RVPC is a member of the Scientific Advisory Board for Visunex Medical Systems (Fremont, CA). No conflicting relationship exists for any of the other authors.

Telemedicine programs provide an affordable method to screen for eye conditions in resource limited settings, but are impeded by costs of expensive imaging systems.

We performed a prospective pilot study at the Illinois Eye and Ear Infirmary to evaluate the accuracy and reliability of detecting retinal pathology using the Forus 3nethra (Forus) fundus camera compared to ultra wide-field (UWF) imaging with Optos 200Tx. Images were compared against clinical diagnosis by ophthalmoscopy as the reference standard.

Patients underwent mydriatic imaging with Optos and Forus. Three double-blinded graders independently evaluated Forus (45 degree, central, nasal, superior, and inferior views) and Optos (200 degree) images (Fig. 1) for the presence or absence of pathology, image clarity, and specific clinical diagnoses of diabetic retinopathy, choroidal lesions, or uveitis. Graders were asked to choose a diagnosis, rate confidence level in the diagnosis and determine if referral was needed. Responses were captured via a closed-ended survey (Qualtrics).

Fig. 1
figure 1

Fundus images of proliferative diabetic retinopathy OS. a Forus. b Optos

Full size image

35 eyes of 18 patients were included. The accuracy of detecting any ocular pathology was similar between the Forus and Optos images (aggregate calculation of 3 graders): sensitivity 71% vs. 77% (p = 0.60); specificity 43% vs. 48%, (p = 0.85). Image quality results are summarized in Table 1. There was greater sensitivity for detection of choroidal lesions for Optos compared to Forus (93 vs. 33.3%) but similar sensitivity for uveitis (66.3 vs. 100%) and diabetic retinopathy (67 vs. 75%).

Table 1 Comparison of image quality between Forus and Optos
Full size table

Inter-grader agreement was moderate among graders for both Forus and Optos with kappa statistics of 0.50 and 0.40, respectively. Rate of referral for clinical exam based on images were similar among graders at 74 and 76% for Forus and Optos, respectively.

This pilot study showed similar sensitivity and specificity for detecting any pathology with the Forus camera compared to UWF imaging. Forus’ overall sensitivity of 71% in detecting any ocular pathology falls within the sensitivity range of 71–97.9% [1, 2] to detect referral-requiring pathology. Referral rates for clinical examination were similar between both modalities. The Forus images were graded as good or acceptable more often than UWF images, consistent with newer fundus cameras found to provide similar image quality to their standard counterparts [3]. Forus’ moderate inter-grader agreement (kappa 0.5) falls below that of other nonmydriatic cameras with good inter-grader agreement (kappa 0.64–0.77) [1,2,3,4]. This difference is likely due to the variability in detecting pathology per grader.

Advantages of Forus when compared to UWF imaging include affordable cost of $8000–$10,000 [5] compared to the average fundus camera at $20,000-$50,000 [3], convenience, portability, and ease of use in allowing non-ophthalmologists to capture images for viewing and grading by ophthalmologists.

Limitations of this study included its limited sample size, variability in graders, dilation of patients, pixilation of images, and field of view.

In conclusion, the Forus fundus camera demonstrates similar accuracy and reliability with UWF imaging in detecting ocular pathology. Further data must be collected in order to validate Forus’ screening capabilities for specific ocular pathologies and use in telemedicine.

Disclaimer

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH, NSF, or RPB. The sponsors and funding organizations had no role in the design or conduct of this research.