SD-OCT. SD-
OCT can be implemented in two formats, FD-OCT and SS-OCT.
Their utility for retinal imaging has been presented in several
recent review articles in this journal (Costa et al., 2006; Drexler
and Fujimoto, 2008; van Velthoven et al., 2007). We will shortly
review them, to compare their performance and discuss how they
can be best combined with other retinal imaging modalities. Each
method has its own merits and deficits.
4.1. Time domain optical coherence tomography
An A-scan is produced by varying the optical path difference
(OPD) in the interferometer to output a square root of reflectivity
profile in depth. En-face flying spot OCT belongs to the same
category. A T-scan is produced by transversally scanning the beam
over the target maintaining the reference mirror fixed to output a
square root of reflectivity profile versus angle or lateral position.
In both cases, the envelope of the interferometric temporal signal
is processed in time.
4.2. Spectral domain optical coherence tomography
In the last 5 years, considerable research has been devoted by
different groups developing OCT for tissue imaging into the
spectral OCT method. SD-OCT is attractive because it eliminates
the need for depth scanning in TD-OCT, performed usually by
mechanical means. Recent studies (Choma et al., 2003) have
shown that SD-OCT can provide a sensitivity, which is more than
10 times higher than that of TD-OCT.
FD-OCT and SS-OCT output A-scans, therefore they do not
allow real-time C-scan imaging. En-face (C-scan) sections can only
be obtained in FD-OCT and SS-OCT by sectioning the 3D volume
generated from a series of B-scan images taken at different
transverse coordinates, i.e. as a post-acquisition process only.
Therefore, essential in comparing the different OCT technologies is
the time required to complete a volume acquisition of the retina,which is then sectioned to create C-scan slices
ARTICLE IN PRESS
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499 469
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
7/36
4.2.1. Fourier domain optical coherence tomography
This refers to Fourier transformation of the optical spectrum of
a low-coherence interferometer (Hausler and Lindner, 1998). This
method is an extension of the work on white light interferometry
with initial applications in absolute ranging and sensing (Smith
and Dobson, 1981). The operation of FD-OCT is based on the
demodulation of the optical spectrum output of an interferometer.
The spectrum exhibits peaks and troughs (channeled spectrum)and the period of such a modulation is proportional to the OPD in
the interferometer (Jenkins and White 1957). If multi-layered
objects are imaged, such as retina, each layer imprints its own
modulation periodicity, depending on its depth. A linear CCD
camera can be used to transform the optical spectrum into an
electrical signal, which exhibits ripple of different frequencies. A
fast Fourier transform (FFT) of the spectrum of the CCD signal
translates the periodicity of the channeled spectrum into peaks of
different frequency, related to the path imbalance (Costa et al.,
2006). Such a profile is essentially the A-scan profile of the square
root of reflectivity in depth. Due to its sensitivity advantage, the
FD-OCT became the method of choice in current OCT investigation
of the retina (at least in the past 3 years) with video-rate images
from the retina demonstrated (Nassif et al., 2004; Cense et al.,2004; Wojtkowski et al., 2004; Gotzinger et al., 2005; Jiao et al.,
2005, 2006). The majority of FD-OCT reports employ linear
cameras at 29 kHz (and faster rates are already available), which
represents a line scan rate at least twice faster than en-face
imaging using a resonant scanner at 16 kHz and more than 30
times faster than line scan rates in en-face OCT using galvan-
ometer scanners. This also reflects in large values of the voxel
number Mv/s $10 to a few tens (Wojtkowski et al., 2004). Using
line field (LF) FD-OCT, where a line is projected to the retina,
eliminating the mechanical transverse scanner, frame rates as
high as hundred of Hz have been achieved (Nakamura et al.,
2007). However, even with such fast line scan rates, because
C-scans are perpendicular to the main scanning direction, which
is axial, the time to provide a C-scan, Tenface, is several seconds
(Table 1), much larger than the time achievable with en-face OCT
(tens of ms to sub-seconds, Table 1).
4.2.2. Swept source optical coherence tomography
Recent progress in the fast tuneability of laser sources has revived
the interest in SS-OCT. The achievable signal-to-noise (S/N) ratio is
similar to that of FD-OCT, i.e. at least 10 times better than TD-OCT
(Choma et al., 2003). The time required to tune the wavelength
determines the time to produce an A-scan. Tuning speeds in excess
of 10 MHz makes the SS-OCT the fastest scanning OCT method
(Moon and Kim, 2006) to date. Lower rates, of a few hundred kHz
have been reported in imaging the retina in-vivo. These values are
close to one order of magnitude higher than those achievable using a
CCD camera implementing FD-OCT. This method leads to muchlarger values of Mv/s$100 ((Huber et al., 2007) achieved a
Mv/s 122 for 400 frames of 512512 pixels cross sections
collected in 0.87s) and to a reduced value for the Tenface, of sub-
second to second. This value is comparable to that reported in typical
en-face OCT, but still larger than the ultrafast Tenface$20ms
achievable with resonant scanners (Hitzenberger et al., 2003).
4.3. Full field or en-face non-scanning systems
Full-field or coherence radar (Dresel et al., 1992) operates
according to the scanning operation described in Fig. 2(iii). The XY
scanning is provided in the process of reading a 2D charge-
coupled device (CCD) photodetector array. The eye is flood
illuminated, i.e. all pixels in transversal section are simulta-neously lit, in contrast to the flying spot method where each
transverse pixel in transversal section is independently lit at a
certain time only. The information acquired along a line of pixels is
equivalent to a T-scan in Fig. 1. Telecentric optics are used to transfer
the object beam from the target to the camera. Practically, every-
thing happens as for each pixel in the transversal section, an object
beam ray can be identified in the object beam, which interferes with
one ray within the cluster of reference rays. C-scans and B-scans can
be produced with no need for lateral scanning. The line rate is that ofreading 2D CCD cameras, which at video rate means more than
10kHz. The frame rate is in the range of tens to hundreds of Hz and
the depth scanning is the slowest, the OPD change rate being slower
than the CCD-frame rate. One of the main advantages of the method
is that it can use incoherent spatial optical sources, such as tungsten
lamps. These represent low-cost versions of low coherent sources
and easily exhibit bandwidths larger than 300 nm. Using such large
band spatial incoherent sources, and a Linnik interference micro-
scope (Dubois et al, 2002), images with submicron lateral and depth
resolutions have been reported (Grieve et al., 2004) from ocular
tissue in-vitro. Images from bovine retina in-vitro have also been
obtained using a superluminescent diode (SLD) (Qu et al., 2004). The
same system with AO enhancement (see Section 8.2.4. below) was
used to produce C- and B-scans of a living eye (Miller et al., 2003).Tremendous progress in sensitivity allowed the imaging of the
anterior chamber of rat eyes (Grieve et al., 2005) with a very high
value of Mv/s 33328 (depending on the frame averaging mode)
and a Tenface 4ms.
As a disadvantage of the FF-OCT, the amplitude of the
interference signal is recovered using phase-stepping algorithms.
Phase shifts are introduced by exact path difference steps, which in
total add up to a wavelength, or by a continuous change of the OPD
and comparing the sequences obtained. This means that real-time
processing is not possible, however, this is not important for fast
acquisition systems where this leads to a mere short delay in the
display, equal to the number of frames used for phase shifting
multiplied by the frame acquisition time. Using a fast acquisition
camera (Miller et al., 2003; Grieve et al., 2005), this amounts to some
tens of ms. As another disadvantage, the detection of reflective
interfaces in a multilayer object using the coherence radar method is
limited by the dynamic range of the analog to digital (A/D) converter
of the combination CCD-frame grabber system. The interference
signal sits on a large constant value, which consumes a large part of
the dynamic range of the CCD cameras.
The limited dynamic range of 16 bit CCD cameras, limits the
smallest signal to 1/65,365 of the digital value, which makes the
FF-OCT method less sensitive than the flying spot imaging
method. In principle, the flying spot can provide signals for
variations in the interference of less than 1013 in a 1 Hz electric
bandwidth (Takada et al., 1991).
A faster processing method uses an array of photodetectors in a
smart chip (Ducros et al., 2002). A photodetector is employed for
each pixel in the en-face image, followed by a processingelectronics channel (demodulation, rectifier, amplifier, condition-
ing). One pixel consists of a small silicon photodiode coupled to a
complementary metal-oxide semiconductor electronic circuit. A
maximum size chip of 5858 smart pixels was reported, which
limits the numbers of pixels in the images to similar values.
Because there is no transverse scanning to alter the OPD, a phase
modulator is used. The reading is sequential, similar to the
reading of a CCD camera in a coherence radar system but
the output is a fully demodulated OCT signal. The amplitude of
the signal provided by each channel is proportional to the
envelope of the OCT interference signal. The smart chip can also
operate in the C- and B-scan regimes. The technology still evolves;
for the moment there is no report on using the smart sensor on
ocular tissue. Medium values of Mv/s 2.5 have been achieved,however, with a record Tenface 1.33ms.
ARTICLE IN PRESS
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499470
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
8/36
4.4. Line-field-SD-OCT
This is a combination of the FF and SD-OCT, where instead of a
2D image, a lD transversal image is collected using principles of
SD-OCT. A recent report on combining the line imaging and FD-
OCT has been developed (Nakamura et al., 2007) for very high-
speed 3D retinal imaging. By this technique, the A-line rate
significantly improved to 823,200 A-lines/s for single frameimaging and 51,500 A-lines/ s for continues frame imaging. A
high-speed 2D CMOS area camera with effective pixels of 1104
(horizontal)256 (vertical) was used. The columns of 256 pixels
are oriented along the vertical lateral size of the image while the
horizontal arrays of pixels are used for spectral analysis of
the channeled spectrum, supplying the depth information in the
B-scan. With an image size (the length of the vertical line
projected on the retina) of 2.1 mm and a pixel size of 16.4mm, the
number of pixels along the vertical line is 128. Due to the decay of
sensitivity of the FD-OCT method with depth, a maximum depth
range of 2.88 mm was achieved and images with sufficient
contrast presented up to 0.8 mm in depth, which gives 108 pixels
of 7mm depth resolution (Table 1).
The frame rate at continues frame imaging is 201fps. This 3Dacquisition speed is more than two-fold higher than the
acquisition speed of standard flying spot SD-OCT. To enhance
the sensitivity, pulse illumination was used for the duration
of the frame, of 0.3 ms. The in-vivo 3D retinal imaging with
256 B-scan image frames was successfully performed in
1.27s, which gives a Mv/s 2.8 for a collection of
128140256 pixels. The time to produce a C-scan image,
Tenface, is quoted as 1.27 s in Table 1 considering the number of
lateral pixels of 128256 in the images reported. However, if
512512 C-scan images are required, Tenface scales eight times
larger to over 10 s.
4.5. Multiplexing in OCT
Multiplexing refers to acquisition of data from several points at
the same time. Extending this notion to OCT, would mean
acquisition of several OCT images simultaneously.
4.5.1. Multiplexing in A-scan-based OCT imaging
One reason for the recent success of SD-OCT method in the eye
imaging is the fact that the depth information is somehow
multiplexed, i.e. one FFT of the photodetected spectrum contains
the OPD of all resolved scattering points along the depth of the A-
scan. One spectrum contains all depth information, where each
depth is coded in the number of peaks in the channeled spectrum
at the interferometer output. This is not the case in TD-OCT, where
when scanning the depth, the same Doppler shift is obtained, for
any of the scattering points resolved.To a larger extent, multiplexing of images means different
procedures in A-scan-based OCT and T-scan-based OCT. 3D
complete information could be collected in different ways, either
acquiring many B-scan OCT images at different en-face positions,
as shown in Fig. 2(iv) or many C-scan OCT images at many depth
positions, as illustrated in Fig. 2(iii). Different possibilities exist
and the choice will depend on the best way to use the available
S/N ratio and the available bandwidth to collect simultaneous
images from the retina. Collecting several B-scan OCT images as in
Fig. 2(iv) would involve splitting the light in the sensing arm,
possible by using mirrors and beam-splitters. The optical
elements in the sensing arm have to be arranged in such a way
that in each new arm the optical path is the same. Consequently,
such a procedure would involve cumbersome optics before thetransverse scanner along with matching the optical paths of the
different sensing arms within a few wavelengths. Additionally, the
signal in each sensing arm and so in each channel decreases
proportionally with the number of channels.
4.5.2. Multiplexing in T-scan-based OCT imaging
In the case of T-scan-based imaging, the reference paths can be
split instead, thus avoiding any disturbances with the transversescanner in the object arm. Different optical paths can be employed
to obtain C-scan slices at different depths. This is why the en-face
OCT imaging is more tolerant to simultaneous collection of slices.
Although apparently it should be equivalent to build the 3D
profile from either longitudinal slices or en-face slices, the latter
procedure is less cumbersome in terms of technical implementa-
tion and less lossy in term of the object signal power. For the same
voxel volume (resolution) and number of voxels, the time taken
and the amount of memory required for storage is the same
irrespective of the method, Fig. 2(iii) or (iv).
Such a procedure, of dividing the power in the reference path
has been explored in two prior reports. Two OCT channels have
been demonstrated using a two splitter configuration (Podoleanu
et al., 1997). A different configuration employed an integrated
MachZehnder modulator, where two delays have been intro-
duced in the reference arm, each with its own RF modulation
(Podoleanu et al., 2001). The frequency modulation limit of the
first and the dispersion of the modulator of the second rendered
these approaches unsuitable for in-vivo applications.
Another possibility is to split both the object and reference
arms as shown in Fig. 3 (Podoleanu et al., 2004b). In this way, two
independent OCT imaging channels are assembled. The depth
scanning proceeds simultaneously in the two OCT channels and
from the same range, however, a differential optical path
difference can be introduced between the two channels. In this
way, two simultaneous images are generated where the depth
differs in each transversal pixel by the differential optical path
difference. A dual OCT system, OCT/OCT, working at 850 nm was
devised and its capability demonstrated by simultaneouslyacquiring images from the optic nerve and fovea of a volunteer.
The configuration ensures a strict pixel-to-pixel correspondence
between the two images irrespective of the axial eye movements,
while the depth difference between the corresponding pixels is
exactly the set differential optical path difference. The images are
collected by fast en-face scanning (T-scan), which allows both
B- and C-scan acquisitions. The reference light is passed via mirrors
M1 and M2 and the depth is selected in both channels at the same
time. A differential path difference between the two interferom-
eters is created by moving the mirrors M1 and M2. In this way, any
differential delay, d, of 1 to 1500 mm could be introduced.
ARTICLE IN PRESS
Low
coherence
source
INTERFEROMETER
1
INTERFEROMETER2
Depth 1
Depth 2
ImageDepth2
ImageDepth1
PC
Control of depth 1
Control of depth 2
M 1
M 2
Fig. 3. Schematic diagram of the OCT/OCT system to provide C-scan images at
different depths. The output beam from the low-coherence source is sent to two
interferometers,1 and 2, with independent adjustment of the imaging depth, using
mirrors M1 and M2, which are simultaneously and synchronously controlled by apersonal computer, PC.
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499 471
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
9/36
Fig. 4 presents images from the optic nerve of a volunteer. Pairs
of images have been collected at D 100mm OPD steps, obtained
by moving the two mirrors M1 and M2 simultaneously, whilemaintaining their differential separation d. The separation
between the pairs may be different from D in reality due to axial
eye movement. However, the images in each pair are from depths
separated by d 100mm exactly (the bottom image is deeper by
100mm), because d is independent to the eye movement. Within
each pair, the differential depth difference is certain because the
two images are collected simultaneously, while the depth
difference between the pairs is only approximate due to eye
movements. Therefore, the images should be interpreted strictly
within the pair, on each vertical in Fig. 4, as associated pairs of
pixels displaced in depth by 100 mm.
The problem for the current technology is how to produce at
the same time 10 or 100 of such images from different depths. If
such a technological difficulty can be overcome, en-face OCT could
reach Mv/s values similar to those obtained with the SD-OCT
method. So far, only two channels could be successfully demon-
strated.
4.6. Comparative assessment of the OCT methods
Each method has several advantages and disadvantages.
Advantages of SD-OCT: The main performance that dictates the
supremacy of SD-OCT in comparison with the TD-OCT is its
superior sensitivity or enhanced S/N ratio. This reflects in better
penetration and/or faster acquisition speed. Theoretical models
estimate that SD-OCT is better than TD-OCT by a factor of 10 at
least which will be considered for our comparative analysis in
what follows.Disadvantages of SD-OCT: Although the SD-OCT method is
currently favored for its speed, it has inherent limitations. In both
formats (FD-OCT and SS-OCT), SD-OCT has three main disadvan-
tages: (i) decay of sensitivity with the OPD, which means that the
relative intensity of layers along the depth in the retina is not real,
in fact their intensity depends on how far the retina is from the
value of OPD 0, which varies due to head position relative to the
chin rest; (ii) dynamic focus not possible (see below), i.e. ensuring
that each depth in the OCT depth is in focus when acquiring signal
from that depth; (iii) the optical spectrum of the interferometer
output consists of symmetric spectral terms, i.e. the same image
results for positive and negative OPDs. For the latter, an initial
adjustment of the OPD 0 outside the range of interest is
required. This is not possible all the time, especially whenimaging moving thick organs or tissue, and this is a problem for
imaging the eye too. Different methods have been devised to
attenuate the symmetric terms in order to obtain a correct image
such as phase-shifting interferometry, or complex signal proces-sing (Targowski et al, 2004), which are cancellation techniques
(and so sensitive to movement) requiring several images or steps
(at least 3). An independent method to the target movement was
also developed (Podoleanu and Woods, 2007).
Unexploited potential in multiplexing of en-face OCT images: This
refers to the possibility of improving the overall number of Mv/s.
In terms of line rate, TD-OCT could reach fast line-scanning rates
using resonant scanners (4, 8 and 16 kHz are available). This is 27
times less than the scanning rate of modern line scan cameras
used in FD-OCT and more than an order of magnitude smaller
than the rate achievable using SS-OCT. It looks unlikely that the
line rate in en-face OCT can be further increased. Polygon mirrors
may achieve faster line-scanning rates, but they introduce a non-
linear dependence of the OPD with scanning and the main
limitation is the S/N ratio in TD-OCT. However, for any given line
rate, en-face OCT has an unexploited potential in the possibility of
simultaneous acquisition of several C-scan images, at the
expense of power division in the reference path, where power is
normally attenuated to reduce the noise, as explained above in the
Section 4.5.2.
Table 1 presents the performance of the OCT technology in
imaging the retina in-vivo as presented in a selection of reports.
The first columns show the acquisition rate and number of pixels
in the image, depending whether B-scan or C-scan, while the last
two columns on the right are the most significant. They display
two main parameters: the number of Mv/s and the time required
to produce a C-scan image, Tenface. The table starts with the 1st
report of OCT from retina at MIT (Huang et al., 1991) and ends
with the highest OCT speed reported using chirped lasertechnology (Moon and Kim, 2006) to implement SS-OCT method
(although not on the eye). Where the authors have not specified
the numbers of pixels, such numbers were inferred from the data
available in each report, as detailed in the footnotes. The size of
the image varies from report to report, therefore the number of
Mv/s represents a suitable performance to compare the different
methods. The number of pixels in Table 1 are those specified in the
images presented in the reports. A rigorous comparison would
require an evaluation of the techniques mentioned based on a
similar size image, however, this was not possible. Anyway, the
important numbers are the order of magnitude of the Mv/s and
Tenface and not their exact value.
The time to produce an OCT cross-section (B-scan) image is not
used as a comparison criterion because on one hand is the inverseof the frame rate in A-scan-based systems and on the other,
ARTICLE IN PRESS
Fig. 4. Images from the optic nerve. Separation between pairs: D 100mm. Separation between images in the pair: d 100mm. Lateral size: 2.5mm2.5mm. Pairs
collected at 2Hz (Podoleanu et al., 2004b).
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499472
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
10/36
a T-scan-based OCT system can be switched into B-scan
regime and produce a B-scan in the same time as for
generating a C-scan, Tenface. In other words, producing a B-scan
image is not a challenge for T-scan-based OCT systems while
producing a C-scan from A-scan-based OCT systems is. The
parameter Tenface also represents the time required to produce
an SLO-like image in the A-scan-based OCT systems, and this has
clinical relevance in the case of SD-OCT, which generates B-scans
and not C-scans.
The graph in Fig. 5 is assembled based on the exemplary reports
in Table 1 only, for simplicity. In reality, a large spread of points could
be placed in the graph based on the articles mentioned in the
reference list, however, when this is completed, they display
the same trend as shown using only the few points representing
the publications in Table 1. The graph shows extraordinary progress
in terms of the number of Megavoxels imaged in a time unit, due to
the progress in SD-OCT and especially in the case of SS-OCT imaging
using mode-locked SS technology (Huber et al., 2007). The SS-OCT
has further potential for improvement in data acquisition rate since
the chirped laser principle allows line rates in excess of tens of MHz.
However, the increase in the line rate is currently accompanied by
reduction in the signal and increase in noise due to large electronic
bandwidth required.
For A-scan-based OCT systems, the time to collect the wholevolume of voxels determines the time required to produce a
C-scan image, Tenface, since such an image is only available once
all data have been acquired. Where data were not available,Tenface was inferred considering a number of 256 frames. The
graph in Fig. 6 illustrates the progress over the years in reducing
Tenface. For A-scan-based TD-OCT systems, Tenface exceeds tens of
seconds, too long for imaging a moving eye (although the majority
of reports on OCT before 1998 required such large time intervals
for acquiring multivoxel data). To the right of the graph, progress
in mode-locked SS-OCT lead to a time of less than 1s, which
becomes useful in practice. However, even if the progress was
substantial in the last few years, this value is still more than an
order of magnitude larger than the time to produce a C-scan
image using an en-face OCT system equipped with resonantscanners (Hitzenberger et al., 2003).
In the past few years, progress has been reported not only
in the improvement of the two parameters, Mv/s and Tenface,
but in the depth resolution too. Excellent accounts on the
resolution improvement in OCT are presented elsewhere (Drexler,
2004). This is however implied in the number of pixels in depth in
the Table 1.
5. Depth of focus range and dynamic focus in OCT
In order to obtain images with high transverse resolution
throughout the whole depth of the retina, dynamic focus is
essential. Dynamic focus means maintaining the coherence gate
and the focus gate in synchrony in OCT. The confocal core of the
OCT channel is what determines the depth of focus in the OCT, and
this is an important issue in the progress towards high resolution,
sometimes ignored. A good S/N ratio requires that the confocal
core of the OCT channel focuses at the same depth where the
coherence gate of the OCT selects signal from. The procedure is
often utilized in the TD-OCT, where the focus and the coherence
gate can be synchronously scanned. A TD-OCT A-scan-based
system requires that the focus scanning be performed at the line
rate. In an A-scan-based TD-OCT system, dynamic focus is in
principle possible, but unachievable technically due to the high
speed required for focus change (it is difficult to move a lens at
kHz rate). A T-scan-based OCT system relaxes this demand, as the
focus needs to be changed at the frame rate, of the order of Hz or
tens of Hz, which is much smaller.
In opposition to the TD-OCT methods, dynamic focusis not applicable to FD-OCT and SS-OCT due to the very
principle employed. Therefore, for such systems, the interface
optics are devised with a large depth of focus, to accommodate
the entire range of the A-scan, usually, 12 mm. Not providing
focus change with depth is visible especially in imaging the
optic nerve, which extends for more than 2 mm, and where
the contrast in the image decreases at the image edges.
The constant focus in SD-OCT precludes the possibility of using
a high NA objective to enhance the transverse resolution and also
limits the efficiency of the combined SD-OCT/AO method as
detailed below.
Dynamic focus applied to a TD transversal (en-face) scanning
system for retinal imaging (Pircher et al., 2006b) demonstrated
that a transverse resolution of 4.4 mm can be achieved over anoptical depth of 1 mm in a model eye.
ARTICLE IN PRESS
Max SS-OCT
SS-OCT:
FD-OCT:
345:
TD-OCTResonantscannerTD-OCT
LF-FD-
OCT:0
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
20
40
60
80
100
120
140
Megavoxel/s
Year
Fig. 5. Evolution of OCT technology including TD (time domain), FD (Fourier
domain), SS (swept source) and LF (line field) for imaging the retina in terms of
Megavoxel/s. The horizontal line at 345 Megavoxel/s represents the maximum
achieved today in terms of Megavoxel/s, using SS-OCT (Moon and Kim, 2006).
0
2
4
6
8
10
12
14
16
18
TimetoproduceaC-scan(s)
Year
Minimum, using transverse resonantscanner in en-face OCT
Mode LockedSS-OCT:
FD-OCT:
0.019
TD-OCTResonantscanner
SS-OCT:
SS-OCT
FD-OCT
1998 2000 2002 2004 2006 2008
Fig. 6. Evolution of Tenface, the time required to produce a C-scan image of the
retina, using longitudinal OCT (including TD and SD). The horizontal line at 19 ms
represents the time reported for en-face OCT using a resonant scanner
(Hitzenberger et al., 2003).
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499 473
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
11/36
6. Combining OCT with SLO
OCT has mainly evolved in the direction of producing cross-
sectional images, most commonly perpendicular to the plane
orientation of images delivered by a microscope or by an SLO.
Several groups have shown that the flying spot concept, utilized in
the SLO hardware can be combined with the OCT technology to
produce en-face OCT images from the anterior and posterior pole.The depth resolution in SLO is 30100 mm coarser than that in OCT
while the transversal resolution in OCT is affected by random
interference effects from different scattering centers (speckle),
inexistent in SLO images. Therefore, there is scope in combining
SLO with OCT. Once C-scan OCT images have had been demon-
strated, sharing the same natural orientation to that of SLO
systems, the next step was to combine OCT with SLO. Different
avenues have been evaluated, to provide SLO and en-face OCT
images simultaneously, quasi-simultaneously or sequentially.
The simultaneous OCT/confocal technology has been exten-
sively evaluated on more than 2000 eyes with pathology, using
what is now called the OCT/SLO or the OCT/Ophthalmoscope
instrument (Rosen et al., 2003). A variant of this instrument, the
OCT/SLO/indocyanine-green (ICG), has also been tested provingthe potential of multi-modal imaging. This allows collection of
simultaneous en-face OCT and ICG fluorescence images from the
retina. The sequential OCT-confocal imaging procedure is still in
laboratory phase. This concept provides better S/Nratio in the OCT
channel and better depth resolution in the confocal (SLO) channel
than the OCT/SLO system. The performances of the sequential
procedure will be discussed in comparison with the simultaneous
procedure.
Different solutions have been provided to assemble OCT/SLO
systems, depending on the scanning type and on the OCT regime
of operation. The main motivation for OCT/SLO combination is to
provide orientation to the OCTchannel. Crucial for the operation is
pixel-to-pixel correspondence between the two channels, OCT and
SLO, which can only be ensured if both channels share the same
transverse scanner to scan the beam across the eye. Different
possible configurations are shown in Fig. 7. Any OCT system is
constructed around an interferometer illuminated by a low-
coherence source. The interferometer is equipped with some
means to adjust the OPD to determine the axial (or depth)
scanning. To send the signal towards the eye, the OCT is equipped
with an interface optics, which contains lenses or curved mirrors
and the XY scanner. Light collected from the eye is received either
in a pinhole or via a single-mode fiber in order to ensure a high
visibility for the interference signal. This acts also as a confocal
core, which can be used for implementing an SLO channel,
however, this is not possible all the time as explained below,
therefore different possibilities exist.
Fig. 7(i) illustrates the principle of the OCT/SLO instrument
where the confocal core of the OCT is used to produce the SLOsignal. No splitting of the light from the eye is performed. Fig. 7(ii)
shows a different configuration, where light for the SLO channel is
produced using a separate optical source and light received from
the eye is diverted towards a separate confocal receiver. This
requires a splitter to divert light into the two paths, OCT and SLO.
Fig. 7(iii) is a simplification of Fig. 7(ii) where the same optical
source is shared by both channels. Again, this requires a splitter to
divert light into the two paths, OCT and SLO.
Historically, while the simplest configuration looks like
that in Fig. 7(i), the first OCT/SLO (Podoleanu and Jackson,
1998) implemented employed the configuration in Fig. 7(iii).
It was 7 years before a sequential version of Fig. 7(i) to be
described, and a simultaneous implementation has not yet been
reported, the main problem being the associated noise with thehigh reference power.
Fig. 7(ii) is useful in combining OCT with fluorescence. In this
case, the separate SLO source excites fluorescence, which is
processed in a separate SLO receiver. Such a configuration would
be useful in combining OCT with fluorescein angiography (FA),
where the OCT operates at 700900nm and the separate source
operates in greenblue to excite fluorescein. Such a OCT/
fluorescence configuration has been reported in endoscopy
(Barton et al., 2004).
The three generic configurations presented in Fig. 7 help to
reveal the diversity of possible combinations of OCT with SLO in
terms of the scanning regimes. SLO provides C-scan images.
Therefore the natural combination of the two channels would be
that where the two images generated are both C-scans. However,
solutions have been provided where the SLO channel maintains its
natural C-scan orientation while the OCT channel operates in
B-scan regime, in order to implement SD-OCT. In time, the two
images can be generated simultaneously or sequentially. Simulta-
neous generation could be implemented in C-scan orientation by
both channels. C-scan SLO and B-scan OCT, however, can onlybe sequential, if the same source and transverse scanner is to be
shared. (The operation in C-scan SLO and B-scan OCT can also be
achieved by combining an OCT channel and an SLO channel each
equipped with its own XY scanner, via a splitter, however, such a
configuration is not practical and will be difficult to ensure pixel-
to-pixel correspondence when using two different, independently
ran XY scanners).
The main advantage of the en-face imaging is that it allows
integration of SLO with OCT. This has proved useful in allowing
ophthalmologists and visions scientists to associate features seen
in cSLOs and SLOs with those highly fragmented due to enhanced
depth resolution in en-face OCT. Such a method allows a dual
presentation of high-resolution images (OCT and SLO) in different
regimes of operation, B- or C-scan, providing cross sections indepth or constant depth images, respectively. Other current
ARTICLE IN PRESS
OPD
control
SLO/OCT Aperture
Interferometer
Object
arm
Reference
armOptical
source
OPD
control
OCT Aperture
Interferometer
Object
arm
Referencearm
OCT
Optical
source
SLO channel
OPD
control
OCT Aperture
Interferometer
Object
arm
Reference
armOCT/SLO
Optical
source
SLO receiver
SLO
Optical
source
(i)
(ii)
(iii)
Fig. 7. Generic configurations used in the combination of OCT with SLO.
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499474
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
12/36
developments include sequential OCT-confocal regime of opera-
tion, dual OCT/confocal fluorescence imaging and triple OCT/SLO/
confocal fluorescence imaging.
6.1. Simultaneous OCT/SLO
Simultaneous acquisition of images in two channels, OCT andSLO requires configurations using splitters, according to Fig. 7(iii).
In (Podoleanu and Jackson, 1998), a small fraction of the light
returned is diverted towards the SLO channel using a separate
splitter. There is an optimum splitting ratio which ensures
sufficient and similar S/N ratio in both channels (Podoleanu and
Jackson, 1999).
New imaging technology brings not only new information to
the clinician, but with it, the requirement of interpretation. En-
face OCT is no exception in this respect. The higher the depth
resolution of the OCT system, the more fragmented the en-face
OCT image looks like (Podoleanu et al., 1999). The fragmentation
is especially visible when the plane of the retina is tilted in
relation to the scanning plane. First, the en-face OCT image
appears fragmented, and on its own, such an image is difficult tointerpret. Second, variations in tissue inclination with respect to
the coherence wave surface alters the sampling of structures
within the depth in the retina, creating unbalanced distortions
among the elements being sampled (Podoleanu et al., 2004a). The
bright patches in the OCT image represent the intersection of
the surface of OPD 0 with the tissue. Due to the particular way
the retina is scanned, with the fan of rays converging on the eye
pupil, the surface of OPD 0 is an arc circle with the center in the
eye pupil. Depth exploration requires that the radius of the arc is
altered. If the arc has a small radius, it may just only intersect the
top of the optic nerve with the rest of the arc in the aqueous. The
radius of the arc is changed by changing the length of one of
the arms of the interferometer in the OCT channel to explore the
retina up to the RPE and choroid. The orientation of the retina
tissue at the back of the eye is not planar and this complicates theinterpretation of the image even further. Despite scanning images
in an en-face plane, the result is that the images may display the
structure in depth like in any B-scan OCT image. This is especially
visible in the high-resolution en-face OCT as shown in Fig. 8 (Cucu
et al., 2006), where the C-scan slice is very thin (3 mm). These two
effects, (i) fragmentation and (ii) multiple depths simultaneously
displayed in the C-scan images are present in a cSLO with high
depth resolution as well, however, at a scale where they are
regularly discarded. In a cSLO, the images do not look fragmented
and the depth structure is barely visible due to the coarse depth
resolution, 0.3 mm, comparable to the retina thickness. Going in
and out of focus results in a smooth transition from dark to bright
areas in the image. Both problems mentioned above are brought
about by the high depth resolution of OCT. Providing an SLO imagesimultaneously guides the user and addresses the fragmentation
problem.
In terms of data acquisition, the confocal image adds further
versatility. The design ensures a strict pixel-to-pixel correspon-
dence between the two C-scan images, OCT and SLO. This helps in
two respects: for small movements, the SLO image can be used to
track the eye movements between frames and for subsequent
ARTICLE IN PRESS
Fig. 8. Images acquired from the foveal region of the retina in-vivo using en-face technology and 120nm band SLD source. SLO images on the left and en-face high-resolution
OCT (3 mm depth resolution) images on the right for two different depths in the retina. The structure of layers normally encountered in B-scan OCT images is also visible
here: ILM, the inner limiting membrane; NFL, nerve fiber layer; GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ELM,external limiting membrane; IS/OS, junction between the inner and outer photoreceptors; RPE, retinal pigment epithelium; CC, choriocapillaris; C, choroid; V, the vitreous.
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499 475
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
13/36
transversal alignment of the OCT image stacks; for large move-
ments and blinks, the SLO image gives a clear indication of the
OCT frames that need to be eliminated from the collected stack. As
a reference for the aligning procedure, the first artifact-free
confocal image in the set is used.
In the B-scan regime, movements of the eye are indicated by
lateral shifts of the confocal traces (Podoleanu et al., 2004a). Each
horizontal line in the SLO image when the OCT/SLO is switched to
B-scan, corresponds to a depth position. The relative lateral eye
movements lead to slight deviations of contours in the SLO image,
which can be employed to correct the lateral shift of the lines in
the B-scan OCT image (also illustrated here in left insets below theB-scan OCT images in Figs. 1014 below).
A bulk interferometer solution for simultaneous acquisition of an
OCT and an SLO image was also reported (Pircher et al., 2006a). The
bulk configuration allows placing the optical splitter close to the
optical source with the advantage of no signal lost towards the OCT
channel (however requiring to compensate for loss of power by
increasing the optical source power). Using such a system equipped
with dynamic focus, the cone mosaic was imaged simultaneously in
the SLO and OCT channels without AO elements.
Combination of techniques allows correlation of information in
orthogonal planes and facilitates more accurate diagnosis. For
instance, Fig. 9 top presents a B-scan OCT, which represents a
20mm thick slice through the central macula. Since the slice iscollected as the patient is instructed to fixate on a specific target,
ARTICLE IN PRESS
Fig. 9. Illustration of synergy between C-scans and B-scans.
A.Gh. Podoleanu, R.B. Rosen / Progress in Retinal and Eye Research 27 (2008) 464499476
7/30/2019 2011 PRER Elsevier_Podoleanu_retina OCT
14/36
the image captures a macular region where the visual acuity is
best and suggests that the macular anatomy of this patient is
normal. Ho