“Be careful, now that I can see the Inner You.”
-Imaging modalities today
Retinal imaging has come a long way today. From times of appropriately concentrating the fluorescein dye from scratch in the lab itself to times where we have a dye free angiography system, technology has covered vast boundaries. There is a platter of different modalities in the market today, each having their own ebb and flow. And what the future beholds for us is certainly inexplicable. To throw a light on these modalities, we have a special interview with someone who continues to extensively work on these technologies and modalities day in and out, Dr Muna Bhende. Dr. Muna Bhende is asenior consultant and deputydirector of Shri Bhagwan Mahavir VR Services, Medical research foundation, Sankara Nethralaya, Chennai.
eOphtha: There is a wide discussion on OCT Angiography (OCT-A) replacing FFA in the field of the medical retina in the near future. What is your take on that? How do you compare the clinical output of OCT+FFA vs OCT-A, if we place the non-invasiveness and repeatability perks of the latter aside?
Dr. Muna Bhende:OCTA has definitely brought to us the biggest advantage – its non-invasive nature that removes the necessity for a more complex backup in the event of complications associated with any injectable dye-based procedure. And this perk can never be set aside. Having said that, FFA has been around since 1959 and is a time tested procedure with classic patterns of hypo and hyperfluorescence that we have grown up with. It still remains critical for evaluation of vascular integrity and leakage and in combination with distinct patterns and biomarkers on OCT gives us a very accurate idea of disease processes involving the retinal vessels, the superficial capillary plexus, and most pathology involving vessels above the RPE, regardless of their flow rates. However, the fact that leakage from ischemic, inflamed, and abnormal retinal vessels obscures many landmarks and the outline of the vessels themselves as the phases of the angiogram progress, the time frame to actually study these vessels is very narrow and may result in loss of very important information. This is where the OCTA has an edge. It combines angiographic visualization of the exact site, size, and extent of neovascular lesions irrespective of leakage, together with the cross-sectional representation of the structure and the effect of the lesion on the surrounding layers on the corresponding B Scan OCT. It has given us the ability to image 3 distinct vascular complexes - superficial, intermediate, and deep, and also vessels in the avascular slab, choriocapillaris, and deeper choroidal slabs. OCTA operates by capturing two consecutive B-scans at each sampling location, from which the red blood cell flow information is extracted using various machine-specific algorithms; therefore it is limited by flow speed within these vascular lesions. Certain lesions with a very low or very high flow may be missed altogether. Another disadvantage is the relatively limited field of view with most regular OCTA systems compared to widefield FFA, though this has to a great extent been overcome with the newer swept-source OCTA machines. In addition, as with all new technologies, there is a learning curve that is steep and involves painstaking segmentation in many instances to accurately localize the pathology along with an understanding of the array of motion and segmentation artefacts that are associated with the procedure, especially in patients with poor fixation and eyes with large irregular highly elevated lesions or variations in macular anatomy and contour. Currently, with a large amount of published research and various algorithms built into the systems, we can fairly confidently use OCTA (along with the corresponding structural OCT) as a safer and more easily repeatable investigation in a large number of ischemic and neovascular retinal, subretinal and sub RPE pathologies including diabetic retinopathy, retinal vein occlusions, AMD and other neovascular pathologies such as Mactel, inflammatory CVN, RAP lesions and PCV. We are still in the process of understanding OCTA findings in some choriocapillaris and choroidal pathologies, inflammations and choroidal tumors, optic nerve, and anterior segment diseases. Algorithms that measure vascular density, flow, and leakage are still mainly used in research.
What will the future bring? Given the speed at which technology is changing our understanding of disease processes by increasing our ability to be able to view structures non- invasively, in real-time and at the microscopic level, we will lose the mystery and challenges that make the practice of retina so fascinating!
IN A NUTSHELL: OCTA certainly has a bright future. So understanding this modality would be extremely useful, especially in early residency or fellowship days itself. On the other hand, FFA will have its minor yet strong stand in certain diseases and situations and is unlikely to be completely replaced even in the future.
eOphtha: How do we differentiate a Neovascular frond, IRMA, and collateral vessels using OCT-A?
Dr. Muna Bhende:The essential feature in a neovascular frond is the breach of the ILM by the new vessel that grows on the surface of the retina and is visualized in the vitreoretinal interface (VRI) slab, a feature typically absent in IRMA. An interesting paper by Pan et al classified NVE in DR into 3 types based on OCTA. Type 1: Arising from the ganglion cell layer and the nerve fiber layer with only 1 main trunk that extends in a tree-like pattern after breaching the ILM and growing along the posterior hyaloid. Type 2: arising from the inner nuclear layer (INL) from capillary networks in the CNP and rapidly branching in an octopus pattern that adheres firmly to the retinal surface. Type 3: Originate from the sea-fan-like IRMAs within the CNP, located more peripherally from the arcades, along the length of the major retinal veins with the main part being intraretinal except for some advancing tips that breach the ILM. NVD is typically seen in the VR interface (VRI slab) and can arise from the artery, vein, or the choroid, either at the peripapillary rim in an arcuate fashion or at the bifurcation of vessels within the cup.
IRMAs usually originate from the major retinal vessels at the IPL to the nerve fiber layer at the margin of the CNP, are confined to that area, originate and drain into major retinal vessels, and do not breach the ILM except for the sea fan variant that forms type 3 NVE. They are visualized in the SCP.
Collaterals are best described in vein occlusions, where blood flow surrounding the occluded segment is channeled into areas of lower capillary pressure, opening up collateral vessels that link vessels that are anatomically the same – vein to vein in a venous occlusion. In CRVO the collaterals are seen at the optic disc. Collaterals are classically noted to be localized to the deep capillary plexus (DCP). A more recent study by Lee et al noted four types of collaterals in RVO (true superficial, true deep, superficial diving, and foveal collateral) based on whether they coursed through the SCP or DCP, and the location of parent vessels connected to the collateral itself. CRVO and HRVO tended to have true superficial collaterals, whereas BRVO tended to have superficial diving collaterals.
IN A NUTSHELL: A neovascular frond is the breach of the ILM by the new vessel that grows on the surface of the retina and is visualized in the vitreoretinal interface (VRI) slab, a feature typically absent in IRMA. IRMAs usually originate from the major retinal vessels and are visualized in the superficial capillary plexus (SCP). Collaterals are vessels that arise due to low capillary perfusion pressure and link anatomically similar vessels (vein-vein). They are classically noted to be localized to the deep capillary plexus (DCP).
eOphtha: Could you briefly discuss En face imaging? How much importance do En face images weigh in your OPD today? Any specific diseases where we specifically need to look for En face images?
Dr. Muna Bhende:Our OCT systems have the ability to capture depth information in slices of tissue at orientation parallel to the retinal surface. These images are called en face images or C scans. To obtain this scan, first a volumetric OCT dataset is created (which is a huge file size, hence not routinely used) and B scans at right angles to the traditional B scans along with enface images are generated. Using multilayer segmentation, various 2D projection images at multiple retinal slabs are created eg. Vitreoretinal interface, RPE, choroid. This scan can be generated on all OCT systems – time domain, spectral domain, and swept-source, however, the best images are obtained with SLO viewing.
The major benefit of an en face image is that it resembles a fundus photograph but at different layers. In fact the term en face OCT is currently used interchangeably with OCTA that provides depth-resolved angiographic maps from the vitreoretinal interface right down to the choroid.
Areas where I personally look with interest at the enface image (commonly in conjunction with the OCTA) are cases of cystoid macular edema, CSCR, preoperatively in an eye with PDR to assess areas of traction, PCV, AMN, PAMM, Geographic atrophy, IRD, MEWDS. However, I admit that it is currently not an indispensable part of my practice but is more of an attempt to understand the disease process and these by no means are general recommendations.
IN A NUTSHELL: Our OCT systems have the ability to capture depth information in slices of tissue at orientation parallel to the retinal surface. These images are called En face images or C scans. The major benefit of an En face image is that it resembles a fundus photograph but at different layers of the retina. It’s a good learning tool to understand the extent and prognosis of certain diseases.
eOphtha: What is the importance of imaging the choroid on OCT today? Does it substantially aid you in clinical decision making?
Dr. Muna Bhende:The choroid has always been elusive because of its inaccessibility to traditional visualization techniques and ICGA was the only way we could get some reliable information in vivo. Imaging the choroid is probably one of the greatest benefits of the newer OCTs, as it has changed our understanding of not only primary choroidal pathology, whether vascular, neovascular, inflammatory or neoplastic, but has also identified the role of the choroid in diseases previously thought to involve only the retina. This is possible with relative ease in most commercially available machines, and to an unprecedented extent in machines that are based on swept-source technology. Choroidal imaging does aid substantially in diagnosis, treatment monitoring, and prognostication in many situations such as CSCR, VKH, and other posterior segment inflammations, PCV, posterior segment tumors, AMD, and even some inherited retinal disorders such as choroideremia. In fact a lot of our understanding of CSCR and PCV has changed with the identification of the pachychoroid spectrum of disease, purely due to choroidal imaging. It has limited routine use in retinal vascular disorders such as vascular occlusions and PDR and disorders that preferentially involve the inner retina and vitreomacular interface.
IN A NUTSHELL: Choroidal imaging aids substantially in diagnosis, treatment monitoring and prognostication in many situations such as CSCR, VKH and other posterior segment inflammations, PCV, posterior segment tumors, AMD and even some inherited retinal disorders such as choroideremia.
eOphtha: Could you briefly elaborate on the difference in imaging via routine Fundus camera vs Optos vs Clarus vs Multicolor imaging?
Dr. Muna Bhende:The simplest way to describe conventional fundus photography is to call it the technique that gives the “best match” for clinical evaluation, the one closest to reality. So it is invaluable as a record of the posterior segment. However, it does have certain limitations in terms of the extent of the fundus imaged in a single frame (max 500), the need for clear media and a well-dilated pupil to give the best quality image.
The Optos camera is an ultrawide field imaging system, which produces a 200° view of the retina (>80% of the surface area). This utilizes a combined SLO with an ellipsoidal mirror to obtain images of the retinal periphery in a single frame, most importantly in the absence of a fully dilated pupil. This is a two-color image as it uses red and green wavelengths. Additional features include the capability of FFA, ICGA, red reflectance, and green reflectance images. It is particularly useful in non-dilating pupils, lesions of the extreme periphery or multiple quadrants, post-operative documentation of findings, and in children where the examination of the periphery is difficult.
The Clarus is a true color ultrawide field retinal imaging device using 3 LEDs which gives images of 7-micron resolution. A single field is 133o with the auto-merge feature increasing it to 2000. It comes with FAF and FFA capabilities.
Multicolor imaging that is developed on the Heidelberg systems gives a combined pseudo-color image using three different wavelengths including blue (486 nm), green reflectance (518 nm), and infrared (IR) reflectance (815 nm). The laser wavelength enhances each separate layer, with the surface of the retina captured by the short-wavelength (486 nm), the retinal vascular and inner retinal layers by the medium wavelength (518 nm), and the retinal pigment epithelium (RPE) and other deep layers by the long-wavelength (815 nm). The multicolor fundus image is a composite of all three, though each can be viewed separately. It is currently limited to viewing the posterior pole.
IN A NUTSHELL: Fundus photography is a technique that gives the “best match” for clinical evaluation, the one closest to reality, with a 50° view of the retina. Optos camera is an ultrawide field SLO based imaging system, which produces a 200° view of the retina (>80% of the surface area), and can be used in undilated pupils. The Clarus is a true color ultrawide field retinal imaging device using 3 LEDs and a single field of 133o with the auto-merge feature increasing it to 2000. Multicolor imaging systems give a combined pseudo-color image using three different wavelengths including blue (486 nm), green reflectance (518 nm), infrared (IR) reflectance (815 nm), and a composite of all three. It is currently limited to viewing the posterior pole and can be taken in the undilated state.
eOphtha: Do you believe the multicolor and reflectance imaging has an edge over conventional fundus photography?
Dr. Muna Bhende:As mentioned in the previous question, multicolor and reflectance imaging does have its benefits, but I think the role of conventional “true color” fundus image will always be the first option for documentation and to understand and explain fundus pathology. Also, since multicolor imaging is currently available only on select machines, it is unlikely to be accepted by all retina specialists as the imaging modality of choice.
IN A NUTSHELL: Multicolor and reflectance imaging does have its benefits, butthe role of conventional “true color” fundus image will always be the first option for documentation and to understand and explain fundus pathology.
eOphtha: How frequently do you advise FFA in cases of diabetic retinopathy in your routine practice? Does the rule of baseline FFA for all diabetic retinopathy patients apply in routine clinical practice?
Dr. Muna Bhende:FFA is still the gold standard of evaluating a diabetic prior to any treatment especially laser for diabetic macular edema and in cases where one suspects macular ischemia or neovascularization. However there are instances where one is comfortable evaluating the patient only with a fundus photograph, OCT, and OCTA, all of which have the advantage of being non-invasive and repeatable at frequent intervals. In early stages of retinopathy or even advanced cases where the findings are evident and correlate with vision, a simple fundus photograph at baseline will suffice, with an OCT if intravitreal injections are planned for macular edema.
With increasing ease of interpreting OCTA and access to wide-field fundus photographs and OCTA, we are slowly moving away from FFA in many cases of diabetic retinopathy. This is a huge benefit for patients who come alone, are systemically unstable and have renal disease, situations where one would like to avoid intravenous injections of contrast agents. Most guidelines do not advise an FFA at baseline or even prior to treatment for all stages of DR.
IN A NUTSHELL: With increasing ease of interpreting OCTA and access to wide-field fundus photographs and OCTA, we are slowly moving away from FFA in many cases of diabetic retinopathy. Having said that, FFA is still the gold standard of evaluating a diabetic prior to any treatment, especially laser for diabetic macular edema and in cases where one suspects macular ischemia or neovascularization.
eOphtha: Why has PCV suddenly become the newly famed retinal entity discussed everywhere today? Since most of the setups in India do not have a facility of an ICG, how do you diagnose a case of PCV otherwise?
Dr. Muna Bhende:Idiopathic polypoidal choroidal vasculopathy (IPCV) was named in 1990 when it was described as peculiar polypoidal, subretinal, vascular lesions associated with serous and hemorrhagic detachments of the retinal pigment epithelium. Initially the diagnosis was based on clinical suspicion but the gold standard was ICGA, to which access is still very limited. However, with a better understanding of its typical OCT and OCTA findings, along with our ability to image the choroid better, we are making a confirmed diagnosis much more frequently and are also able to attribute differing responses to recommended treatments for classical n-AMD to the associated choroidal changes seen on OCT, such as pachychoroid and PCV. Our understanding of the pachychoroid spectrum has also placed PCV at one end, the other end being CSCR.
Recent studies have compared other investigative modalities with ICGA and have noted that a combination of color fundus photograph and an enhanced depth OCT is able to arrive at a diagnosis of PCV in most cases. The advent of OCTA, especially the swept-source OCTA has further improved our ability to diagnose PCV without ICGA. The current limitations of OCTA make it easier to identify the branching vascular network component than the polyps which may have the flow that is outside the limits of capture by the machine. OCTA is also unable to identify leaking pachyvessels with certainty. All said and done, if there is access to ICGA, it is still the best modality to diagnose and plan treatment of PCV.
IN A NUTSHELL: A combination of colour fundus photograph and an enhanced depth OCT is enough to arrive at a diagnosis of PCV in most cases without ICGA. However, if there is access to ICGA, it is still the best modality to diagnose and plan the treatment of PCV.
eOphtha: How much importance do you attribute to autofluorescence imaging in your regular OPD? Could you briefly elaborate on the difference between green light vs blue light vs infrared autofluorescence?
Dr. Muna Bhende:Autofluorescence ( FAF) provides information regarding the functioning of retinal cells particularly the RPE and the photoreceptors. This facility is available on most fundus cameras, particularly the c- SLO based systems. The photoreceptors shed their damaged outer segments which the RPE ingests through phagocytosis. The molecules are stored in liposomes as lipofuscin. Hyperfluorescence reflects increased lipofuscin accumulation due to photoreceptor damage in the presence of functioning RPE. Hypoflourescence occurs when the RPE cells die and hence lipofuscin is not produced.
Situations, where autofluorescence imaging is useful in OPD:
AMD – to delineate areas of geographic atrophy that are hypofluorescent and also identify hyperfluorescent junctional zones of imminent atrophy that are hyperfluorescent. RPE rips can also be identified easily as sharply demarcated areas of hypofluorescence.
Inherited retinal degenerations and dystrophies. FAF imaging is particularly useful in dystrophies involving the RPE – Best’s vitelliform dystrophy, Stargardts dystrophy, Central areolar choroidal dystrophy, and pattern dystrophy. It is also useful in studying the macular region in Retinitis pigmentosa and LCA where the parafoveal hyper autoflourescent ring constricts over time, indicating the progression of the disease.
CSCR – the area of neurosensory detachment is hyperfluorescent, making it easy to delineate.
Drug toxicity, especially hydroxychloroquine. FAF shows central mottled hypo autofluorescence with a surrounding rim of hyper autofluorescence, due to the RPE loss It is a sensitive indicator of RPE degeneration.
AZOOR – which has the characteristic trizonal pattern of FAF abnormalities.
The wavelengths differ based on the system used. Conventional fundus camera FAF-imaging method uses filters exciting in the green spectrum and recording emission in the yellow-orange spectrum. This technique images fluorescence from the retina and the RPE at the same time and gives a very strong signal. The c-SLO based system from Heidelberg uses blue light and the Optos SLO, green light. These selective isolate signals from a single plane, minimizing incidental fluorescence from other structures like the lens. Greenlight may provide further details about the fovea as blue light absorbs xanthophyll, it may also be beneficial in eyes with cataract as the lens is autofluorescent with blue light. Infrared wavelengths excite molecules other than lipofuscin, most notably melanin, and are useful to study the distribution of melanin within lesions.
IN A NUTSHELL: Autofluorescence provides information regarding the functioning of retinal cells particularly the RPE and the photoreceptors. It is useful in diseases such as AMD, inherited retinal degenerations and dystrophies, CSCR, Drug toxicity, AZOOR etc. The wavelengths differ based on the system used.
eOphtha: With the advent of extensive imaging, telemedicine, and artificial intelligence coming in, do you feel the practice of seeing the patient clinically will fade by the end of this century?
Dr. Muna Bhende:I would say each of these has its own benefits. Imaging has enabled us to understand the disease process better, without having to rely on histopathology or biopsy specimens for a confirmed diagnosis. Non-invasive techniques are repeatable frequently thus helping to assess treatment response as well as look for reasons for lack of response. We have enough examples where the erroneous interpretation of images has led to a wrong diagnosis and thereby wrong treatment being prescribed, so imaging alone will not replace a complete examination.
Telemedicine has the advantage of reaching inaccessible areas and probably the best examples would be the ROP and diabetic retinopathy screening programs and more recently the teleconsultations that increased during the recent lockdown. In the former, the facility is linked to a device operated by a trained technician that can provide reliable information of the retinal findings and a decision regarding the need for further evaluation and treatment can be made and conveyed. In case of teleconsultations from home during the current COVID crisis that caught health systems worldwide unawares, various tools like smartphone “selfies”, WhatsApp and video calls have served the purpose to a great extent, though definite guidelines would need to be followed in future if these have to play a role in the long run.
Artificial intelligence and machine learning in ophthalmology are enabling computer-assisted screening, diagnosis, and prognostication of ophthalmic disease and even systemic diseases. Ophthalmic imaging, and in particular retinal imaging is unique in that it allows direct evaluation of blood vessels, neural tissue, and connective tissue in vivo with excellent image quality and without the need for pathologic specimens. The retinal disease has been most actively researched, in particular diabetic retinopathy, ROP, and AMD, given the fact that retinal photographs and OCTs are standard modalities of investigation with disease-specific findings. AI is unlikely to replace the clinical visit, it is aimed at improving screening if combined with telemedicine, supporting diagnosis by creating algorithms, aiding decision making for treatment, and estimating the progression of the disease. It would be of use in areas where the population is high and resources are low, to rationalize clinic visits and workforce. The flip side is that there may be false negatives (more dangerous) as well as false positives and the need to frequently revisit the existing algorithms based on feedback, eventual blunting of clinical abilities to distinguish subtle differences due to increased reliance on computer-aided diagnosis, and trust issues from the patient and the clinician that would need to be resolved.
To answer the second half of the question, there is still no replacement for humans treating humans! But it is only 2020 and we still have a long way to go.
IN A NUTSHELL: Imaging has enabled us to understand the disease process better, without having to rely on histopathology or biopsy specimens for a confirmed diagnosis. Telemedicine has the advantage of reaching inaccessible areas. AI is mainly used in areas where the population is high and resources are low, to rationalize clinic visits and workforce. Having said all of this, all these modalities have their own flaws, which can be more dangerous ethically, morally, and legally. Hence, a sound clinical examination and judgment is here to stay and absolutely irreplaceable.