Released by Newswire.com
SAN FRANCISCO – AI researchers around the world are developing and publishing algorithms that can detect COVID-19 in medical images, including CTs and x-rays. To support the development and validation of those algorithms, Silicon Valley startup Arterys has launched its open innovation platform, the Arterys Marketplace. AI developers can publish their algorithms to the Arterys Marketplace to make them available for testing, validation, and trial by the global community.
According to Arterys co-founder and radiologist Dr. Albert Hsiao, researchers, machine learning scientists, and clinicians need to work together to develop these AI tools. “Our lab at UC San Diego has shifted focus to tackle this pandemic with the hope that we may leverage artificial intelligence to expedite accurate diagnosis of COVID-19 to reduce unnecessary spread of this disease, and assist in management of patients with severe infection. We hope the global community will come together and join us in this fight,” said Dr. Hsiao.
The Marketplace went live in late March and already includes a COVID-19 AI model from Chinese developers at Aitrox, a chest x-ray analyzer from Korean developers at Vuno, and a chest opacity and bone lesion detection model from French developers at Milvue.
Marketplace users can test algorithms by uploading anonymized medical images through a web browser. AI-enhanced results are delivered to the user’s email inbox. Links to interactive analyses can be shared with colleagues and collaborators for additional review. Users can then connect directly with algorithm developers to provide feedback. Developers can use that feedback to improve their models, moving them closer to being ready for use in clinical settings.
According to Rem Darbinyan, CEO of SmartClick.AI, which recently released its COVID-19 chest x-ray algorithm, “Joining the Arterys Marketplace Community has given our team advantages in developing technology in the medical field … a unique opportunity to get an immediate response in peer and medical review through testing and feedback.”
Independent innovators are finding value, too: “I joined Arterys in hopes of finding new and better detection methods for the COVID-19 but it soon gave me a bigger window to the world,” said AI developer Darshan Deshpande. “People here are extremely helpful … Not only did my work get recognition but the community also gave me an amazing chance to contribute to a noble cause.”
Since launch, over 100 volunteers from the clinical research and machine learning communities have begun working together via a public Marketplace instant messaging channel to develop, share, and improve the algorithms. Community members are helping each other by aggregating COVID-19 medical image datasets from around the world, sourcing and developing new models, and providing feedback to improve the performance of existing ones.
“We are trained to be public health leaders,” said one volunteer, a Schulich MPH graduate student in Canada named Eemaan. “Healthcare workers, such as my mother and some of my classmates, are fighting COVID-19 on the frontlines. I cannot sit at home and do nothing. I have always been intrigued by the scope of AI in medicine. I took a hard look at my skills and decided to support the Marketplace. I source X-rays and CT scans of COVID-19 patients and contact the owners of the datasets for permission to use in our initiative. Then they are validated and improved through the Arterys Open Innovation Platform.”
Model developers who want to bring their medical AI online can get started by joining the Slack community, reviewing the Arterys SDK, and submitting this form.
Clinical researchers and innovators with anonymized medical images can test out AI models on the Arterys Marketplace at http://marketplace.arterys.com.
“It is absolutely amazing what energy is here … a perfect place for innovation,” said Dr. Raluca Chelu, Rotterdam-based radiologist and Marketplace community contributor. “I was trained as a radiologist and further I studied Python and Machine Learning in order to ‘speak the language’ of the engineers. I want to provide UX feedback and to contribute with my radiology knowledge to help these AI tools to succee