Radiology is evolving with the adoption of deep learning models for the recognition of lesions in the body, to prioritize cases for the direct treatment of patients at risk and to predict the evolution of pathologies. Artificial intelligence (AI) affects all the imaging modalities, in particular Magnetic Resonance Imaging (MRI), Computed Tomography (CT) scanning, X-rays and ultrasound imaging.
In its latest report, “Artificial Intelligence for Medical Imaging” Yole Développement’s analysts estimate that the total market revenues in 2025 for AI software tools will reach $2.9B with a Compound Annual Growth Rate for 2019 to 2025 (CAGR-2019-2025) of 36%. This will be shared between the main applications: improved image capture, noise reduction, image reconstruction, screening, diagnostic and treatment planning.
Arterys, a key player in the ecosystem of startups developing AI algorithms for medical imaging applications, has responded to Yole’s questions. Marjorie Villien, PhD, Technology & Market Analyst, Medical & Industrial Imaging at Yole interviewed Maya Khalifé, PhD, Product Manager, and Stephane Boyer, General Manager at Arterys.
Yole Développement (YD): Please could you introduce Arterys and its activities?
Maya Khalifé & Stephane Boyer (MK & SB): Arterys is a San Francisco-based company, created in 2013. Our mission is to create an automated, intelligent diagnostic platform that unleashes real-world clinical data to make healthcare more accurate and data driven.
Our Cloud-based platform is accessible by everyone from everywhere and is able to process medical imaging and patient data with the help of AI based algorithms.
Due to a native Cloud based dynamic diagnostic user interface, Arterys Marketplace enables frictionless collaboration between producers and consumers of clinical AI models.
YD: Could you tell us more about your products and what applications you are targeting?
MK & SB: Arterys has started by creating its own clinical applications. We are currently regulated for premium commercial applications in Cardiac MRI, Lung CT and MSK/Chest X-ray. Because we want to offer a much wider catalog of applications than what we can build ourselves, Arterys has moved recently into a collaborative product development by launching Arterys Marketplace.
The Arterys Marketplace was launched in March 2020 and has already five AI models freely accessible, including two models focused on COVID-19 automatic diagnostic in CT and X-ray.
YD: The current COVID-19 pandemic is having a huge impact on the medical imaging industry. We currently see a lot of tools using AI being developed to help radiologists diagnose the disease. Can you tell us more about your developments and where research in general is heading to help medical practitioners deal with this global crisis?
MK & SB: AI researchers around the world are developing and publishing algorithms that can detect, track or quantify COVID-19 in medical images, including CTs and X-rays. Arterys, through its open innovation platform, the Arterys Marketplace, is supporting the development and validation of those algorithms. Clinicians can have access today on the Arterys Marketplace to five AI models from academics and industrial AI contributors, including a model that automatically detects pneumonia on chest X-rays, and another model that segments infection on CT scans, and the list is still growing. By putting those algorithms into the hands of clinicians, they 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.
YD: A majority of the models empowered by AI are trained on CT scans and MRI images for the detection of brain and vascular pathologies. How do you see the market changing in terms of modalities used and pathologies studied? Will this impact the way data is currently processed?
MK & SB: There are three current trends that we can observe in the models empowered by AI creation.
The first one is to provide algorithms for exams with less economic value but massive volume. The best example is X-ray for bones, chest and mammography but we see also trials for knee and spine exams in MR.
The second one is the combination of multiple models for analyzing the same exam. We know that a radiologist looking at a body CT exam or an oncology MR is looking at multiple pathologies. If AI answers only to one clinical question this bring too low value for adoption by the radiologist.
The last one is the capability of the AI models to process image data and patient data to provide more accurate, predictive and personalized diagnostics. This is a challenge for our industry because patient Electronic Medical Record (EMR) data are not structured and can come from a huge number of different IT infrastructures.
YD: Your marketplace is filled with algorithms you developed but also algorithms developed by other software companies. What is your business model regarding your AI solutions?
MK & SB: The business model is still a work in progress, but currently the rules are the following:
- Free for producers, except if the model integration requires a specific development for Arterys
- Fixed fee to access the Marketplace depending on the size of the consumer. This fee can vary depending the volume of exams or the volume of data to store and process.
- Pay per use by model. This is today the only way to pay the producers in the Marketplace.
YD: We expect the market to grow from $154M in 2018 to $2.9B in 2025 with a CAGR(2019-2025) of 36%. How do you see the market adopting AI solutions after the significant investments in this field?
MK & SB: We think that this will be a marathon and not a sprint. After more than three years of massive investments from major companies and venture capitalists, the 150 companies working in diagnostic AI in medical imaging field are still not generating revenue and are still looking for a viable business model. I think that we need two to three years before the start of a real business and five before entering into a profitable one.
YD: RSNA 2019 was the opportunity for all imaging machine manufacturers to present their new products empowered by AI and a lot of these companies launched products using AI. How is Arterys positioned next to the medical imaging giants?
MK & SB: Medical imaging giants are based on an industrial culture. This is very different from a digital culture. Our experience, skills and culture are very complementary to these giants. For companies like Arterys the future could be partnership or acquisition. Now there are also other giants that are interested by our domain. Pharma and digital giants are also thinking of entering this business.
YD: What are your future perspectives concerning the adoption of AI in the medical field?
MK & SB: Up until now, we have seen an increasing number of developers and startups creating AI, publications on AI applied to medical imaging, and invested capital. However, the promised clinical use and improvements in patient care have been unrealized. Arterys believes that to deliver on AI’s promise and achieve adoption of AI, we need bridge the gap and close the virtuous feedback loop between innovators and those clinical users benefiting from their innovations through a toolkit and frictionless channel of distribution of AI.
YD: With the increased adoption of models empowered by AI, how will the work of radiologists evolve? Could this profession slowly disappear?
MK & SB: I don’t think that there are any risks for radiologists in the next 20 years if they adopt AI as a new opportunity. AI will complement their capabilities and will make them even more powerful for the health organizations. AI could also impact radiologist’s profession by decreasing low value tasks, allowing them to focus more on patients and clinic and finally position radiologists at the center of the diagnostic pathway.
Stephane Boyer, General Manager EMEA
Since January 2018, Stephane is managing the commercial development of Arterys for Europe, Middle-East and Africa. Previously he has spent 15 years in GE Healthcare where he had many positions in Sales, Marketing and Business Development. His last job in GE HC was Head of MR business in Western Europe.
Maya Khalifé, Product Manager
Maya is product manager at Arterys working on getting AI-based medical imaging algorithms into the hands of clinicians through the Arterys Marketplace. She’s a biomedical engineer and has earned her PhD in MRI Physics from Université Paris-Sud. Her previous experiences involves working on MRI and PET-MRI methodology and clinical applications.
As a Technology & Market Analyst, Medical & Industrial Imaging, Marjorie Villien, PhD., is member of the Photonics & Sensing activities group at Yole Développement (Yole).
Marjorie contributes regularly to the development of imaging projects with a dedicated collection of market & technology reports as well as custom consulting services in the medical and industrial fields. She regularly meets with leading imaging companies to identify and understand technology issues, analyze market evolution and ensure the smart combination of technical innovation and industrial application.
Marjorie Villien graduated from Grenoble INP (France) and holds a PhD. in physics & medical imaging.
Artificial Intelligence for Medical Imaging
With the emergence of AI in imaging, the medical industry and the radiology profession have begun to dramatically change.