RAPID SARS-COV-2 TESTING WITH AI ANALYSIS OF EYE IMAGE
iDetect has found a way to achieve SARS-COV-2 testing by only using images of the sclera of the eye.
Fortem Genus Labs (FG Labs) a service-disabled veteran-owned corporation and Fayetteville State University (FSU), a Historically Black College or University (HBCU), and a constituent institution of the University of North Carolina (UNC) have partnered to train, test, and operate iDetect. iDetect is a patent-pending combination of existing open and proprietary technologies to create a rapid (less than 1 minute). Sensitive and specific eye imaging test to identify subjects infected with SARS-CoV-2, the pathogen responsible for SARS-COV-2. The principal components of the system are High-Resolution Retinal, Iris and Scleral images, collected with existing, commercially available fudiscopic cameras (see pic below) equipped with Bluetooth or Wifi capability, data repository software, and a Convolutional Neural Network (CNN) trainable with the Inception Resnet deep neural network architecture.
FSU is currently conducting SARS-CoV2 testing for 3000 subjects, who respond to notices offering free testing, at locations in and around Fayetteville, NC. In conjunction with the FSU SARS-COV-2 testing, FG Labs is conducting eye scan imaging on the subjects participating in the SARS-COV-2 testing. The eye images are tagged as SARS-COV-2 positive or negative and then uploaded to iDetect for the purpose of training the model to identify subjects as SARS-COV-2 positive or negative.
Approximately 1/3rd of the COVID testing subjects are agreeing to have their eyes imaged. Therefore, it is estimated that the FSU viral testing and FG Labs imaging process will render approximately 50 SARS-COV-2 positive subject eye images. We estimate that a minimum of 750 to 1,500 SARS-COV-2 positive eye images and an additional 750 to 1,500 SARS-COV-2 negative eye images will be needed to validate our hypothesis by training iDetect to be able to recognize a SARS-COV-2 signature in the eye images with a reasonable degree of accuracy (see pic below of eye imaging done in conjunction with COVID19 testing). Therefore, FG Labs is working to form an iDetect consortium and expand its access to SARS-COV-2 positive eye images by partnering with additional public and private organizations associated with the UNC system that is conducting SARS-COV-2 testing. In this regard, FG Labs and FSU are forming a unit to provide eye imaging and data processing training to iDetect mobile teams that will be deployed to consortium partner locations throughout North Carolina where SARS-COV-2 testing is conducted. The mobile iDetect teams will conduct eye imaging and data gathering in collaboration with the UNC consortium partners who are conducting SARS-COV-2 testing. Basic demographic and medical history information is collected on each subject volunteer.
This project was approved and is supervised by the UNC - IRB. FG Labs in collaboration with FSU is currently operating under a 5-month UNC grant ending December 2020. We are requesting COVID-19 testing partners to enable us to jointly continue the eye imaging beyond December 2020 until July 2021 and to expand the scope of the current eye imaging to include additional UNC COVID-19 testing affiliates.
The additional eye imaging will enable us to more rapidly acquire the sample data and images necessary to train our CNN and conduct preliminary pivotal clinical trials.
WHO WILL USE THE OUTPUTS OF THIS PROJECT? WHAT VALUE WILL THIS PROJECT PROVIDE TO THAT USER?
The proposed approach is envisioned to be easily accessible to the health-care community world-wide. The use of this technology will provide not only a direct clinical benefit to providers and patients, but provide hospitals, health departments, and other testing centers with rapid (, 1 min from acquisition to outcome), inexpensive, noninvasive tests to detect SARS-CoV-2, and other diseases. The test is as quick as checking temperature plus a symptom review and could be used for access control as well as diagnostic triage.
iDetect TECHNOLOGY, JUSTIFICATION, AND WORK TO DATE
Since the beginning of the SARS-COV-2 pandemic until now there has been an acute shortage of SARS-COV-2 testing capability in the U.S. and worldwide. Therefore, in April 2020 FSU in collaboration with its partner FG Labs we began research to develop an (AI) artificial intelligence-based, rapid, non-invasive SARS-COV-2 screening solution.
The diagnostic test for SARS-COV-2 infection is a reverse transcription-polymerase chain reaction (RT-PCR) test. However, there has been a severe shortage of test-kits worldwide and laboratories in most countries have struggled to process the available tests within a reasonable timeframe. The US Dept of Health, Inspector General conducted a SARS-CoV-2 Hospital Experience Survey  from March 23-27, 2020, with hospital administrators from 323 hospitals across 46 States. The ensuing IG report released on 3 April stated that: “Hospitals reported that their most significant challenges centered on testing and caring for patients with SARS-COV-2 and keeping staff safe. Hospitals said that severe shortages of testing supplies and extended waits for test results limited hospitals' ability to monitor the health of patients and staff”. Furthermore, RT-PCR has limited sensitivity for SARS-COV-2 infection at 71% . The paucity of testing kits warrants a system of ‘triaging’ patients most needful of a confirmatory RT-PCR test for SARS-COV-2. A rapid, non-invasive and readily available screening tool based on machine learning and computer vision could fill this gap by identifying patients who need further testing. It should detect both symptomatic and asymptomatic subjects therefore will be more useful for contract tracing and access control purposes. Identifying and isolating asymptomatic, infected “carriers” remains well beyond the capabilities of current tests. The ability to readily identify such “silent carriers” will dramatically improve our approach to contagion management. The US testing capacity is currently approximately 800,000 test per day, however, the current demand for tests is 1.5 million tests/day and growing (WJS reference)
Due to the SARS-COV-2 testing shortage, FG Labs and its partner FSU decided to explore options for SARS-COV-2 testing using artificial intelligence (AI). After researching a range of AI solutions, it was decided to develop “iDetect” a non-invasive, rapid, means of risk assessment for SARS-CoV-2. Our iDetect proof-of-concept (POC) research uses fundus eye images captured using FDA-approved handheld portable non-mydriatic camera technology. The eye images are then processed through our (CNN) Convolutional Neural Network model that employs transfer learning with deep neural network architecture - Inception Resnet to re-train and fine-tune the model. This approach has been implemented in our POC and evaluated to effectively detect Diabetic Retinopathy from retinal scans. An overall validation accuracy of 92.58% has been achieved by our POC over the validation dataset of retinal scans. Our deep neural network has been trained over a dataset of 1368 retinal scans split into two classes - “No Diabetic Retinopathy” and “Diabetic Retinopathy”. In addition to the retinal scans, our iDetect POC was re-trained, fine-tuned, and evaluated on Chest X-Ray images to detect normal pneumonia and SARS-CoV-2 cases. Our POC model trained and evaluated over 1196 Chest X-Ray images resulted in a validation accuracy of 89.29% over the validation dataset. We are encouraged that iDetect.01 differentially diagnosed SARS-CoV-2 pneumonia based on imaging analysis alone.
We hypothesize that our Convolutional Neural Network-based approach could prove useful for SARS-CoV-2 risk assessment from not only retinal scans but scleral and iris scans as well. The success of this approach for retinal imaging Diabetic Retinopathy Detection as well as X-Ray imaging SARS-COV-2 detection reinforces our hypothesis of using CNNs for effective risk assessment for SARS-CoV-2 from retinal as well as scleral and iris scans using handheld portable non-mydriatic camera technology.
SIGNIFICANT DATASETS NEEDED:
The data set used to train our POC model was provided by Kaggle. The CNN model was trained to detect diabetic retinopathy in retinal images and COVID-19 in X-ray images. We implemented a dataset of 1368 retinal scans split into two classes - “No Diabetic Retinopathy” and “Diabetic Retinopathy”. An overall validation accuracy of 92.58% has been achieved by our POC model over the validation dataset of retinal scans. We also implemented a dataset of 1196 Chest X-Ray images split into 3 classes – “normal”, “pneumonia” and “SARS-CoV-2 cases”. Our POC model achieved a validation accuracy of 89.29% over the validation dataset.
In collaboration with FSU, we are currently collecting a data set of COVID-19 negative and positive eye images of both the retina and the eye surface (iris and sclera). This data set is being collected from 3000 subjects beginning mid-July 2020 ending December 2020 or sooner depending on the date of the last of the 3000 COVID-19 tests administered. We estimate the FSU viral testing and FG Labs imaging process will render between 150 and 300 SARS-COV-2 positive subject eye images. We hypothesize that a minimum of 750 to 1,500 SARS-COV-2 positive eye images and an additional 750 to 1,500 SARS-COV-2 negative eye images will be needed to validate our hypothesis by training the iDetect to be able to recognize a SARS-COV-2 signature in the eye images with a reasonable degree of accuracy.
Our goal is to collect the additional eye imaging datasets needed by partnering with UNC affiliates who are scheduled to conduct an additional 7500 to 15,000 COVID-19 viral tests.
RISKS AND PROPOSED MITIGATION
There is a risk that a much larger data set of 10,000+ sample eye images may be required to train and test our CNN and achieve a reasonable degree of accuracy, In this event, it will require us to partner with an increased number of COVID-19 testing partners to gather the necessary eye images.
We may also determine that there is a fidelity issue with the eye image datasets as a result of the capability of the currently available retinal imaging devices. In that regard, a much deeper retinal scan may be required using hand-held OCT (optical coherence tomography) retinal scanning devices which are still in the prototype phase and commercially unavailable. In this event, we will need to purchase and/or construct OCT prototypes.