Curr Diab Rep (2014) 14:514 DOI 10.1007/s11892-014-0514-0
MICROVASCULAR COMPLICATIONS—RETINOPATHY (JK SUN, SECTION EDITOR)
Ultra Widefield Fundus Imaging for Diabetic Retinopathy Szilárd Kiss & Thomas L. Berenberg
# Springer Science+Business Media New York 2014
Abstract For decades, the standard method for screening and grading severity of diabetic retinal disease has relied upon a montage of photographs using normal angle fundus cameras. With the development of ultrawide field (UWF) fundus imaging, more of the retina can be imaged with fewer pictures, less dependence on photographer skill, and, often, greater ease on the patient. Recent studies have shown comparability between traditional and UWF imaging for standard grading of diabetic retinopathy. Moreover, UWF images can detect peripheral pathology not typically seen in standard photographs, which may enlighten our understanding of disease severity and suggest new indications for treatment.
sites and 7SF photographs or 7 modified fields based on the ETDRS protocol continue to serve as the standard for DR grading in large multicenter trials. In recent years, wide field imaging and ultrawide field imaging have been developed that can image larger degrees of the retina in less time with fewer flashes per eye. As of yet, no major clinical trial in diabetic eye disease has replaced the standard photographs with UWF images. This article will review the evolution of UWF imaging and its use in diabetic eye disease. In addition, existing studies evaluating the utility and accuracy of UWF imaging in diabetic eye disease will be discussed and future directions will be considered.
Keywords Diabetic retinopathy . Fundus photography . Wide field imaging . Ultrawide field imaging . Screening . Grading
Standard Fundus Camera
Introduction Fundus photography has been the basis for screening and grading of severity of diabetic eye disease since the seminal clinical trials were performed over 30 years ago, including the Diabetic Retinopathy Study (DRS) and Early Treatment of Diabetic Retinopathy Study (ETDRS) [1, 2]. These photographs were acquired in a standard protocol referred to as ETDRS-protocol 7 standard field (7SF) stereoscopic photographs and have served as the gold standard for detecting and classifying diabetic retinopathy (DR). This protocol allowed for standardization of DR assessment across multiple study This article is part of the Topical Collection on Microvascular Complications—Retinopathy S. Kiss (*) : T. L. Berenberg Weill Cornell Medical College, 1305 York Avenue, 11th Floor, New York, NY 10021, USA e-mail: [email protected]
ETDRS photographs use 35 mm slide photographs that capture 30° retinal fields after pupil dilation (each field covers approximately 5 % of total retina) [2]. At least 7 paired stereoscopic images for the 7-standard fields and 1 paired anterior segment photograph were taken per eye per study participant. Over recent years, the imaging technique has transitioned from film to digital, while maintaining comparable agreement [3–5]. Together the 7SF cover approximately 90° of the retina (approximately 30 % of the entire retinal surface) with 3 overlapping fields capturing the optic nerve, central macula, and temporal macula (1 M, 2, and 3 M, respectively), and 4 fields capturing the superotemporal, inferotemporal, superonasal, and inferonasal quadrants (4–7, respectively) [6••]. Several modifications to the original traditional 7SF have been evaluated, validated, and are widely used in clinical practice. These modifications include substitution of uncompressed, digital images and the acquisition of fewer 45° to 60° wide angle images compared with the standard ETDRS 30° to 35° 7 fields. From either the 7SF or the modified 7SF images, the risk of DR progression and vision
514, Page 2 of 7
Curr Diab Rep (2014) 14:514
loss can be identified [7]. Obtaining these stereoscopic images requires wide pupillary dilation, at least 16 images/flashes per eye, good patient fixation and cooperation, and skilled photographers. More recently, the Diabetic Retinopathy Clinical Research Network (DRCR.net) has utilized 4 wide-angle stereoscopic fields with a camera capturing 45°–60° views, but has not replaced them with UWF images [4]. In addition to the acquisition-related disadvantages of these 4- and 7-field imaging methods, there are large areas of retina in the midperiphery and far periphery not visualized by the standard photographs (Fig. 1).
Table 1 Comparison of imaging platforms
A Review of Wide-Field Imaging
Adding a contact lens, adjusting the location and mechanism of the light source, and utilizing a confocal scanning laser as opposed to traditional photography have all been used to increase the field of view. In 1975, Pomerantzeff reported a contact lens based system that was able to image up to 148° of the retina [10]. Others have reported using wide-angle lenses to image up to 160°of the retinal periphery [11–13]. In 1997, the Retcam (Clarity Medical System, Inc., Pleasanton, CA), became commercially available. Using a contact lens and fiber optic cable light source, 120° of the retina can be visualized [14]. This technology has been used primarily in the pediatric population, both because of its ease of use in sedated pediatric patients but also because of its poor image quality in adults with any media opacities. The Panoret-1000 was a hand-held digital camera (Medibell Medical Vision Technologies, Haifa, Israel) that imaged 100°of the retina. This was based on the Pomerantzeff concept of trans-scleral illumination [15].
In 1926, the Carl Zeiss Company produced the first commercially available fundus camera, which provided a 20° field of view [8]. The same company later produced a camera with a 30° field of view, which then became the standard. Fundus cameras with image angles up to 50°emerged and have been used widely in clinical practice. Subsequently, any camera with a field of view larger than 30° to 50° has been termed “wide angle” [9•]. Table 1 displays the fields of view of various imaging platforms. The field of view of a fundus camera is a function of its optical system’s focal length and the dimension of the image being photographed. The power of the lens system and field of view are directly correlated, while the focal length of the lens system and the field of view are indirectly correlated. Wide field imaging therefore expands the field of view by either increasing the lens power or decreasing the focal length. Fig. 1 An UWF-FA image of a diabetic patient reveals areas of neovascularization (yellow arrows) and nonperfusion (red arrows) outside the area visualized by the ETDRS 7-fields (blue circles)
Type
Contact vs noncontact
Maximum field of view (degrees)
Standard small Angle Pomerantzeff Retcam Panoret 1000
Noncontact Contact Contact Contact
60 148 120 130
Optos Staurenghi Heidelberg
Noncontact Contact Noncontact
200 150 150
Curr Diab Rep (2014) 14:514
In 2000, the Optos camera (Optos PLC, Dunfermline, UK) became commercially available. Utilizing a confocal scanning laser ophthalmoscope (SLO), the camera creates an image of the retinal periphery with a single capture, without the need for a contact lens or pupillary dilation. To overcome problems imaging the retinal periphery with SLO, the Optos system uses the optics of an ellipsoid mirror. There are 2 focal points, 1 through which the laser is directed and the other located inside the patient’s eye, thus, allowing for large scan angles of up to 200° (approximately 80 % of the retina) in a single shot [16]. Optos technology utilizes green (532 nm) and red (633 nm) laser wavelengths that can be captured simultaneously to produce pseudocolor images of the retina. In addition, autofluorescence, red-free images, green-free images, fluorescein angiography, and the option of a platform for wide field indocyanine green angiography are available. Disadvantages include low posterior pole resolution, unrealistic color rendering, and peripheral image distortion. In 2005, Staurenghi et al reported the use of an integrated, multi-element contact lens-based system, which is compatible with a confocal SLO-based imaging platform (Ocular Staurenghi 230 SLO Retina Lens; Ocular Instruments Inc, Bellevue, WA) [17]. Achieving up to a 150°field of view, this lens system consists of 2 biconvex aspheric lenses and a 2-element convex-concave contact lens. It can be used for fluorescein and indocyanine green angiography. More recently, Heidelberg Engineering (Heidelberg, Germany) has developed a noncontact ultrawide field angiography module for the Spectralis and Heidelberg Retina Angiograph (HRA 2) [18]. The noncontact lens attaches to the camera head to provide high-contrast, undistorted, and evenly illuminated images with a wide field of view. Previously, the Spectralis provided a 25° and 35° field of view with the possibility of a 55° noncontact lens attachment or the addition of the Staurenghi contact lens to provide up to a 150° field of view for angiography.
Studies Ultrawide field images have been used to capture many types of nondiabetic ocular pathology, including retinal vein occlusions [19, 20], choroidal pigmented lesions and choroidal masses [15, 21, 22], cytomegalovirus retinitis [23], sickle cell retinopathy [24], uveitis [25], choroidal dystrophies [26], retinal detachment [27, 28], congenital hypertrophy of the retinal pigment epithelium [29, 30], choroidal detachment [31], giant retinal tear [32], trauma-related injuries [33], and pediatric retinal disease [34–36]. From the beginning of ultrawide field imaging, the potential value for screening, grading, and management of diabetic retinopathy has been recognized [16]. Even prior to our ability
Page 3 of 7, 514
to reliably and easily image the peripheral retina, participants in the 1968 Airlie House Symposium recognized the importance of the periphery in diabetic retinopathy. More recently, Friberg et al suggested that wide field imaging could be used to study the relationship between peripheral capillary nonperfusion and the development of neovascularization, a precursor to proliferative diabetic retinopathy (PDR).
Comparing Wide Field with Standard In 2003 Friberg et al compared UWF images with the Optos nonmydriatic 200° system with clinical examination, finding moderate specificity and sensitivity for the detection of DR of 76 % each [16]. There have been a number of small studies comparing Optos images to clinical exam and ETDRS-protocol 7SF photographs for the evaluation of DR severity. One study comparing grading of DR severity using nonmydriatic Optomap images vs dilated clinical funduscopic examination in 51 eyes of 51 diabetic patients found good levels of agreement between the 2 modalities, although 9.8 % of the images were ungradable [37]. A sensitivity of 94 % and specificity of 100 % were found for graders’ assessment of more than mild nonproliferative DR (NPDR) on the UWF images. Assessment of clinically significant macular edema (CSME) on UWF images revealed sensitivities of 89 %–93 % and specificities of 72 %–89 %. Another study compared 200° images with ETDRS standard photographs in 66 eyes of 34 patients. In the 48 sets that could be graded for both ETDRS standard and UWF images, kappas of 0.70 and 0.66 were obtained for agreement of DR severity level and 0.68 and 0.74 for diabetic macular edema (DME) severity [38]. Yet another study also found good agreement between UWF and ETDRS standard grading for both DR and for DME severity (Kappa=0.79, 0.77 for DR and 0.73, 0.77 for CDME) [39]. At a different site, an imaging validation study was conducted to compare nonmydriatic Optomap UWF images with mydriatic ETDRS protocol 30° 7SF stereoscopic fundus film photographs for the grading of DR and DME [40]. Subjects underwent nonmydriatic 100° and 200° imaging with Optos, dilated ETDRS photography and dilated funduscopic examination by a masked retina specialist. Images of 200 eyes of 103 patients with type 1 or 2 diabetes were evaluated for the presence or absence of DR and further categorized in to mild, moderate, severe, or very severe NPDR, and PDR. In the ETDRS 7SF photographs 0.5 % were deemed ungradable. The presence of DME was also evaluated and 5.5 % of images were considered ungradable. Exact agreement of DR severity grading between 100° images and ETDRS photographs occurred in 84 % with agreement within 1 DR severity level in 91 % (KW=0.85, K=0.79). Optomap images
514, Page 4 of 7
Curr Diab Rep (2014) 14:514
exactly matched DR grading by clinical exam in 70 % and were within 1 DR severity level in 93 % (KW=0.77, K = 0.61). Exact agreement with ETDRS photographs for DME graded on a 3-category scale (No DME, DME
MICROVASCULAR COMPLICATIONS—RETINOPATHY (JK SUN, SECTION EDITOR)
Ultra Widefield Fundus Imaging for Diabetic Retinopathy Szilárd Kiss & Thomas L. Berenberg
# Springer Science+Business Media New York 2014
Abstract For decades, the standard method for screening and grading severity of diabetic retinal disease has relied upon a montage of photographs using normal angle fundus cameras. With the development of ultrawide field (UWF) fundus imaging, more of the retina can be imaged with fewer pictures, less dependence on photographer skill, and, often, greater ease on the patient. Recent studies have shown comparability between traditional and UWF imaging for standard grading of diabetic retinopathy. Moreover, UWF images can detect peripheral pathology not typically seen in standard photographs, which may enlighten our understanding of disease severity and suggest new indications for treatment.
sites and 7SF photographs or 7 modified fields based on the ETDRS protocol continue to serve as the standard for DR grading in large multicenter trials. In recent years, wide field imaging and ultrawide field imaging have been developed that can image larger degrees of the retina in less time with fewer flashes per eye. As of yet, no major clinical trial in diabetic eye disease has replaced the standard photographs with UWF images. This article will review the evolution of UWF imaging and its use in diabetic eye disease. In addition, existing studies evaluating the utility and accuracy of UWF imaging in diabetic eye disease will be discussed and future directions will be considered.
Keywords Diabetic retinopathy . Fundus photography . Wide field imaging . Ultrawide field imaging . Screening . Grading
Standard Fundus Camera
Introduction Fundus photography has been the basis for screening and grading of severity of diabetic eye disease since the seminal clinical trials were performed over 30 years ago, including the Diabetic Retinopathy Study (DRS) and Early Treatment of Diabetic Retinopathy Study (ETDRS) [1, 2]. These photographs were acquired in a standard protocol referred to as ETDRS-protocol 7 standard field (7SF) stereoscopic photographs and have served as the gold standard for detecting and classifying diabetic retinopathy (DR). This protocol allowed for standardization of DR assessment across multiple study This article is part of the Topical Collection on Microvascular Complications—Retinopathy S. Kiss (*) : T. L. Berenberg Weill Cornell Medical College, 1305 York Avenue, 11th Floor, New York, NY 10021, USA e-mail: [email protected]
ETDRS photographs use 35 mm slide photographs that capture 30° retinal fields after pupil dilation (each field covers approximately 5 % of total retina) [2]. At least 7 paired stereoscopic images for the 7-standard fields and 1 paired anterior segment photograph were taken per eye per study participant. Over recent years, the imaging technique has transitioned from film to digital, while maintaining comparable agreement [3–5]. Together the 7SF cover approximately 90° of the retina (approximately 30 % of the entire retinal surface) with 3 overlapping fields capturing the optic nerve, central macula, and temporal macula (1 M, 2, and 3 M, respectively), and 4 fields capturing the superotemporal, inferotemporal, superonasal, and inferonasal quadrants (4–7, respectively) [6••]. Several modifications to the original traditional 7SF have been evaluated, validated, and are widely used in clinical practice. These modifications include substitution of uncompressed, digital images and the acquisition of fewer 45° to 60° wide angle images compared with the standard ETDRS 30° to 35° 7 fields. From either the 7SF or the modified 7SF images, the risk of DR progression and vision
514, Page 2 of 7
Curr Diab Rep (2014) 14:514
loss can be identified [7]. Obtaining these stereoscopic images requires wide pupillary dilation, at least 16 images/flashes per eye, good patient fixation and cooperation, and skilled photographers. More recently, the Diabetic Retinopathy Clinical Research Network (DRCR.net) has utilized 4 wide-angle stereoscopic fields with a camera capturing 45°–60° views, but has not replaced them with UWF images [4]. In addition to the acquisition-related disadvantages of these 4- and 7-field imaging methods, there are large areas of retina in the midperiphery and far periphery not visualized by the standard photographs (Fig. 1).
Table 1 Comparison of imaging platforms
A Review of Wide-Field Imaging
Adding a contact lens, adjusting the location and mechanism of the light source, and utilizing a confocal scanning laser as opposed to traditional photography have all been used to increase the field of view. In 1975, Pomerantzeff reported a contact lens based system that was able to image up to 148° of the retina [10]. Others have reported using wide-angle lenses to image up to 160°of the retinal periphery [11–13]. In 1997, the Retcam (Clarity Medical System, Inc., Pleasanton, CA), became commercially available. Using a contact lens and fiber optic cable light source, 120° of the retina can be visualized [14]. This technology has been used primarily in the pediatric population, both because of its ease of use in sedated pediatric patients but also because of its poor image quality in adults with any media opacities. The Panoret-1000 was a hand-held digital camera (Medibell Medical Vision Technologies, Haifa, Israel) that imaged 100°of the retina. This was based on the Pomerantzeff concept of trans-scleral illumination [15].
In 1926, the Carl Zeiss Company produced the first commercially available fundus camera, which provided a 20° field of view [8]. The same company later produced a camera with a 30° field of view, which then became the standard. Fundus cameras with image angles up to 50°emerged and have been used widely in clinical practice. Subsequently, any camera with a field of view larger than 30° to 50° has been termed “wide angle” [9•]. Table 1 displays the fields of view of various imaging platforms. The field of view of a fundus camera is a function of its optical system’s focal length and the dimension of the image being photographed. The power of the lens system and field of view are directly correlated, while the focal length of the lens system and the field of view are indirectly correlated. Wide field imaging therefore expands the field of view by either increasing the lens power or decreasing the focal length. Fig. 1 An UWF-FA image of a diabetic patient reveals areas of neovascularization (yellow arrows) and nonperfusion (red arrows) outside the area visualized by the ETDRS 7-fields (blue circles)
Type
Contact vs noncontact
Maximum field of view (degrees)
Standard small Angle Pomerantzeff Retcam Panoret 1000
Noncontact Contact Contact Contact
60 148 120 130
Optos Staurenghi Heidelberg
Noncontact Contact Noncontact
200 150 150
Curr Diab Rep (2014) 14:514
In 2000, the Optos camera (Optos PLC, Dunfermline, UK) became commercially available. Utilizing a confocal scanning laser ophthalmoscope (SLO), the camera creates an image of the retinal periphery with a single capture, without the need for a contact lens or pupillary dilation. To overcome problems imaging the retinal periphery with SLO, the Optos system uses the optics of an ellipsoid mirror. There are 2 focal points, 1 through which the laser is directed and the other located inside the patient’s eye, thus, allowing for large scan angles of up to 200° (approximately 80 % of the retina) in a single shot [16]. Optos technology utilizes green (532 nm) and red (633 nm) laser wavelengths that can be captured simultaneously to produce pseudocolor images of the retina. In addition, autofluorescence, red-free images, green-free images, fluorescein angiography, and the option of a platform for wide field indocyanine green angiography are available. Disadvantages include low posterior pole resolution, unrealistic color rendering, and peripheral image distortion. In 2005, Staurenghi et al reported the use of an integrated, multi-element contact lens-based system, which is compatible with a confocal SLO-based imaging platform (Ocular Staurenghi 230 SLO Retina Lens; Ocular Instruments Inc, Bellevue, WA) [17]. Achieving up to a 150°field of view, this lens system consists of 2 biconvex aspheric lenses and a 2-element convex-concave contact lens. It can be used for fluorescein and indocyanine green angiography. More recently, Heidelberg Engineering (Heidelberg, Germany) has developed a noncontact ultrawide field angiography module for the Spectralis and Heidelberg Retina Angiograph (HRA 2) [18]. The noncontact lens attaches to the camera head to provide high-contrast, undistorted, and evenly illuminated images with a wide field of view. Previously, the Spectralis provided a 25° and 35° field of view with the possibility of a 55° noncontact lens attachment or the addition of the Staurenghi contact lens to provide up to a 150° field of view for angiography.
Studies Ultrawide field images have been used to capture many types of nondiabetic ocular pathology, including retinal vein occlusions [19, 20], choroidal pigmented lesions and choroidal masses [15, 21, 22], cytomegalovirus retinitis [23], sickle cell retinopathy [24], uveitis [25], choroidal dystrophies [26], retinal detachment [27, 28], congenital hypertrophy of the retinal pigment epithelium [29, 30], choroidal detachment [31], giant retinal tear [32], trauma-related injuries [33], and pediatric retinal disease [34–36]. From the beginning of ultrawide field imaging, the potential value for screening, grading, and management of diabetic retinopathy has been recognized [16]. Even prior to our ability
Page 3 of 7, 514
to reliably and easily image the peripheral retina, participants in the 1968 Airlie House Symposium recognized the importance of the periphery in diabetic retinopathy. More recently, Friberg et al suggested that wide field imaging could be used to study the relationship between peripheral capillary nonperfusion and the development of neovascularization, a precursor to proliferative diabetic retinopathy (PDR).
Comparing Wide Field with Standard In 2003 Friberg et al compared UWF images with the Optos nonmydriatic 200° system with clinical examination, finding moderate specificity and sensitivity for the detection of DR of 76 % each [16]. There have been a number of small studies comparing Optos images to clinical exam and ETDRS-protocol 7SF photographs for the evaluation of DR severity. One study comparing grading of DR severity using nonmydriatic Optomap images vs dilated clinical funduscopic examination in 51 eyes of 51 diabetic patients found good levels of agreement between the 2 modalities, although 9.8 % of the images were ungradable [37]. A sensitivity of 94 % and specificity of 100 % were found for graders’ assessment of more than mild nonproliferative DR (NPDR) on the UWF images. Assessment of clinically significant macular edema (CSME) on UWF images revealed sensitivities of 89 %–93 % and specificities of 72 %–89 %. Another study compared 200° images with ETDRS standard photographs in 66 eyes of 34 patients. In the 48 sets that could be graded for both ETDRS standard and UWF images, kappas of 0.70 and 0.66 were obtained for agreement of DR severity level and 0.68 and 0.74 for diabetic macular edema (DME) severity [38]. Yet another study also found good agreement between UWF and ETDRS standard grading for both DR and for DME severity (Kappa=0.79, 0.77 for DR and 0.73, 0.77 for CDME) [39]. At a different site, an imaging validation study was conducted to compare nonmydriatic Optomap UWF images with mydriatic ETDRS protocol 30° 7SF stereoscopic fundus film photographs for the grading of DR and DME [40]. Subjects underwent nonmydriatic 100° and 200° imaging with Optos, dilated ETDRS photography and dilated funduscopic examination by a masked retina specialist. Images of 200 eyes of 103 patients with type 1 or 2 diabetes were evaluated for the presence or absence of DR and further categorized in to mild, moderate, severe, or very severe NPDR, and PDR. In the ETDRS 7SF photographs 0.5 % were deemed ungradable. The presence of DME was also evaluated and 5.5 % of images were considered ungradable. Exact agreement of DR severity grading between 100° images and ETDRS photographs occurred in 84 % with agreement within 1 DR severity level in 91 % (KW=0.85, K=0.79). Optomap images
514, Page 4 of 7
Curr Diab Rep (2014) 14:514
exactly matched DR grading by clinical exam in 70 % and were within 1 DR severity level in 93 % (KW=0.77, K = 0.61). Exact agreement with ETDRS photographs for DME graded on a 3-category scale (No DME, DME
