Fundusscreening durch Assistenzpersonal

Abstract

Ocular fundus photography allows detection of both ocular and systemic diseases. This study investigated the efficacy of a broad screening in a department of internal medicine using nonmydriatic digital fundus photography. For 8 weeks a medical technician was trained in using the camera as well as interpreting the photographs. The medical technician and an ophthalmologist evaluated the fundus photographs separately by using a self-developed questionnaire. The fundus camera was user-friendly and after several weeks of adjustment and practical application the medical technician was able to detect the majority of pathological fundus photographs. Out of 218 patients examined 148 (68%) were identified as pathological by the medical technician and 163 (75%) by the ophthalmologist (p = 0.0003). The medical technician missed 15 (7%) patients. Furthermore the diagnoses made by the medical technician were faulty. In summary an ophthalmological screening by a medical technician is feasible but the diagnosis still remains the responsibility of ophthalmologists. Such a compromise could facilitate the examination of a large number of patients and disclose previously unrecognized diseases.

Diabetic Retinopathy Screening Ratio Is Improved When Using a Digital, Nonmydriatic Fundus Camera Onsite in a Diabetes Outpatient Clinic

“…One might argue that for diabetic retinopathy screening an ophthalmologist is essential and digital fundus imaging alone might not be sufficient. However, recent studies comparing nonophthalmologists with ophthalmologists for the diagnosis of DR have shown similar accuracy in the detection of changes of the fundus even with paramedical staff performing the screening procedure [ 25 , 38 – 40 ]. This was also the case in the study presented here.…”

“…Evaluation of the fundus photography by a reference ophthalmologist revealed that no change of diagnoses of DR categories was necessary (data not shown). Furthermore, several studies have evaluated the sensitivity and specificity of digital fundus screening [ 20 , 41 ], showing that instant fundus screening by trained nonophthalmologists was accurate and safe in the detection of DR [ 25 , 42 ]. However, since the number of detected DRs in this study is clearly higher than preexistent DRs, it seems unlikely that a remarkable proportion of patients with DR might have been missed in the study although this cannot be fully excluded, since stereoscopic colour fundus photographs as gold standard were not used in this evaluation.…”

“…Before commencing the study, the medical technician was instructed in using the camera and interpreting the retinal photographs, until he felt safe on carrying out the examination. Studies verify that this way of screening correlates with a high level of accuracy [ 24 , 25 ]. If the original image was judged unsatisfactory by the photographer, the image acquisition process was repeated.…”

“…In total the examination process did not exceed five minutes of time. The trained technician and an ophthalmologist evaluated the fundus photographs separately by using a self-developed questionnaire as published previously [ 25 ].…”

The relationship between Fully Connected Layers and number of classes for the analysis of retinal images

“…Fundus screening allows detection of both ocular and systemic diseases that is diabetes, glaucoma, cataract, age-related macular degeneration (AMD) and other causes [11]. In this paper, data sets consisting of retinal images labelled as normal, cataracts, age-related macular degeneration and myopia are used to understand the effect of increasing the number of fully-connected layers in a Convolutional Neural Network.…”