Ultrawide field non-mydriatic portable pediatric retinal imaging - Siloam Vision aims to revolutionize pediatric retinal imaging by introducing two advanced, laser-based, artificial
intelligence (AI) enabled retinal imaging systems. These innovations are primarily targeted at addressing an
unmet need in retinopathy of prematurity (ROP), which is a leading cause of child blindness in every part of the
world, even though blindness is nearly always preventable with accurate diagnosis and timely treatment. Current
examination methods for ROP require either in person dilated examination with ophthalmoscopy by a skilled
clinician, or digital imaging using an available widefield digital fundus imaging (WDFI – up to 130 degrees)
system. Both of these methods are stressful to neonates because they use bright illumination, and are expensive
because they either take the limited time of a clinician, or require purchase of a WDFI system ($50,000 –
150,000). Siloam Vision's novel approach involves an infrared laser-based camera capable of ultra-widefield
imaging (UWF - up to 165 degrees) using a scanning laser ophthalmoscope (SLO) or an optical coherence
tomography (OCT) system, offering 2D or 3D imaging, respectively. There are no commercially available
handheld OCT systems on the market, and none of the current WDFI systems match the field of view (FOV) and
optical advantages of these systems, which do not require dilation, can image the entire retina in a few seconds,
and use infrared light to reduce neonatal stress. Our prototype UWF OCT system has already been successfully
utilized in over 2000 ROP exams, demonstrating the potential of quantitative OCT biomarkers in clinical disease
staging, but as built is too expensive for commercialization. To address this, and to bring this technology to
market faster, we are leveraging the existing dataset of UWF-OCT images to develop super resolution AI
algorithms that can match the resolution and FOV of our current prototype, using lower-cost components, a
strategy that has been used successfully in radiology. Thus, there are two key innovations that we are
developing: a wider FOV camera design that can be paired with either an SLO or OCT, and super-resolution AI
to make UWF-OCT commercially viable. The project is structured around two specific aims: Aim 1: Optimize the
performance and cost of the OCT device using super-resolution techniques, like generative adversarial networks,
to improve image quality while reducing costs. This involves comparing images from 100 kHz and 400 kHz
lasers, aiming to achieve similar clinical utility with cheaper components. Aim 2: Develop commercial ready SLO
and OCT systems, ready for regulatory approval. This includes refining the design to reduce cost, size, and
weight, and conducting a feasibility study for non-mydriatic imaging. For the ROP market, the ultimate goal is to
combine these innovative devices with Siloam Vision's existing AI software for ROP, currently in clinical trials, to
an integrated AI-camera system for ROP screening and diagnosis. Long-term, both the SLO and OCT systems
are clinically superior to any existing technology on the market and may have widespread impact in improve the
care of pediatric patients with retinal disease.
