The A-Z Guide to radiology AI companies showcasing at RSNA 2017
The largest conference in the world starts on the 26th November 2017. Note I didn’t say largest ‘radiology’ conference. Nope — the world’s largest conference, regardless of sector anywhere in the world, starts once again in the windy city of Chicago. The Radiological Society of North America’s annual conference boasts upwards of 50,000 physical attendees, thousands of virtual ones, and tonnes of vendors and software providers. All of these people will pack into the absolutely vast 2.6million square feet of McCormick Place (the world’s largest convention center), bringing with them, I kid you not, up to 30 full size MRI scanners. It is quite literally the epicenter of the global radiology community’s calendar, and something to breath-taking to behold. Just have a look at the floorplan for the exhibitors — it’s like a mini-city!
As I make my plans to head over, I thought I’d do some research on the expanding number of artificial intelligence start-ups and companies who have booked in to show-off their tech in the Machine Learning Showcase arena. Who will make deals? Who will be acqui-hired by the big fish? And who will be dead in a year? Let’s find out….starting with ‘A’…
Disclaimer: all information in this article is taken from publicly available company websites, LinkedIn, Crunchbase, news searches, and details are accurate at the time of writing (Nov 2017).
Aidence
Location: Amsterdam, Netherlands
Founded: 2015
CEO: Mark Jan-Harte
Funding: €2,250,000 seed funding, June 2017
Product: Lung CAD, pending CE marking
“Aidence’s first solution LUNG CAD provides fully automated analysis and reporting of pulmonary nodules. With accuracy levels equal or even surpassing human capabilities, Aidence’s LUNG CAD can be used both in a clinical routine, as well as in a lung screening setting to improve diagnostic accuracy and reduce reading times.”
The team behind Aidence achieved a top-3 result in the 2017 Kaggle Data Science Bowl to build software to predict lung cancer based on a single chest CT scan. They are ISO 13485 certified, and planning launch of a CE marked product this year — very likely it will be showcased at RSNA. Will lung CAD see clinical value given the recent update to Fleischner guidelines effectively making nodules <6mm pointless to follow up in low risk patients?
AiDoc Medical
Location: Tel Aviv, Israel
Founded: 2016
CEO: Elad Walach
Funding: $10,500,000 in total
Product: Unknown — “AIdoc helps radiologists work through their case load faster, just in time to make a difference.”
“Aidoc utilizes deep learning and AI to analyze medical images and patient data, to detect and pinpoint critical anomalies for radiologists. It is done while simultaneously integrating patient data, resulting in seamless streamlining of their workflow which frees them to focus on what they do best.”
With no official product yet despite over $10million in funding, it’s hard to make any judgement. The team did win Israel’s Digital Health start-up competition in 2016, so that’s something. They claim to have developed “a new kind of deep learning technology that can detect high-level abnormalities”, so let’s hope they have something to show at RSNA.
Arterys
Location: San Francisco, USA
Founded: 2011
CEO: Fabien Beckers
Funding: $13,716,000 in total
Product: Arterys 4D flow and Cardio DL
“Arterys’ 4D Flow technology generates precise measurements of blood flow non-invasively and non-radiatively anywhere in the body. The company’s medical imaging SaaS analytics platform combines the best of cloud computation and machine learning to improve diagnostic and therapeutic decisions.”
Arterys made headlines earlier in the year for being the first deep learning company in the imaging space to get FDA clearance for it’s cloud-based technology, following CE marking in December the year before. Massive funding followed, and Arterys look set to be the firm leaders in cardiac MRI quantitation and segmentation, with other products reportedly in production.
Blackford Analysis
Location: Edinburgh, Scotland
Founded: 2010
CEO: Ben Panter
Funding: £2,410,000 in total
Product: Blackford Workflow Server and Automated Image Registration
“We provide a single platform for quick access to automated image analysis and processing that can be integrated into existing workflows for image capture, storage, interpretation, viewing and sharing.”
While not technically a start-up (almost 8 years, and counting) their measured approach has secured a recent partnership with RadNet across America after FDA clearance. Their first AI product on the platform is automated image registration. Going down the platform route instead of aiming for Clinical Decision Support certainly seems to have paid off, and getting a foot in the door in the huge American market will likely be a smart move, allowing Blackford to roll out new algorithms directly into existing clinical partners’ workflows.
ContextFlow
Location: Vienna, Austria
Founded: 2010
CEO: Markus Holzer
Funding: Undisclosed
Product: Radiology Explorer
“The Radiology Explorer provides contextflow with a simple, accurate and fast image search tool to provide radiologists with a new approach to searching for information. By selecting a region on the radiological image, they can immediately access information, relevant to the actual case, concerning visually similar cases, reference literature and relevant guidelines to provide a faster and more accurate diagnosis.”
With CE and FDA processes ongoing, ContextFlow’s intelligent search is yet another interesting take on the use of AI in imaging (I wrote previously that this is one of the top 5 uses for radiology AI). They will likely be showcasing their new work called Temporal Trajectories for Radiology Image Search (TeTRIS), a lung CT trajectory system that plugs-in to the existing product.
Combinostics
Location: Tampere, Finland
Founded: 2014
CEO: Lennart Thurfjell
Funding: €350,000 seed funding + unknown EU grant
Product: cNeuro cMRI
“Combinostics provides advanced tools for data-driven diagnostics, giving the physician a holistic view of all patient data. We provide cloud-based tools for extracting biomarkers from images, lab data and cognitive tests and integrating and contrasting all data with data from previously diagnosed patients. The combined data is presented in a decision support system using Combinostics’ patented disease-state fingerprint-technology.”
ISO 13485 certified company Combinostics have CE marked their suite of tools for extracting quantified imaging biomarkers from brain MR. A strong scientific background has put them on firm footing, and their tools see use on the academic circuit, with plenty of publications. Whether or not the technology translates to added value in clinical practice is yet to be seen.
CuraCloud
Location: Seattle, USA
Founded: 2016
CEO: Unknown (!)
Funding: Undisclosed
Product: CuraCloud A.I.
“Our proprietary artificial intelligence (AI) algorithms are uniquely tailored to medical data, adding machine intelligence and automation to challenging steps in medical image analysis, clinical results mining, molecular oncology and other areas of medicine and research.”
I couldn’t find much information about CuraCloud online — their CEO is neither on LinkedIn nor listed on Crunchbase, and they don’t appear on Glassdoor. Their website has a lot of ‘fluff’ about enhancing capabilities and spanning modalities… but not much information on what exactly their product does, nor do they have any regulatory clearances— but from what I can tell they aim to triage imaging exams to present to the radiologist’s workflow, and have a partnership with Clario and an un-named academic institution. They also claim that they aid in digital pathology, NLP for EHR automation and some genomics... a mixed bag with little clarity.
CureMetrix
Location: San Diego, USA
Founded: 2014
CEO: Kevin Harris
Funding: Undisclosed Series A
Product: cmAssist (in development)
“Through its research partnerships with leading hospital radiology departments, CureMetrix has evaluated more than 500,000 mammogram images to identify potential false negatives, which are undiagnosed cancers, and false positives, which are unnecessary recalls of patients to review anomalies that turn out to be normal.”
CureMetrix has a pilot partnership with Philips, UC San Diego and MD Anderson, and is developing a breast mammography CAD as an initial offering. It then hopes to extend this on a patented platform to other modalities, starting with CXR lung nodule detection. No word on regulatory status or product availability yet.
Deep Radiology
NO LOGO FOUND
Location: Santa Monica, USA
Founded: 2016
CEO: Unknown
Funding: Undisclosed
Product: Unknown
“DeepRadiology, a deep learning artificial intelligence medical startup is proud to announce the world’s first fully autonomous system for medical image interpretation. The system is able to produce final interpretations and reporting on medical imaging studies without the need for a radiologist.”
Wow… “the system is able to produce final interpretations” — a very bold (but year old) claim from a start-up! I have only one thing to say — good luck with FDA clearance for that! Then again, they are supported by Professor Fei Fei Li, a well-known researcher in the world of AI, and seem to have links to Stanford’s computer vision department. They showcased at RSNA in 2016, but I didn’t come across them. Their website is just a splash page, so no clues yet as to what they can actually do. My guess — this is hype.
DesAcc
Location: Bellevue, USA
Founded: 1993
CEO: Hugh Lyshkow
Funding: Undisclosed
Product: Data Continuity Engine
“DesAcc is a developer of health data migration software and interoperability tools, focusing on eliminating proprietary data silos and moving medical images and paper documents into standardised systems such as PACS and EHR’s. We also provide data migration services for PACS migration.”
Not a start-up! A very well established company that specialises in small and large scale data migration and continuity. Nothing on their website about machine learning, but my guess is that they are in a good position to leverage their 5000+ customers to provide big data access for AI.
DiA Analysis
Location: Be’er-Sheva, Israel
Founded: 2010
CEO: Hila Goldman-Aslan
Funding: $2,000,000
Product: LVIVO EF, SWM and Strain
“DiA’s flagship product LVIVO is an FDA/CE cleared, fully automated and objective cardiac toolbox that helps physicians and echo cardiologists analyses accurately and instantly Echocardiogram scans.”
With a three-part suite of regulatory approved ultrasound-based technology focused on ejection fraction, segmental wall motion and strain for echocardiograms, DiA has taken a niche of the cardiac ultrasound market, with customers across America and China. They say they are looking beyond cardiac ultrasound, and may well be showcasing some alternate tech this year.
Envoy
NO LOGO FOUND
Location: Unknown
Founded: 2017
CEO: Unknown
Funding: Undisclosed
Product: Unknown
Nothing. Zip. Nada. My detective work has me thinking that Envoy is the new company from TeraRecon, as they acquired McCoy Medical in June this year. McCoy was/is a marketplace that allows researchers to securely distribute trained machine learning algorithms to other researchers, as well as to hospitals and practitioners. They have a stellar advisory board (Dr. Eliot Siegel, Dr. Paul Chang and Dr. Khan Siddiqui), however, this one remains a mystery, for now…
Galileo
Location: Austin, USA
Founded: 2015
CEO: Abhijeet Pradhan
Funding: Undisclosed
Product: RadReport, RadCad (coming soon)
“RadReport is a Machine Learning based Clinical Decision Support (CDS) tool that seamlessly introduces quality control and efficiency into radiology workflow. RadReport uses a proprietary Bayesian Expert System (BES) to arrive at a probabilistic Differential Diagnosis (DDx).”
I like the sound of RadReport (again, one of the top 5 radiology AI use cases I have highlighted previously), and I’m interested to see a Bayesian network in action in a live clinical environment. No news on what modality or pathology their planned CDS system handles yet, and regulatory status is unknown.
HealthLevel
Location: Mountain View, USA
Founded: 2010
CEO: Parag Paranjpe
Funding: Undisclosed
Product: Foundations
“HealthLevel provides cloud based analytic solutions that enable healthcare administrators and physicians to easily understand and manage towards clinical, financial and operational performance improvements.”
Health Level provides business analytics and clinical metrics for radiologists, cardiologists and enterprise to improve efficiencies. Already working with UW Medicine Regional Heart Center they have an established footprint in the data analytics market.
HeartVista
Location: Los Altos, USA
Founded: 2006
CEO: Juan Santos
Funding: Undisclosed
Product: RTHawk
“HeartVista’s products promote rapid and robust cardiovascular disease diagnoses through a highly integrated, real-time focused software suite that contains the MRI tools needed for a comprehensive cardiac examination. The HeartVista platform, RTHawk, contains a collection of advanced MRI techniques that acquire high-quality images in a fraction of the time required by conventional MRI solutions.”
Backed by Khosla ventures, this 10 year old company is well established in cardiac MRI acquisition, and RTHawk is used around the world by academics. Nothing on their website about machine learning, so perhaps they have something new to demo?
iCAD
Location: Nashua, USA
Founded: 1984
CEO: Ken Ferry
Funding: Undisclosed
Product: PowerLook Suite
“The PowerLook Breast Health Solutions built on deep learning delivers powerful software solutions for breast tomosynthesis, breast density and 2D mammography. These innovative breast cancer solutions offer clinicians a wide range of tools for disease detection and analysis that enhance workflow and improve overall efficiency. PowerLook Breast Health Solutions support healthcare facilities worldwide in providing quality breast care to their patients.”
iCAD are already a global player in the CAD market, with their FDA-cleared mammography and tomosynthesis PowerLook products being used in many of the leading hospitals in America. They’ll likely have a new iteration out on display. Or perhaps they’ll be eyeing up the competition looking to buy?…
Kheiron
Location: London, UK
Founded: 2016
CEO: Peter Kecskemethy
Funding: Undisclosed
Product: Undisclosed mammography software
“Kheiron Medical is a medical imaging company that uses advanced machine learning technologies to develop and provide intelligent tools for radiologists, radiology departments, imaging centers and hospitals to improve efficiency, consistency and accuracy of radiology reporting.”
There’ll be peaked interest over this London based start-up at their launch. The NHS is a tough-market to crack, so being the only English company in this space at RSNA they are probably in a good position there. The CEO has a track record in teleradiology solutions, and they have backing from Entrepreneur First. They are launching their first mammography product this year at the conference, so watch out for announcements.
LPixel
Location: Tokyo, Japan
Founded: 2014
CEO: Yuki Shimahara
Funding: 8.11 Million JPY
Product: Undisclosed
“Our goal is to continue to be the world’s best provider in image processing and analysis technology for the life sciences. As LPixel, however, we want to be known more than just a life image processing and analysis company-we want to bring colossal change to the world of research.”
LPixel has focused on pathology & life sciences so far, with a platform for automated cell count detection on pathology slides, but states that it aims to enter the radiology market with regulatory clearance pending for it’s new product. Little information is available on their website, but there is a nod towards automated detection of cerebral aneurysms on MR. They are launching their medical image diagnostic support system at RSNA, so keep your eyes peeled.
Lunit
Location: Seoul, Korea
Founded: 2013
Funding: $5,478,910
Product: Unknown
“We are devoted to developing advanced software for medical data analysis and interpretation via cutting-edge deep learning technology. We envision a near-future when our systems would greatly help physicians make accurate, consistent, and efficient clinical decisions, not limited to diagnoses, through our data-driven imaging biomarker technology.”
The Lunit team have focussed on tuberculosis and data-driven imaging biomarkers in breast and lung cancer detection (as displayed last year at RSNA), and were listed as a top-100 start-up to watch by CB insights. They have since gone on to digital pathology, with tumour proliferation assessment algorithms. It will be interesting to see what progress they have made in the radiology space this year.
Mindshare Medical
Location: Seattle, USA
Founded: 2014
CEO: Michael Calhoun
Funding: $3,600,000
Product: Unknown
“Our analytics uses image pattern recognition algorithms combined with patient information through deep learning to build intelligence that assists clinicians and patients in making more effective and efficient clinical decisions.”
Another mysterious company with no official product on market, or radiologist in their leadership — maybe they’ll be demonstrating something tangible this year?
Quantib
Location: Rotterdam, Netherlands
Founded: 2012
CEO: Rudolphe Scholte
Funding: Undisclosed
Product: Quantib NeuroDegenerative & Quantib Brain software
“Our products will assist in automatically analyzing scans to aid in the diagnosis of several neurological diseases such as multiple sclerosis and Alzheimer’s disease. Our clients are major medical industries. Quantib is a spin-off from Erasmus MC in Rotterdam.”
Their existing product ‘Neurodegenerative’ provides quantification of MRI brain scans, and FDA-cleared ‘Brain’ provides segmentation, and they have a partnership with GE. They say they have more products in the pipeline for a range of relevant biomarkers for abdominal, MSK, neurovascular and neuro-oncological conditions, launching Q2 2018. A solid outlook, with strong in-roads to market already. One to watch!
Quantitative Insights
Location: Chicago, USA
Founded: 2010
CEO: Keith Tipton
Funding: $150,000 in total (declared)
Product: QuantX Advanced
“The QuantX-exclusive QI Score™ and Similar Case Database Compare™ provide similarity assessment to a robust database of lesions with known pathology based on biopsy results. QuantX has been shown to reduce missed cancers by 30% and improve reading time for complex multi-modality cases by more than 50%.”
This company broke new regulatory ground by going down the de novo FDA route for a never-before-seen classification of device for analysis of breast MR. This means that others can simply do a 510(k) using them as a predicate device. They have proven an improvement in breast cancer detection against single readers, and are rolling out an early adopter programme, so RSNA is the ideal nurturing ground for them to make new strategic collaborations.
Qure.ai
Location: Mumbai, India
Founded: 2016
CEO: Prashant Warier
Funding: Self-funded
Product: Unknown
“Applying deep learning to medical images poses a special set of challenges. At Qure.ai we’ve tailored deep learning algorithms to deal with these challenges and to make the most of the rich 3-dimensional information contained in medical images. One of our key focus areas is algorithm interpretability, ensuring that the reason for a suggested diagnosis is clear to a doctor.”
Qure.ai won NVDIA’s Social Innovation Award for its vision of AI driven affordable healthcare for medical imaging and diagnostics, and is a subsidiary of Fractal Analytics. They have a range of projects, ranging from tuberculosis screening on CXR, brain segmentation to digital pathology, but no details on their website of products or regulatory clearances. They focus on interpretability, and will be demonstrating their CXR saliency maps at RSNA.
RadLogics
Location: Piedmont, USA
Founded: 2010
CEO: Moshe Becker
Funding: $2,850,000 in total
Product: RADLogics Virtual Resident, Alphapoint
“Your RADLogics Virtual Resident uses machine learning image analysis to process the enormous amount of imaging data associated with CTs, MRIs and X-rays. Within minutes, a draft report — with key images — flows into your reporting system. You can focus your time and attention on diagnosis, improving quality for your patients and institution, and collaborating with colleagues.”
RadLogics actually hold the record for first FDA clearance in machine learning in radiology (not deep learning) for Alphapoint, a cloud-based system for the automatic analysis, review, and interchange of CT scans. There’s planned roll-out of support for other modalities, so expect to see some of those demonstrated this year at RSNA.
Realize
Location: Chicago, USA
Founded: 2016
CEO: Alex Risman
Funding: Undisclosed
Product: Unnamed — CT lung nodule detection
“Realize’s CT lung nodule detection application, available soon for investigational use, provides a sensitivity rate of 96% at 4 false positives per scan. Upcoming new applications include solutions for detecting abnormalities in chest X-ray, pulmonary embolism in CT angiography, and intracranial hemorrhage in head CT.”
Realize’s algorithms are available on the Ambra health platform for investigational use only (presumably no FDA clearance yet?), and they are launching several new algorithms to market soon. Expect demos for these, as well as image prioritisation, a well-needed functionality in radiology workflows.
Riverain
Location: Miamisburg, USA
Founded: 2013
CEO: Steve Worrell
Funding: Undisclosed
Product: ClearRead
“Riverain Technologies is dedicated to thoracic imaging, providing software tools to aid clinicians in the efficient, effective early detection of lung disease. With Riverain’s ClearRead applications, clinicians are empowered by tools that enable them to read faster and more effectively across the entire enterprise.”
With FDA-cleared software that suppresses bone and vessels, detects lung nodules, and highlights lines and tubes, VC-backed Riverain has avoided the cloud option, and installs locally within hospitals. They have a global customer base, and are clearly ahead of the pack in the chest imaging niche. But are they a one-trick pony? I doubt it, and hopefully they’ll be showcasing some new tech this year.
Subtle Medical
Location: Palo Alto, USA
Founded: 2017
CEO: Unknown
Funding: Undisclosed
Product: Unknown
“Subtle, Inc., is a medical imaging software company. Our goal is to improve the quality, value, and accessibility of medical imaging by significantly reducing cost, imaging time, and radiation dose, using deep learning algorithms.”
Minimal information on this brand new start-up (even their website misspells their name as Settle Medical). It appears to be a three man team thus far, and no specifics on product or capability. I’m guessing RSNA is their launch pad…
Visage Imaging
Location: San Diego, USA
Founded: 2009
CEO: Sam Hupert
Funding: Undisclosed
Product: Visage 7, Visage Ease
“Visage Imaging is a proven Deconstructed PACS® leader with best-in-class interoperability, enabling institutions to optimize their informatics investments in VNA and EMR, while also delivering the most technologically advanced enterprise viewer available. As an Enterprise Imaging Platform, Visage 7 also supports the viewing of non-DICOM and medical multimedia objects, as well as diagnostic mobile access with Visage Ease Pro™.”
A long-term player in the vendor neutral scene, Visage will likely be looking to incorporate AI into their offerings. Nothing on their website about deep learning algorithms, however they have announced they will be showcasing their new Visage 7 enterprise imaging platform at RSNA 2017. I expect this may include some ML features for auto-hanging protocols and the like.
Vuno
Location: Seoul, Korea
Founded: 2014
CEO: Yeha Lee
Funding: Undisclosed
Product: VunoMed
“VUNO is currently working with Korea’s largest hospitals to collect data and create an advanced classification index for diseases. With this data, VUNO will be able to provide products that will assist doctors in identifying and diagnosing diseases in patients.”
The first product ‘VunoMed’ calculates probability-based bone age on hand X-rays. They claim to be testing various algorithms that analyse medical images, signals and EMR data, so perhaps some of this tech will be showcased this year.
Zebra Medical Vision
Location: Shefayim, Israel
Founded: 2014
CEO: Elad Benjamin
Funding: $20,000,000 in total
Product: AI1
“Zebra’s mission is to provide radiologists the tools they need to make the next leap in patient care. Zebra is empowering radiologists with its revolutionary AI1 offering which helps health providers manage the ever increasing workload without compromising quality- at a flat, transparent $1 per scan.”
Out of all the medical imaging AI companies at RSNA this year, Zebra has certainly made the biggest splash in the main stream media by announcing a flat rate of $1 per scan analysed in their cloud. To date they have 5 FDA-cleared algorithms performing simple quantitation on CT scans, but with over 50 hospitals signed up around the world, and funding pouring in, they look set to expand their suite of tools, with breast imaging firmly in their sights.
Summary
Well, there you have it — a comprehensive list of AI companies showcasing at this year’s RSNA. There doesn’t seem to be a specific trend, in fact, there’s a real mix of established companies vs start-ups, with some going down the cloud route, others choosing to stay within hospital firewalls. Brain MR, chest CT, X-rays and mammography are common themes, as well as prioritisation and efficiency generators, plus regulatory clearances are coming fast and furious. Business models are also a mixed bag, with some being fairly mysterious about their capabilities and partners, and others going full PR and disrupting the payment system. The big players at RSNA (IBM, GE, Siemens etc) will certainly be keeping an eye out on who is making the right moves!
Final thought: one thing that stood out to me was some notable absences from the conference. Enlitic, Image Analysis, MedyMatch, VoxelCloud, Vida are all missing in action. Others, such as IcoMetrix and ClearView (now rebranded as Koios Medical) aren’t in the machine learning ‘arena’ but have a booth elsewhere. What this says about these companies is hard to tell…nevertheless, RSNA 2017 promises to be a machine learning frenzy judging by the programme, companies and speakers. I look forward to seeing you all there!
If you are as excited as I am about the future of AI in medical imaging, and want to discuss these ideas, please do get in touch. I’m on Twitter @drhughharvey
If you enjoyed this article, it would really help if you hit recommend and shared it.
About the author:
Dr Harvey is a board certified radiologist and clinical academic, trained in the NHS and Europe’s leading cancer research institute, the ICR, where he was twice awarded Science Writer of the Year. He has worked at Babylon Health, heading up the regulatory affairs team, gaining world-first CE marking for an AI-supported triage service, and is now a consultant radiologist, Royal College of Radiologists informatics committee member, and advisor to AI start-up companies, including Kheiron Medical.
