skip to Main Content

Optical Coherence Tomography Imaging of Diseases of the Central Nervous System

The eye as a window to the central nervous system

Optical coherence tomography (OCT) is a non-invasive, non-contact structural imaging modality (Huang D, 1991), that has unique utility in a number of life science applications.  In ophthalmology it is used to generate in vivo, micron scale, cross-sectional imagery of ocular tissues.  The technology has rapidly evolved, facilitating very fast capture of highly detailed, volumetric data, which, coupled with automated analysis algorithms, has made OCT today’s standard of care in the detection and management of ocular diseases.

The above shows a single OCT image as a slice through the macula of a human retina. It also labels the discrete retinal layers that are readily visible. Their thicknesses are used as biomarkers and can be measured using segmentation algorithms.

 

OCT is routinely used to make clinical treatment decisions with more than 20 million patients being scanned each year by an OCT device. Ocular pathologies manifest themselves in different ways in the retina, and the “optical histology” afforded by this imaging modality has, by facilitating objective structural analysis, been a boon to disease interpretation.
An early example was the PrONTO study that evaluated the response of age-related macular degeneration (AMD) patients to intraocular ranibizumab (Lucentis), which used OCT imaging (Rosenfeld et al., 2005). It was shown that OCT was essential to the management and treatment of this disease as it reported central retinal thickness (CRT) and how this changed as a result of administering the anti-VEGF drug.

Other examples abound. Macular edema will result in a swelling of the macula that may not initially be apparent via a traditional ophthalmic viewing, but can be quantified as having changed by automated algorithms measuring the total retinal thickness (TRT). Drusen, an early indicator of dry AMD, will show as bulges between Bruch’s membrane and the retinal pigment epithelium (RPE) layer, and can also be quantified using OCT (Gregori, 2011). Glaucoma, a complex optic-neuropathy, is characterized by progressive death of the retinal ganglion cells (RGCs). This structural change can also be quantified using OCT, either in the macula by measuring the thickness of the RGC layer, or peripapillary to the optic nerve head (ONH) by measuring the thickness of the retinal nerve fiber layer (RNFL) (Ishikawa H, 2002, Wollstein et al., 2004), the axons of the ganglion cells (Mwanza, et al., 2011).

Image Segmentation

The thickness measurements in routine clinical use are generated using an image processing method known as segmentation. The task of segmentation is to automatically delineate coherent regions in image data. In the case of retinal layer segmentation, the task is to identify the different retinal layers. An early example of OCT image segmentation is given in the figure below (Ishikawa et al. 2005), where retinal tissue is delineating using various image processing techniques. But despite many years of development in this area, reliable image segmentation remains a challenging task, especially in diseases of the outer retina where the regular topology of the anatomy can be significantly disrupted. Yet even in cases of less disruptive pathology, where, for example, the structural change is a thinning of the layer, only two or three layers are reliably segmented by current commercial software. The technology has been slow to support fundamental clinical research to determine how these discrete neuronal layers are affected by different ocular diseases and neuropathies. To answer this, large-scale studies are required and findings need to be confirmed across different centers. Each time the software takes a step forward in terms of functionality, we learn something new clinically.

An early example of a OCT image automatically segmented using software developed at Department of Ophthalmology, University of Pittsburgh School of Medicine (Ishikawa 2005). The lines drawn into the image are shown in white (inner limiting membrane), black (posterior of the retinal nerve fiber layer), blue (posterior of the inner nuclear layer), blue (posterior of the outer plexiform layer) and dark blue (outer segments of the photo receptors). The labels in the figure are: macular nerve fiber layer (mNFL), the inner retinal complex (IRC), outer plexiform layer (OPL), and outer retinal complex (ORC) (Ishikawa et al. 2005).

 

Optical Coherence Tomography in Neuro-Ophthalmology

“The eye is the window into the brain and by measuring how healthy the eye is, we can determine how healthy the rest of the brain is”, (Calabresi, 2012).

The retina is an extension of the central nervous system and is made up of different layers of neuronal tissue. Each of these layers is visible and readily measurable using OCT with analysis algorithms. Interest in these measurements as potential biomarkers has existed in the neuro-ophthalmic community for some time, particularly in multiple sclerosis (MS), where up to ~60% of patients will exhibit acute optic neuritis as a symptom of the disease (McDonald & Barnes, 1992, Frohman EM, 2008). Thinning of the RNFL measured using OCT is the first real candidate biomarker for this debilitating disease, but improvements in the analysis algorithms have unearthed more interesting findings and potentially more relevant biomarkers in the inner retinal layers.

Beyond the Retinal Nerve Fiber Layer

As discussed in (Kardon, 2011), the use of the RNFL thickness alone as a structural biomarker has various pros and cons. Among its advantages is its correlation with functional testing of visual fields such as perimetry. With different pathologies, however, this structure-function relationship is uncertain and contends with limitations that include a lack of RNFL thinning in the presence of functional defects (Hood, 2007).
Measuring instead the thickness of the GCL, can potentially overcome some of the aforementioned disadvantages. Without non-neuronal tissue confounding the measurement, a more accurate and direct mapping to the visual field can achieved (Kardon, 2011).
Automatic segmentation of the GCL is more challenging, however, as the interface between the layer and the layer below it, the inner plexiform layer (IPL) has indistinct changes in refractive index, meaning the back-scattered signal, the basis of OCT imagery, changes little and the boundary cannot normally be seen. Instead a so-called ganglion cell complex is used. It is comprised of the GCL + the IPL. The posterior boundary of the IPL is apparent in OCT, but less so than the posterior of the RNFL. Segmentation of the complex is also more challenging and, initially, performance limitations in the OCT retinal layer segmentation software meant the measurements could not be reliably made.
Advances in the technologies often lead to new clinical findings, with one group, for example, able to show how thinning of the ganglion cell complex reflected global CNS pathology in MS (Ratchford, et al., 2013), offering that OCT-derived GCIP thickness measures may have utility as an outcome measure to assess neuro-protection in MS.

Beyond the Ganglion Cell Layer

For inflammatory attacks of the optic nerve, the thicknesses of the RNFL and the GCL are the obvious retinal biomarkers. If, however, the attacks are not limited to the myelin, OCT analysis of deeper neuronal tissues may provide missing insight into other pathological changes. In MS, histological evaluations have shown pathological changes beyond the RNFL and GCL, indicating that retinal injury is indeed more widespread than initially thought (Green, 2010); earlier electroretinography studies implicated inner nuclear layer dysfunction (Gills, 1966). If confirmed, new biomarkers involving inner nuclear layer thicknesses could allow researchers to better target the symptoms of MS.

Again, the technology to support these measures in vivo in a patient population has been lacking, leaving the assessment of these inner retinal layers largely unexplored. Until CirrusTM version 6.0 in January of 2012 was released for their HD-OCT instrument (CZMI, Dublin, CA), it was not possible to purchase commercial software segmenting anything other than total retinal thickness (TRT), RNFL and GC-IPL.

Using a prototype version of the Cirrus 6.0 software, researchers at Johns Hopkins were early to utilize the advances in OCT segmentation and to begin studying the effect of the disease in the deeper retinal layers. With the new software, they initially looked at macular thinning in the absence of optic nerve pathology, as this would implicate primary retinal pathology in MS. They found a subset of patients with significant thinning of the INL and ONL complexes with relative sparing of the RNFL and GCL (Saidha, et al., 2011). Their findings indicating that, neuronal loss in the retina could occur independent of myelin injury and axonal damage, supporting what was initially suggested in (Gills, 1966) and confirming some of the findings of (Green, 2010).

In MS, the consequences are significant as, having seen patients with all the classic clinical symptoms of MS, but void of inflammatory axonal damage as revealed by OCT and instead showing primary retinal neuronal pathology, MS is being rethought as something that is not only an inflammatory disease.

Additional findings continued apace as the same collaborative group found INL thickness to be a predictor of relapses, lesions, and disability progression in MS patients (Saidha, et al., 2012). Later that year, they presented OCT measures correlating with intracranial volume in MS and healthy controls and found that lower ONL thickness appears to reflect lower cerebellar white matter volume (Saidha S, 2012). In the same work, they showed how normalizing to intra-cranial volume (ICV) could allow single measurements to have greater clinical utility and also how certain correlations only become apparent once such normalization is done. In doing this, they were able to associate macula RNFL and GCL-IPL with gray matter thickness across the cortex as measured using MRI and also an association between INL thickness and white matter lesions and atrophy in relapsing-remitting MS patients. In patients with a history of optic neuritis, they were able to elucidate associations between the ONL and most gray matter structures.

OCT-based measurements in the eye are being considered as complementary to measurements made in the brain. These measurements are not just surrogates, but independently useful in the management of diseases of the central nervous system. OCT measures offer a number of possible biomarkers for screening, diagnosis, prognosis and disease management.

Beyond MS

Quantitative OCT measurements in the retina will be used more and more not just in ocular diseases but as complementary and surrogate measures in a multitude of neurological diseases, and not limited to those exhibiting only secondary visual dysfunction. Recent examples include:

  • Parkinson’s, where different Parkinsonian syndromes have been found to be associated with distinct changes in retinal morphology (Albrecht, et al., 2012);
  • Alzheimer’s, where it has been shown that specific RNFL pattern abnormalities were present in early-stages of the disease (Marziani, et al., 2013);
  • Amyotrophic lateral sclerosis (ALS), where OCT retinal layer segmentation showed subtle reduction in the thickness of the RNFL, yet a marked thinning of the inner nuclear layer (INL) (Ringelstein, et al., 2014).

Limitations exist, however, again with the technologies supporting the analysis software. As is shown in above, the segmentation performed is limited to complexes of layers; for example, neither of the nuclear layers can be measured in isolation. Assumptions were necessarily made in the findings reported in the previous section that, for example, the plexiform layers are fairly constant with pathology. Such assumptions are open for debate, with one recent study based on manual segmentations reporting that the OPL was thicker in ALS patients and noted how compensatory processes in the IPL confound measures of thinning in the GC-IPL as a single entity (Ringelstein, et al., 2014).

Given the significant amount of interest from clinical researchers and also pharmaceutical companies in constant search for compelling biomarkers, the current need for more advanced retinal segmentation software is dire (Ng, 2014). This desperate need for biomarkers supporting neuroprotective therapies is unrelenting, while at the same time the burden on society of neurodegenerative diseases increases. Gene-sequencing to learn about various diseases and their treatment is now being supported by the likes of Google, but void of compelling biomarkers indicating an onset of a disease and response to treatment, knowledge of one’s predisposition to, say, Alzheimer’s, is less useful. Quantification of structure with direct association to clinical assessment allows for important end-points in drug trials looking to arrest the onset of such terrible diseases.

And not only is the software limited in the number of layers, it’s limited by device. In the US, where only the Cirrus offers more than two layers in the macula, this software cannot be used to analyze data from a different device. Clinical trials using, therefore, different instrumentation cannot readily compare their findings.

Working to address exactly these issues, Voxeleron is collaborating with groups like the one at Johns Hopkins led by Professor Calabresi.  We are  also working with the neuro-ophthalmology group from the University Hospital in Barcelona led by Professor Villoslada and the one from UCSF led by Professor Green to develop software with more clinical utility. Currently in beta, OrionTM is our segmentation software that automatically segments seven retinal layers in volumetric OCT data. In addition to on-going clinical studies, it has been validated with two publications (Johung, 2013) (Oakley, et al., 2014). An example screenshot of the software is shown below.

The Orion software (Voxeleron LLC, Pleasanton, CA) segments seven retinal interface automatically and is device independent.

 

Conclusion

As an extension to the central nervous system, the human retina consists of layers of neuronal tissue. The ability to precisely measure these retinal structures helps gauge the health of the eye, and offers direct anatomical correlates to brain structure and its overall health.  But this is only possible using new, state-of-the-art analysis algorithms.  Existing technologies have left a critical need for biomarkers supporting neuroprotective therapies in retinal imaging; the need is only heightened as large-scale prospective studies of various neuropathies loom on the horizon.

It is Voxeleron’s vision to address these technology gaps and advance the capabilities of OCT analysis software.  We are currently preparing to release OrionTM with exactly this objective in mind.  And given the interest in retinal OCT as a window to disease processes in the CNS, we can expect in the coming years more technical advances from ourselves and other organizations, many of which will likely be followed by improved clinical understanding as researchers are empowered with the tools they need.  The consequence is that OCT’s utility will reach far beyond ocular diseases, and is poised to become a ubiquitous and powerful tool for a variety of optic neuropathies and neurologic diseases.

References

Albrecht, et al., 2012 – Optical Coherence Tomography in Parkinsonian Syndromes.  Albrecht, P., Müller, A.-K., Südmeyer, M et al., PLoS ONE.

Calabresi, 2012 –  http://www.hopkinsmedicine.org/news/media/releases/using_the_eye_as_a_window_into_the_brain

Frohman EM, 2008 – Optical coherence tomography: a window into the mechanisms of multiple sclerosis.  Frohman EM, Fujimoto JG, Frohman TC, Calabresi PA, Cutter G, Balcer LJ.  Nat Clin Pract Neurol. 2008 Dec;4(12):664-75.

Gills, 1966 – Electroretinographic abnormalities and advanced multiple sclerosis. Gills JP Jr. Invest Ophthalmol 1966; 5: 555–9.

Green, 2010 – Optical coherence tomography: a sign of the times.   Green AJ.  Neurology. 2010 Oct 12; 75(15):1312-3.

Gregori, 2011 – Spectral Domain Optical Coherence Tomography Imaging of Drusen in Non-Exudative Age-Related Macular Degeneration.  Gregori, G., Wang, F., Rosenfeld, P., Yehoshua, Z., Gregori, N., Lujan, B., et al. (2011). Ophthalmology, 1373–1379.

Hood 2007 –  A framework for comparing structural and functional measures of glaucomatous damage.  Progress in retinal and eye research.  Hood D, Kardon R.  2007 November. 26(6):688-710.

Huang D, 1991 – Optical coherence tomography.  Huang D, Swanson EA, Lin CP, Schuman JS, Stinson WG, Chang W, Hee MR, Flotte T, Gregory K, Puliafito CA, et al.  Science. 1991 Nov 22;254(5035):1178-81.

Ishikawa H, 2002 –  Detecting the inner and outer borders of the retinal nerve fiber layer using optical coherence tomography.  Ishikawa H, Piette S, Liebmann JM, Ritch R.  Graefe’s Arch Clin Exp Ophthalmol 2002;240: 362-71.

Ishikawa H, 2005 – Macular segmentation with optical coherence tomography.  Ishikawa H, Stein DM, Wollstein G, Beaton SA, Fujimoto JG, Schuman JS. Inves Ophthal Vis Sci 2005; 46: 2012-7.

Johung, 2013 –  he Prevalence of Cirrus SD-OCT Ganglion Cell Segmentation Errors in High Myopes.  Johung, Oakley, Russakoff, Sabhlok, Li, & Chang  ARVO 2013.

Kardon, 2011 – Role of the macular optical coherence tomography scan in neuro-ophthalmology.  Kardon R.  Journal of neuro-ophthalmology : the official journal of the North American Neuro-Ophthalmology Society. 2011 December. 31(4):353-61.

Marziani, et al., 2013 -Evaluation of Retinal Nerve Fiber Layer and Ganglion Cell Layer Thickness in Alzheimer’s Disease Using Spectral-Domain Optical Coherence Tomography.  Ermengarda Marziani et al.,  IOVS September 2013 Vol. 54 No. 9 5954.

McDonald & Barnes, 1992 – The ocular manifestations of multiple sclerosis. 1. Abnormalities of the afferent visual system.  W I McDonald and D Barnes.  J Neurol Neurosurg Psychiatry. Sep 1992; 55(9): 747–752.

Mwanza, et al., 2011 – Macular Ganglion Cell–Inner Plexiform Layer: Automated Detection and Thickness Reproducibility with Spectral Domain–Optical Coherence Tomography in Glaucoma.  Jean-Claude Mwanza, Jonathan D. Oakley, Donald L. Budenz, Robert T. Chang,  O’Rese J. Knight and William J. Feuer.  Invest Ophthalmol Vis Sci. Oct 2011; 52(11): 8323–8329.

Oakley, et al., 2014 – Assessing Manual versus Automated Segmentation of the Macula using Optical Coherence Tomography”, Jonathan D. Oakley, Iñigo Gabilondo, Christopher Songster, Daniel Russakoff, Ari Green, Pablo Villoslada. ARVO Meeting Abstracts, 2014.

Ng, 2014 – Ophthalmological Imaging and Application.  Ng, Acharya, Rangayyan, & Suri.  May 5, 2014 by CRC Press.

Ratchford, et al., 2013 – Active MS is associated with accelerated retinal ganglion cell/inner plexiform layer thinning. Neurology, 47-54.

Ringelstein, et al., 2014 – Subtle retinal pathology in amyotrophic lateral sclerosis.  Ringelstein et al., Annals of Clinical and Translational Neurology, April 2014, Volume 1, Issue 4, pp 290–7.

Rosenfeld et al., 2005 –  Optical coherence tomography findings after an intravitreal injection of bevacizumab (Avastin) for neovascular age-related macular degeneration.  Rosenfeld PJ, Moshfeghi AA & Puliafito CA. Ophthalmic Surg Lasers Imaging., Vol.36, No.4, (July-August 2005), pp.331–5, ISSN 1542-8877 .

Saidha, et al., 2011 – Primary retinal pathology in multiple sclerosis as detected by optical coherence tomography.  Brain. 2011 Feb;134(Pt 2):518-33. doi: 10.1093/brain/awq346. Epub 2011 Jan 20.

Saidha, et al., 2012 – Relationships Between Retinal Axonal and Neuronal Measures and Global Central Nervous System Pathology in Multiple Sclerosis. Arch Neurol.

Saidha S, et al., 2012 – Microcystic macular oedema, thickness of the inner nuclear layer of the retina, and disease characteristics in multiple sclerocis: a retrospective study. The Lancet (Neurology), 963-72.

Wollstein et al., 2004 – Optical coherence tomography (OCT) macular and peripapillary retinal nerve fiber layer measurements and automated visual fields.  Wollstein G, Schuman JS, Price LL, Aydin A, Beaton SA, Stark PC, Fujimoto JG, Ishikawa H.  Am J Ophthalmol. 2004;138(2):218-25.

Back To Top