2.1 Imaging setup
Figure 1A shows photographic illustration of the nonmydriatic wide-field fundus camera with trans-palpebral illumination. The trans-palpebral illuminator consists of 4 optical fibers which has 600 µm diameter and 0.39 numerical aperture (Fig. 1A). Each fiber is connected to LED light sources which has 530 nm (M530L4, Thorlabs Inc, Newton, NJ, USA), 625 nm (M625L4, Thorlabs Inc, Newton, NJ, USA), 780 nm (M780L3, Thorlabs Inc, Newton, NJ, USA), and 970 nm (M970L4, Thorlabs Inc, Newton, NJ, USA). The wavelengths of LEDs were carefully selected to acquire the retinal and choroidal vasculature. The schematic diagram of the trans-palpebral illuminator and eye illustrates the illumination position and pars-plana location (Fig. 1B). The detailed optical layout of the wide-field fundus imaging system is shown in Fig. 1C. The first, second, and third lens (L1, L2, and L3) of the imager are a meniscus lens (LE1234-A, Thorlabs Inc., Newton, NJ), plano convex lens (67–152, Edmund Optics Inc., Barrington, NJ), and double convex lens (63–688, Edmund Optics Inc., Barrington, NJ, USA), respectively. This lens combination produces an aerial image of the retina in front of the relay optics, triplet achromatic lens L4 (67–422, Edmund Optics Inc., Barrington, NJ, USA). In coordination with the relay optics and a camera lens with a focal length of 12 mm (33–303, Edmund Optics Inc., Barrington, NJ, USA), the aerial image is relayed to the camera sensor. A color CCD camera (GS3-U3-41S4C-C, Flir systems Inc, Wilsonville, OR, USA) and a monochrome camera (GS3-U3-41S4M-C, Flir systems Inc, Wilsonville, OR, USA) were used for MSI (Fig. 3, and Fig. 5) and color fundus imaging (Fig. 4), respectively. Both cameras have a frame rate of 18 frames per second, frame resolution of 2016 × 2016 pixels, with 3.1 µm × 3.1 µm pixel size. The sensor provides quantum efficiencies of 75%, 58%, 22%, and 4% at 530 nm, 625 nm, 780 nm, and 970 nm wavelength, respectively.
2.2 Optical simulation
The optical system for the wide-field fundus imaging (~ 93° visual-angle; ~140° eye-angle) was designed and evaluated by Zemax simulation (Zemax OpticStudio 18.7, ZEMAX LLC., Kirkland, WA, USA) to optimize the image quality and to ensure optimum performance. As shown in Fig. 1C, the simulation starts from the eye pupil. All off-the-shelf lenses were selected from the lens catalog in Zemax libraries (Fig. 1C). Paraxial surface was used for last lens surface to mimic the camera lens which is not available in the lens catalog and also the aperture was set to working as a stop to simulate ray vignettes. The system optimization was performed by field of view, spot diagram, and modulation transfer function (MTF) whose modulus of optical transfer function evaluation from various field angles at 0°, 15°, 30°, and 45° visual-angle (0°, 23°, 45°, and 68° eye-angle). Note that the field angle is half of the visual-angle. The minimum eye pupil size for the wide-field fundus image was validated by simulating the entrance pupil shape and position deformation from various field angles. The eye pupil size was changed from 2.5 mm to 5mm diameter with fixed camera aperture size as 6.66 mm. To validate the effect of eye pupil size on the image quality, the relative illumination was simulated.
2.3 Human subjects and fundus imaging
This study was approved by the Institutional Review Board of the University of Illinois at Chicago and followed the ethical standards stated in the Declaration of Helsinki. Seven healthy subjects with no history of eye disease were recruited to validate the proposed fundus camera prototype. The informed consent was taken from each subject.
The fundus images were taken in a dark room condition. The subject head was placed to forehead/chin rest for stable imaging (Fig. 1A). The trans-palpebral illuminator was positioned to the eyelid (Fig. 1A and 1B). The illuminator can adjust transverse, vertical position, and angle depending on the subject. Considering the pars-plana width and distance from the limbus, the center of illuminator was placed ~ 6 mm away from the limbus and the orientation of the illumination was aligned to normal direction of the sclera. The optimal illumination location, i.e., the pars-plana, could be identified based on the image quality by fine adjusting of the illuminator. During the image acquisition, live view of fundus image was streamed to monitor imaging location and performed the fine focus adjustment. The illumination location was maintained during the multispectral fundus imaging. The exposure times were set as 500 ms and 100 ms for 530 nm and rest of other wavelength LEDs, respectively. For the color fundus image, 530 nm and 625 nm LED was turned on and off sequentially for green and red fundus image, respectively. The camera exposure was set as same for all imaging sequence and the illumination power was adjusted to 40 mW and 1.5 mW for green and red optimized fundus images. Five fundus images were acquired from different location to expand the FOV of MSI by utilizing fixation target. First, center aimed fundus was acquired and other four image locations were positioned roughly 33°~40° visual-angle (50°~60° eye-angle) away from the center fundus image to each diagonal direction.
2.4 Ocular light safety
The ocular light safety was evaluated according to ISO standard “Ophthalmic Instruments - Fundus Cameras” (10940:2009) [24] which safety limits are at least 10 times lower for retinal threshold damage. Both photochemical and thermal hazards of the retina were quantitatively evaluated. To acquire the fundus images within the safety limit, maximum permeable exposure time was calculated from all wavelengths. The illumination power of each wavelength were 40 mW, 15 mW, 10 mW, and 4 mW for 530 nm, 625 nm, 780 nm, and 970 nm, respectively. The transmission of the sclera and eyelid for different wavelengths were considered. According to the ISO standard, a maximum of 10 J/cm2 weighted irradiance is allowed on the retina without photochemical hazard concern. The weighted irradiance was calculated using the photochemical hazard weighting function provided in the ISO standard. For conservative estimation of the worst case, assuming all light directly reaches to the retina behind the illuminated sclera area, the illuminated retinal area was estimated as 0.2826 mm2 considering that the fiber diameter is 600 µm. The details about the retina safety calculation was described in [8]. The maximum permeable exposure time was ~ 35 minutes for the 530 nm illumination and > 24 hours for the rest of wavelengths. The maximum weighted power intensity allowed on the sclera without thermal hazard concern is 700 mW/cm2. The equivalent powers for thermal hazard estimation were 191 mW/cm2, 127 mW/cm2, 154 mW/cm2 and 62 mW/cm2 for 530 nm, 625 nm, 780 nm and 970 nm light sources, respectively, which is 4–11 times below compared with the maximum weighted power intensity allowed on the sclera without thermal hazard concern.