Eyeworld

C M Y CM MY CY CMY K 26 Ophthalmology Business • March 2018 clinical judgment. The photos are not as good as seeing someone in person," Dr. Woodward said. Christopher Rapuano, MD, chief of the cornea service, Wills Eye Hospital, Philadelphia, agreed. "A finite set of pictures doesn't compare to examining someone at a slit lamp where you can adjust the slit beam width, height, and angle at will," he said. Dr. Woodward is continuing research in the area of low cost tech- nologies that could be used effective- ly by general ophthalmologists and primary care physicians to improve access to and quality of care for pa- tients with corneal diseases. Dr. Rapuano thinks the technol- ogy for cornea telemedicine is not too far away in the future. He also thinks that having a trained image interpreter is key. He said teaching first-year residents how to use the slit lamp is like teaching someone to drive a car. With experience and training, he wants those residents to become Indy 500 race car drivers in their detection of cornea disease. Dr. Park also sees a brighter future for corneal telemedicine. "In- creased magnification, 3-D imaging, and ease of use will help improve the use of telemedicine. On the clinical side, establishing standardized guide- lines for evaluating these images will make this a more robust resource," she said. OB Reference 1. Woodward MA, et al. Teleophthalmic approach for detection of corneal diseases: accuracy and reliability. Cornea. 2017;36: 1159–1165. Editors' note: The physicians have no financial interests related to their comments. Contact information Park: lp2736@cumc.columbia.edu Rapuano: cjrapuano@willseye.org Woodward: mariawoo@med.umich.edu The future of corneal telemedicine With a sensitivity of 80% thought to be the standard for the accuracy of telemedicine screening for diabetic retinopathy and other eye diseas- es, this method did not meet that standard overall, the researchers concluded. "Our findings indicate that the interpretation of images from portable cameras had high specificity but low sensitivity," the researchers wrote. Dr. Woodward said that while it was encouraging that the cameras could reasonably detect ulcers and pterygia, it was disappointing that they could not accurately detect all other corneal pathology. Lisa Park, MD, associate pro- fessor of ophthalmology, Columbia University Medical Center, New York, was surprised that the corneal findings were not easily detected, particularly corneal abrasions. "Pa- tients commonly will present to an emergency room or urgent care clinic for this type of injury, and telemed- icine would be quite useful in this case, but the technology does not yet appear to be useful for this diagno- sis," she said. The problems with corneal pathology detection occurred even with the use of an ophthalmic pho- tographer who knows how to obtain relevant images of the eye, she said. One drawback for the portable camera technology is that corneal specialists are accustomed to high quality slit lamp photographs, Dr. Woodward said. That's probably yet another reason that the photos analyzed in the study did not seem to help the graders effectively find pathologies. The study is also a reminder of a situation in which many ophthal- mologists find themselves—when family, friends, or patients send them smartphone photos of eye problems. "This grounds us in the fact of using Photographers took seven images of each eye: straight gaze as well as right, left, up, down, with eyelids closed, and straight gaze with cobalt blue light after instillation of topical fluorescein dye. An ophthalmic pho- tographer took the photos. Three corneal specialists inter- preted the images for pathology. The graders judged the presence or absence of disease in each eye, and they had an additional category for "suspicious for pathology present." Graders determined both the severity of disease as well as disease location. They graded each photo quality from 1 (lowest) to 9 (highest). The study included 198 eyes. Per the clinical diagnosis, 30% of eyes had corneal scars, 17% had ulcers, 7% had abrasions, and 5% had ptery- gia. Forty-one percent of the eyes had no corneal disease. There was a sensitivity of 54% to 71% to detect anterior segment pathology for the iTouch photos and 66% to 75% for the VersaCam across graders. Specificity ranged from 82% to 96% for the iTouch and 91% to 98% for the VersaCam. However, the sensitivity and specificity differed based on the corneal pathology in- volved: the sensitivity was especially low for corneal scars (29% to 54% for the iTouch and 42% to 58% for the VersaCam) compared with corneal ulcers and pterygia. "For both cameras and all grad- ers, each grader had significantly greater sensitivity to detect corneal ulcers than to detect corneal scars for each of the cameras," the researchers wrote. When grading image quality, two of the three graders rated signifi- cantly lower image quality from the iTouch compared with the VersaCam (both P<0.004). The graders also found that having additional images was helpful to detect the presence or absence of pathology. Overall, image quality differences were not consis- tent among diagnoses, camera types, or graders. continued from page 25

• Blank Page
• Blank Page (1)