Imaging of the eye using smartphones has become increasingly popular and allows for an inexpensive and mobile fundus examination and documentation [1,2]. Although initially the lens needed to be held in the optical path manually [1-3], different proprietary adapters are available now [4-7]. Smartphone-based fundus imaging can be performed based on both direct [4,5] and indirect ophthalmoscopy [6,8-10]. Given the low costs and great mobility, smartphone-based fundus imaging has the potential to revolutionize eye care, especially in lower-resource settings [11,12]. Furthermore, smartphones allow for continuous connectivity, and hence, smartphone-based fundus imaging might pave the way for telemedicine in ophthalmology [13].

With the advent of smartphone-based fundus imaging, multiple applications in ophthalmology including smartphone-based diabetic retinopathy screening [14-20] and smartphone-based optic nerve head evaluation [5,21,22] have become available. In addition, smartphone-based fundus imaging has successfully been used for documentation of hypertensive retinopathy [23] and its utility for fundus documentation in a pediatric emergency department has been evaluated [24]. Furthermore, direct smartphone-based fundus imaging might have merits in teaching medical students and other health care trainees fundus examination as it seems much easier to understand and master than conventional direct funduscopy [25,26].

Here, we evaluated a novel, high-quality optics adapter for smartphone-based fundus imaging in single-image acquisition or video mode for documentation of a variety of retinal/vitreous pathologies in a patient sample with wide refraction and age ranges.

A total of 47 eyes of 32 participants with healthy eyes or the following vitreous/retinal disease were recruited (see Table 1 for characteristics of the sample): healthy (8 eyes), exudative age-related macular degeneration (6 eyes), retinal detachment (4 eyes), adult-onset vitelliform macular dystrophy (4 eyes), intermediate-stage age-related macular degeneration (3 eyes), status post retinal surgery (3 eyes), dome-shaped maculopathy (2 eyes), choroidal nevus (2 eyes), uveal coloboma (2 eyes), central serous chorioretinopathy (2 eyes), central retinal artery occlusion (1 eye), retrohyaloidal hemorrhage (1 eye), Coats disease (1 eye), status post blunt ocular trauma with choroidal rupture (1 eye), posterior uveitis (1 eye), branch retinal vein occlusion (1 eye), retinoschisis with retinal detachment (1 eyes), status post retinal gunshot injury (1 eye), asteroid hyalosis (1 eye), peripheral retinal hole (1 eye), and endogenous endophthalmitis (1 eye). Visualization of the fundus using smartphone-based fundus imaging was possible in all eyes following dilation, irrespective of refraction. Image quality was sufficient to document various specific fundus pathologies including pathologies with only subtle findings as in macular dystrophy, as demonstrated in the exemplary images of the posterior pole in Figure 2. A total of 14 eyes were imaged by both smartphone-based fundus imaging and conventional fundus imaging (see Figure 3 for exemplary images). Images acquired with conventional fundus imaging were sharper and had less reflex artifacts, compared with images acquired by smartphone-based fundus imaging (P<.001 and P=.03; Figure 4). There was no significant difference in contrast and illumination (P=.10; Figure 4), and usable field of view was not different between smartphone-based fundus imaging and conventional fundus imaging—0.99% (0.03%) and 100% (0%), respectively (P=.07).

Adjustable depth of focus allowed for documentation of retinal detachments in varying focal planes (Figure 5). Applicability in children was demonstrated by imaging a 7-year-old child with Coats disease (Figure 6). We found image quality to be limited in the far periphery, but sufficient to detect the Coats-specific pathological alterations (Figures 6C and 6D). In contrast to the far periphery, the midperiphery could be imaged with only slightly reduced image quality as compared with the posterior pole (Figure 7).

Documentation of pathology was more easily achieved using the video mode than the single-image acquisition mode. Multimedia Appendix 1 shows an exemplary smartphone-based fundus examination in video mode where first the healthy right eye and then the left eye exhibiting vitreous synchysis were imaged. Furthermore, smartphone-based fundus imaging in video mode with slightly altering camera directions conveys a 3D impression of structures such as deepened optic nerve cups (Multimedia Appendix 2). We termed this mock 3D impression pseudo-biomicroscopic effect.

Glare of the image periphery (Figures 2F and 7B) and minor reflexes (Figures 5A and 5B) were image artifacts that we encountered during both single-image acquisition and video mode, with the latter artifact being predominantly present in pseudophakic eyes. Glare increased as images were acquired from more peripheral parts of the retina (Figures 6C and 6D). Smartphone-based fundus imaging with this device was also possible without pupil dilation but we found it to be more difficult and time consuming (>2 minutes for one image of the posterior pole); furthermore, reduced field of view due to increased glare limits its application (see Figure 8 for an exemplary image acquired without pupil dilation).

Smartphone-based fundus imaging with this device allows for retinal imaging of a diverse group of retinal pathologies in adults and children. Best image quality is achieved at the posterior pole and decreases with more peripheral imaging. Use of the video mode had several advantages over single-image acquisition mode but comes with a higher data volume, reduced image resolution, and the need for postprocessing (ie, selection of relevant parts of the video).

Our study further supports existing data showing that retinal imaging using smartphone-based fundus imaging can be successfully applied to a variety of ophthalmologic diseases and might be implemented in health care systems [5,14-19,21-24]. Smartphone-based fundus imaging might be especially applicable in children as they are often familiar with and interested in smartphones. For example, it has been shown that smartphone-based fundus imaging can be used in a pediatric emergency department setting for screening for fundus pathologies [24].

Image quality of conventional fundus imaging was superior to that of smartphone-based fundus imaging, yet smartphone-based image quality was sufficient to document various fundus pathologies, including only subtle findings. Imaging the peripheral retina is possible; however, due to optical limitations—also related to the eye not being a perfect optical system [27]—image quality is reduced. Glare increased with more peripheral imaging and was particularly pronounced without pupil dilation. This might indicate a correlation of image glare with the effective aperture in the optical path. As the viewing angle increases in imaging of the mid and especially far retinal periphery, the effective aperture at the level of the pupil gets smaller and exhibits asymmetric distortions [28]; therefore, among other imaging artifacts, light scatter increases, which leads to increased glare in the image periphery. In addition, we observed characteristic small reflex artifacts which were predominantly seen in pseudophakic eyes, albeit not reducing overall image quality. Cataract might be a further confounder of image quality in smartphone-based fundus imaging; however, there were no major lens opacities in our study population.

Smartphone-based fundus imaging without pupil dilation might appear promising as no topical application of drugs is needed and examination can be performed directly without any waiting time. Although this device was not specifically built for imaging without pupil dilation, we showed that it is in principle possible, but not feasible due to reduced image quality and increased examination time. Yet, there are other approaches for smartphone-based fundus imaging which are capable of image acquisition without pupil dilation [4,24,29]. However, pupil dilation is still recommended to achieve best image quality in smartphone-based fundus imaging [22].

Smartphone-based fundus imaging using this device allows for retinal imaging of a diverse group of retinal pathologies in adults and children. Using the video mode has several advantages over single-image acquisition mode but comes with a higher data volume, reduced image resolution, and the need for postprocessing. We showed that video documentation of deepened optic disc cups conveys a mock 3D impression, which we termed pseudo-biomicroscopic effect. Although smartphone-based fundus imaging using video mode has already been reported [30], this is the first time this mock 3D impression is described. The pseudo-biomicroscopic effect might aid evaluation of optic nerve heads, for example, in the setting of a lower-resource glaucoma screening.

The strengths of our study are a comparison with conventional fundus imaging, semiquantitative analyses of image quality, a detailed description of the device, an application to a variety of different retinal diseases, wide refraction and age ranges of participants, and a novel approach for video-based documentation of the 3D structure of the optic nerve cup. However, we did not compare this device with other smartphone-based fundus imaging adapters, had a relatively small sample size, and did not evaluate applicability in eyes with clinically relevant lens opacities.

Image quality of conventional fundus imaging was superior to that of smartphone-based fundus imaging in terms of sharpness and reflex artifacts, although this novel smartphone-based fundus imaging device achieved image quality high enough to document various fundus pathologies, including only subtle findings. Hence, smartphone-based fundus imaging with this device might represent a mobile alternative for high-quality fundus documentation, for example, in immobilized intensive care patients. An image quality comparison between different smartphone-based fundus imaging approaches and further studies on the applicability of this device are warranted.