From the University of Pittsburgh School of Medicine (TRF, JY, LH), UPMC Eye Center, Pittsburgh Pennsylvania; UCLA Jules Stein Eye Institute (AG, SDS), Los Angeles, California; Duke Eye Center (IS), Durham, North Carolina; and Bascom Palmer Eye Institute (CAP), Miami, Florida.
Drs. Friberg, Gupta, Suner, Puliafito, and Schwartz are or have been on the Scientific Advisory Board for Optos PLC.
Dr. Puliafito did not participate in the editorial review of this manuscript.
Presented in part at the American Academy of Ophthalmology annual meetings, New Orleans, Louisiana, October 23–26, 2004, and November 11–14, 2006, and the Macula Society annual meeting, Key Biscayne, Florida, February 26, 2005.
Address correspondence to Thomas R. Friberg, MD, UPMC Eye Center, 203 Lothrop Street, Suite 824, Pittsburgh, PA 15213.
Diabetic retinopathy is a leading cause of world blindness, making its detection an important priority.1 Recent technological advances, such as the development of non-mydriatic cameras for rapid screening, facilitate the goal of early diagnosis.2 Devices that can display a wide field of view have an additional theoretical advantage and are likely to increase the sensitivity of diabetic retinopathy detection by displaying regions not imaged using more limited fields.3 Furthermore, because fluorescein angiography displays vascular abnormalities in high contrast, intraretinal microvascular abnormalities, retinal ischemia, and proliferative retinopathy are better demonstrated with angiography compared to conventional photography. Hence, fluorescein angiography is widely used once threshold diabetic retinopathy is detected.4
Typically, conventional fundus cameras cover 45° to 60° of field in one exposure. Obtaining extremely wide field images during fluorescein angiography is therefore challenging. Often, an experienced ophthalmic photographer is needed to document the peripheral retina using multiple fields, and these must be quickly acquired to cover the entire transit of dye. These individual fields can later be assembled to create a multi-field montage. However, because the image fields must necessarily be acquired separately, the dye passage within the separate images represents different angiographic time points or phases. The resulting montage thus has little time value because it is a mosaic of different angiographic stages in different regions. Therefore, multiple injections are required to capture the angiogram in all its phases across all subfields if a representative study is required. To image the peripheral fields of the retina, the subject must also alter his or her gaze position, which induces some optical astigmatism and therefore image distortion. In aggregate, such complexities are often overwhelming for the patient and for the photographic staff, so wide field angiography obtained in this way was never practical and was seldom used.
The Optos Panoramic200 prototype device manufactured by Optos Inc. (Dunfermline, Scotland) captures a wide field (200°) in high resolution (2,000 × 2,000 pixels) without requiring pupillary dilation in a single exposure. The commercially available device captures a higher resolution image (3,000 × 3,000 pixels). Montages of multiple images are therefore not needed. The Optos device is based on scanning laser ophthalmoscopy technology but differs by imaging an ultrawide field.5 A proprietary ellipsoidal mirror is incorporated that creates a virtual scanning head within the patient’s eye, behind the pupil. Hence, the laser beam can scan virtually the entire retina and captures a single, high-resolution, 200° digital color image in 0.25 second.
Theoretically, a careful review of ultrawide field images should be more sensitive than conventional fundus imaging in the detection of diabetic retinopathy. Although the Optos Panoramic200 device would likely be valuable during remote screening endeavors, we were especially interested in assessing the entire retinal vasculature in a single image using fluorescein angiography. By adding a blue scanning laser head to the Panoramic200 unit, along with a barrier filter and a fast recycling electronic flash, an ultrawide angiography unit was created (Optos P200A). In contrast to creating a digital montage, the prototype P200A obtains images of the peripheral and central retina simultaneously in each 200° frame as the subject maintains central fixation.
We were interested in comparing the Optos P200A prototype device to conventional digital angiographic cameras with respect to just how much of the field is captured, the image quality, and whether they display capillary dropout and retinal neovascularization. We performed this assessment across three centers in a consecutive series of 10 patients, each of whom had diabetic retinopathy.
Patients and Methods
After obtaining Institutional Review Board approval, we performed angiography on 10 subjects with diabetes mellitus each at three separate centers in an open, nonrandomized pilot trial (The University of Pittsburgh Eye Center, Pittsburgh, Pennsylvania, The Jules Stein Eye Institute, University of California, Los Angeles, California, and the Bascom Palmer Eye Institute, Miami, Florida). Both Optos P200A and conventional digital angiograms were obtained independently for each subject. Subjects all had moderate nonproliferative diabetic retinopathy as assessed by clinical examination. Enrolled patients underwent fluorescein angiography of one eye with the P200A imaging on visit 1, followed by repeat fluorescein angiography with standard digital camera imaging of the same eye 1 week later (Fig. 1). Depending on the center, either a Top-con TRC 501X (Topcon Inc., Paramus, NJ) or a Zeiss RC6000 (Carl Zeiss Meditec, Dublin, CA) was used as the conventional digital camera. Inclusion and exclusion criteria are summarized in Table 1.
Figure 1. Ultrawide Panoramic 200A (Optos Inc., Dunfermline, Scotland) Fluorescein Angiographic Image of an Eye with Diabetic Retinopathy, Peripheral Retinal Ischemia, and Retinal Neovascularization. the Field of View of This Image Is Much Greater than the Field of View of a Typical Standard Digital Image, Which Is Indicated by the Black Circle.
Table 1: Study Inclusion and Exclusion Criteria
Images of both the arteriovenous transit and late phases were then assessed by a masked independent image reviewer with the aid of a grid that was digitally applied over each angiogram. This image grid was a simplification of that published by Shimizu et al.1 and comprised 12 concentric rings (each 1 disc diameter [DD] apart) that were each further subdivided into 8 equal 45° sectors labeled A through H (Fig. 2).
Figure 2. Digital Grids Have Been Superimposed on the Images in Figure 1 to Facilitate the Quantification of the Field of View Differences and Areas of Ischemia and Retinal Neovascularization. the Distance Between Consecutive Rings Is 1 Disc Diameter. the Ultrawide Device Images Significantly More of the Fundus (A) Compared to the Image Obtained by a Conventional Digital Camera (B).
The grid size was carefully normalized for each image so that the optic nerve diameter in the fluorescein frame exactly equaled the spacing between the concentric rings. The center of the image grid was a circle and was designated as ring #1. Adjacent concentric circles (rings) were numbered consecutively outward to the peripheral retina up to ring #12. The center of each grid was digitally placed to coincide with the center of the foveola. The grid thus created a total of 96 possible image sectors (12 concentric circles each divided into 8 sectors).
An unbiased reader (LH) measured the peripheral extent of the angiographic images by simply marking on the grid how many disc diameters (rings) were needed to contain the image in each of the 8 sectors. These numbers were summed and the average represented the mean extent of the image. To measure retinal vasculature ischemia (capillary dropout), sectors where any ischemia was detected were counted individually in the arteriovenous and late phase images. Similarly, the number of sectors harboring any retinal neovascularization were counted for each eye using both the conventional and P200A images.
The image quality was subjectively scored as excellent, fair, or poor by the reader and was logged for each image. The study participants undergoing the angiography were also asked to rank their experience after each angiogram was obtained. They graded the P200A and conventional angiographic method on a scale of 1 to 10, where 1 was an unpleasant, totally dissatisfying experience and 10 indicated total satisfaction. Each subject had had at least one angiogram prior to entry to the study as an entry criterion. All data were entered onto a separate data sheet for each image and the collected data sheets were sent to a masked off-site center so that appropriate statistical tests could be conducted independent of the reader. This trial was completed in accordance with ICH Guidelines on Good Clinical Practice and the Health Insurance Portability & Accountability Act (HIPPA) of 1996.
Extent of the Image Fields
The total image extent for conventional versus P200A systems are compared in Table 2. The P200A understandably yielded a much greater view compared with standard camera images in both the arteriovenous and late phases. The magnitude of the difference was approximately threefold (Table 2).
Table 2: Retinal Image Area
Comparison of the Number of Image Sectors Displaying Neovascularization Lesions
Some subjects had proliferative diabetic retinopathy. The mean number of image sectors displaying neovascular lesions in both the arteriovenous and late phases was significantly greater using the P200A images as opposed to the conventional images (Table 3; Fig. 3).
Table 3: Number of Image Sectors Displaying Detectable Retinal Neovascularization
Figure 3. In an Eye Found to Have Proliferative Diabetic Retinopathy and Retinal Ischemia, the Conventionally Acquired Image (A) Misses Many Areas of Capillary Dropout and Local Retinal Neovascularization Compared to the Ultrawide Image (B ). Such Pathology Is Important to Detect from a Patient Care Standpoint and Often Signals the Need for Panretinal Laser Photocoagulation.
Comparison of the Number of Image Sectors Demonstrating Retinal Ischemia
More ischemic sectors were displayed with the P200A compared to the standard digital camera in both the arteriovenous and late phases (Table 4). The difference was significant (P < .05).
Table 4: Number of Image Sectors Displaying Ischemic Lesions
Comparison of Image Quality
A higher proportion of images taken with the standard digital camera in the arteriovenous and late phases were judged to be excellent (93% and 93%, respectively) compared to the P200A (83% and 90%, respectively). This difference in image quality was most apparent for images taken in the arteriovenous phase (Table 5), was present across all three study centers, and was statistically significant (P < .05).
Table 5: Image Quality
Patient Satisfaction with the Angiographic Method
Patients graded their satisfaction with each method using a scale ranging from 1 (not satisfied) to 10 (very satisfied). No significant difference was seen in the satisfaction indices between the two different methods (Table 6).
Table 6: Patient Satisfaction
Uncontrolled hyperglycemia in patients with diabetes mellitus can lead to various well-known microvascular changes that include the development of basement membrane thickening, pericyte loss, and microaneurysm formation.6 These microvascular changes are related to many pathophysiological alterations, including polyol accumulation, an increase in reactive oxygen species formation, and protein kinase C activation. Long-term microvascular damage results in profound ischemia that leads to the formation of growth factors causing neovascularization.6 Once the cascade of biochemical pathways is activated, the management of the disease becomes more challenging. Hence, early detection and diagnosis of diabetic retinopathy has been emphasized by clinicians.
Standard ophthalmoscopic examination performed by ophthalmologists has only a moderate sensitivity in detecting ischemic and neovascular lesions in diabetic retinopathy and depends on the skill and observational powers of the examiner. The fact that the Early Treatment Diabetic Retinopathy Study defined grading of stereoscopic color fundus photographs in seven standard fields as the standard for detection of diabetic retinopathy is not surprising.7 Although such an approach is both accurate and reproducible, it is labor intensive, requires experienced photographers and readers, and is impractical when dealing with the assessment of many patients in succession. Furthermore, the acquisition of seven photographic fields is not well tolerated by the patients, who must endure multiple exposures and must follow detailed instructions with respect to their gaze positions. In the experience of the authors, many patients do not wish to undergo additional assessments using such an approach. The introduction of wider field devices thus facilitates the more complete detection of diabetic lesions because few exposures are required, the gaze is fixed, and compliance is less of an issue.
With respect to angiography, imaging the peripheral retina in all quadrants simultaneously has heretofore been both impractical and difficult to execute. We found that the P200A system imaged a greater number of ischemic and neovascularization lesions in the arteriovenous transit and late phase of the fluorescein angiography compared to more common digital imaging systems. These differences are most likely due to the capture of a significantly wider field with the Optos angiographic unit, which extended approximately 9 DD from the foveola compared to approximately 3.3 DD for conventional fundus images (Table 2).
The image quality of the P200A prototype was judged to be inferior to the standard digital camera in this trial. This occurred in part because of the distortion produced by the P200A as a result of taking a curved three-dimensional structure (the retina) and stretching it on a two-dimensional screen and less than optimal resolution inherent in the three prototype units. However, the lower image quality did not greatly affect the display of pathological lesions such as capillary dropout and retinal neovascularization, most likely because of the high contrast of the angiography images (wide angle or conventional). Patients did not prefer one examination method over the other, suggesting that the P200A angiography method could substitute for or replace the angiography method done using the standard digital cameras from the patient’s point of view.
The P200A displayed a substantially larger proportion of the retina versus conventional imaging devices when comparing a single image centered at the foveola. Although the resolution of the images from the prototype was not optimal and was judged to be less than that of conventional systems, the P200A nonetheless revealed significantly more neovascular and ischemic lesions. When the first digital image systems were introduced to clinical ophthalmology more than 20 years ago, one of the authors (TRF) commented that the resolution of these early images was inferior to the then-standard film-based systems.8 Predictably, because of the many advantages of digital over film-based images, the digital systems evolved technically and eventually revolutionized both ophthalmic and consumer photographic markets. As observers of this evolution, we believe that as additional technical and software modifications are made to the prototype P200A system, the quality of the images will substantially improve. In the future, such a system may prove indispensable.
- : Kohner EM, Barry PJ. Prevention of blindness in diabetic retinopathy. Diabetologia. 1984;26:173–179. doi:10.1007/BF00252402 [CrossRef]
- : Boucher MC, Gresset JA, Angioi K, Olivier S. Effectiveness and safety of screening for diabetic retinopathy with two nonmydriatic digital images compared with seven standard stereoscopic photographic fields. Can J Ophthalmol. 2003;38:557–568.
- : Friberg TR, Pandya A, Eller AW. Non-mydriatic panoramic fundus imaging using a non-contact scanning laser-based system. Ophthalmic Surg Lasers Imaging. 2003;34:488–497.
- : Ivanisevic M, Stanic R. Importance of fluorescein angiography in the early detection and therapy of diabetic retinopathy. Ophthalmologica. 1990;201:9–13.
- : Manivannan A, Plskova J, Farrow A, Mckay S, Sharp PF, Forrester JV. Ultra-wide-field fluorescein angiography of the ocular fundus. Am J Ophthalmol. 2005;140:525–527. doi:10.1016/j.ajo.2005.02.055 [CrossRef]
- : Ciulla TA, Amador AG, Zinman B. Diabetic retinopathy and diabetic macular edema: pathophysiology, screening, and novel therapies. Diabetes Care. 2003;26:2653–2664. doi:10.2337/diacare.26.9.2653 [CrossRef]
- : Early Treatment Diabetic Retinopathy Study Research Group. Photocoagulation for diabetic macular edema. Early Treatment Diabetic Retinopathy Study Report No. 1. Arch Ophthalmol. 1985;103:1796–1806.
- : Friberg TR, Rehkopf PG, Warnicki JW, Eller AW. Use of directly acquired digital fundus and fluorescein angiographic images in the diagnosis of retinal disease. Retina. 1987;7:246–251. doi:10.1097/00006982-198707040-00010 [CrossRef]
Study Inclusion and Exclusion Criteria
|Inclusion Criteria||Exclusion Criteria|
|Male or female||Known sensitivity or allergy to fluorescein sodium|
|Age > 18 years||Lens opacity preventing high quality digital photography|
|Diabetic retinopathy requiring angiography assessment||Epilepsy (which could be triggered by multiple flashes)|
|History of a previous fluorescein angiogram that had been performed without sequelae||Pregnant or nursing women|
|Diabetic retinopathy classified either as moderate nonproliferative retinopathy, severe nonproliferative retinopathy, or proliferative retinopathya||Poor peripheral veins or presence of hemodialysis shunt|
|Written informed consent||Unable to comply with trial procedures;previously treated with panretinal photocoagulation|
Retinal Image Area
|Characteristic||Panoramic 200A (DD)||Digital Camera (DD)||Difference|
|Maximum possible radial image extent in DD||12||12|
| Mean ± SD||8.7 ± 1.60||3.4 ± 0.76||5.3 ± 1.78|
| Range||3 to 11||2 to 5||−1 to −8|
| 95% CI||8.51, 8.91||3.27, 3.46||5.12, 5.57|
| Mean ± SD||9.0 ± 1.27||3.30 ± 0.79||5.71 ± 1.48|
| Range||5 to 12||2 to 5||1 to 9|
| 95% CI||8.88, 9.20||3.19, 3.39||5.56, 5.93|
Number of Image Sectors Displaying Detectable Retinal Neovascularization
|Sectors Showing Neovascularization|
|Characteristic||Panoramic 200A||Digital Camera||Difference|
| Mean ± SD||1.8 ± 3.16||0.5 ± 1.28||1.33 ± 3.08|
| Range||0 to 14||0 to 5||−3 to 14|
| 95% CI||0.7, 2.9||0.1, 1.0||0.2, 2.4|
| Mean ± SD||2.6 ± 3.35||0.61 ± 1.19||2.0 ± 3.17|
| Range||0 to 10||0 to 4||−2 to 10|
| 95% CI||1.4, 3.8||0.2, 1.0||0.9, 3.2|
Number of Image Sectors Displaying Ischemic Lesions
|Characteristic||Panoramic 200A||Digital Camera||Difference|
| Mean ± SD||16.9 ± 15.04||3.4 ± 4.26||13.5 ± 13.04|
| Range||0 to 48||0 to 13||−3 to 36|
| 95% CI||11.5, 22.3||1.9, 5.0||8.8, 18.1|
| Mean ± SD||17.6 ± 16.77||3.1 ± 3.36||14.5 ± 14.89|
| Range||0 to 50||0 to 9||−3 to 41|
| 95% CI||11.6, 23.6||1.9, 4.3||9.1, 19.8|
|Device||Grade||Center 1||Center 2||Center 3||Overall|
| Arteriovenous phase||grade 1||0%||0%||0%||0%|
| Late phase||grade 1||0%||0%||0%||0%|
| Arteriovenous phase||grade 1||0%||0%||0%||0%|
| Late phase||grade 1||0%||0%||0%||0%|
|Patient Satisfaction Score||Panoramic 200A||Digital Camera||Overall|
|No. (%)||30 (94)||30 (100)||60 (97)|
|Mean ± SD||8.6 ± 2.33||9.1 ± 1.44||8.9 ± 1.93|
|Range||1 to 10||5 to 10||1 to 10|
|95% CI||7.8, 9.4||8.6, 9.6||8.4, 9.3|