Explore GEN family of websites, social media experiences, blogs and more.

Unleashing ideas.
Strengthening startups.

Back to search results
    Date submitted
  • 22-Sep-2015

OptDx

Abstract

OptDx is an image pattern recognition software that enhances a doctor’s ability to screen infants for retinopathy of prematurity (ROP) which can cause blindness.

Video

Original YouTube URL: Open

Introduction Video

Additional Questions

Who is your customer?

Primarily, the demographic that will most benefit is the Neonatal Intensive Care Unit (NICU) teams. Our technology will help the screening process and lighten up the workflow for the NICU teams and pediatric ophthalmologists. By extension, our technology can help screen more patients so that they can be prevented from ROP related blindness. This will have a tangible impact on the baby’s ability to thrive. We especially target developing middle income countries. Often, these countries have the funding to build the infrastructure such as a NICU and a professional ROP camera called the RetCam, but not necessarily the knowledge base that allows for proper diagnosing. Hence many children in these environments survive their prematurity but unfortunately lose their visions. With our subscription based program, we can help these health networks improve their quality of service. We have seen a need of such programs in India, Armenia, China, and the Philippines. Hence our program aims to not only impact childhood blindness related to ROP nationally but also globally. We believe that OptDx and our automatic screening function is not only a useful tool but also a future necessity.

What problem does this idea/product solve or what market need does it serve?

The current model for screening for ROP is time intensive - it requires an ROP specialist to take multiple photos of patients and then examine these photos to identify which zone, stage and how much risk each child has in developing ROP-related blindness. Furthermore, to help train younger doctors on the field, each mentor must examine a high volume of images to decipher and explain which babies are at risk for developing ROP. This process is very personnel heavy and requires a substantial amount of time from a doctor to not only diagnose the disease, but also to train new practitioners in the diagnostic process. OptDx addresses this bottleneck problem. We are collaborating with Dr. Lee from CHLA who has been teaching ROP to ophthalmologists in Armenia since 2009. His experiences show the need for an automatic screening tool. His training efforts have been incredibly successful with his students achieving success rates higher than or equal to that of doctors in the US. Recently, in order to make training more effective, he has started using a facebook group. In the group, the Armenian ophthalmologists upload photos of their patients’ retina and Dr. Lee comments with recommendations and diagnoses. While training on Facebook has been helpful, his current impact is significantly limited by the number of images he can review. Hence, if we can automatically screen out benign cases, we could help scale his efforts to reach a wider patient base. We at OptDx believe that with a good algorithm and a right business model, we can popularize ROP knowledge worldwide to protect the vision of our premature infants.

What attributes will make this idea/product successful? Why do you believe that those features will create success?

There is both a need and a desire for a training program in the diagnosis of ROP in developing middle income countries. Collaborations with Dr. Lee and discussions with providers from Armenia and the Philippines, we have identified that automatic screening of ROP is an absolute necessary to evolve the current training models to the next level. The quality of current retinal imaging devices has increased dramatically in recent years as machine learning algorithms have assisted physicians in diagnosing diseases and disorders. One example of the latter is the active use of the ADNI (Alzheimer's Disease Neuroimaging Initiative) dataset to develop machine learning models that achieve high accuracy in diagnosing people with Alzheimer’s disease. For ROP, we have a growing dataset of retinal images by cooperating with Dr. Thomas Lee. As ROP is image based, we can breakdown the image pixel by pixel and take pertinent information. This sizeable dataset provides the raw material necessary for the development of machine learning algorithms that can assist in screening for high-risk ROP and training medical practitioners. On the operations side, in addition to the training programs led by Dr. Lee, we aim to secure the cooperation of new hospital networks and NICUs. OptDx is an enterprise subscription based model as we aim to secure collaboration with founding hospital networks such as Interhealth Canada.

Explain how you (your team) will execute to make this idea/product successful? What gives you (your team) an advantage over others already in the market or new to this market?

Our OptDx team has developed a partnership with Dr. Thomas C Lee, MD, the director of The Vision Center at Children’s’ Hospital Los Angeles and an Associate Professor at the USC Eye Institute. Currently, Dr. Lee uses telemedicine to diagnose ROP through a Facebook group. Physicians from around the world upload retinal images to his Facebook page and Dr. Lee, along with his associates, diagnose the stage and severity of ROP in these young patients. As he is collaborating with more centers not only abroad but also rural areas in the U.S., we will have access to the practice workflow to verify our algorithm and prove the trustworthiness of the technology. We have gained approval to use his database with retinal images that are diagnosed by world-renowned specialists as input into our software. We have also developed working relationships with ROP specialists who have agreed to advise on the accuracy of our algorithm. We have yet to acquire any funding, but we have spoken with an enterprise venture firm based in San Francisco unofficially. One of the associates is helping us fine-tune our pitch and is providing us advice and guidance. We aim to look for funding more aggressively when we have a stronger algorithm.