50,000. That is the number of images a radiologist can see every day. A figure almost impossible to grasp or believe, processing such a large amount of data just exceeds the capacities of the human brain. But medical images volume has been growing at a hyper-fast pace, in a combined effect of technology advances, Western world population ageing and more systematic use of imaging, not only for diagnosis but also for screening, monitoring, minimally invasive surgery… Moreover, imaging today involves quantifying, measuring, and many repetitive tasks. Nowadays, the number of images per exam can reach 1,000 to 2,000 images, that is… 40 times more than 30 years ago, allowing physicians to see more details, more resolution, earlier…
The problem is that this ability to produce data is growing faster than our ability to process them. Radiologists are at risk under this overwhelming flow of information, to miss something, to look more at their screen than at their patients, to be used as reading machines and, eventually, to get sued if they miss something, and burnt out, facing an ever-growing workload.
Thankfully, as technology progresses, so does computing power, allowing artificial intelligence and deep learning algorithms to produce, for the first time, concrete advances in healthcare. Computers which can generate images at a vertiginous speed can also analyse them at such pace, without being subject to distraction, fatigue or stress. They can detect abnormalities – presumptions of cancer or vascular accidents – through signals almost invisible to the human eye. They can recognise patterns in data and make a correlation to hypothesise a diagnosis, using the vast amount of existing knowledge, only like expert physicians can do. And so, the prospects for public health are immense, potentially leveraging levels of care to the highest. Automatic image processing and analysis can free up radiologists to spend more time with their patients and focus on complex cases. It can also enable access to medical imaging in areas where there is no expert physicians to interpret them. It is estimated that today 4.7 billion people across the world, simply do not have access to medical imaging.
Looking closer, the real challenge today will be to smoothly integrate these algorithms in the clinical workflow, without disruption, or stressful adaptations. Physician’s training will be of utter importance, and so will be easy-to-use interfaces. We are at the dawn of sudden acceleration in the implementation of technological solutions.
Today, one hundred applications, more or less advanced, using artificial intelligence exist or are under development in the world and USD 800 million are currently invested in this field. Navigating in this rapidly growing market requires time and expertise. Hospitals cannot do it on their own, nor can they hire dozens of small companies for each of their specific need.
Since its creation, just a year ago, Incepto has been bringing together physicians, developers and data scientists to create and identify meaningful applications that deliver clinical value. These applications need a new but yet crucial mix of skills:
a) advanced medical expertise: the central link in the chain will remain the doctor. As a user first, able to determine the best use cases and validate the clinical utility of an innovation. But also co-producer of new applications.
b) advanced expertise in mathematics, computer development and data science, being able to attract and facilitate the work of the very best coders and computer scientists.
c) business and entrepreneurial expertise – if artificial intelligence is omnipresent in the public debate, it must be understood that the advent of a revolution in medical imaging, and, therefore, in medicine as a whole, depends on mastering a much more complete process. It requires quality data, identified and duly qualified by doctors, treatment capabilities (powerful equipment and algorithms), but also communication tools (information systems, cloud solutions) and, lastly, ergonomic platforms to enable doctors to use the tools in a simple way. All of this under economically viable conditions for all actors. Therefore, a new business model must be invented and federate very heterogeneous actors, which requires a sharp knowledge of the economy of medical imaging and the health sector.
At Incepto, this amalgam of a new genre is embodied by our three founders: Gaspard d’Assignies, MD, radiologist, PhD in quantitative imaging and co-founder of the SFR-IA working group; Florence Moreau, a leading engineer, specialised in cloud-based medical imaging software; and Antoine Jomier, business leader, former imaging sales manager for GEHealthcare in France, specialist in high growth innovative activities.
Around the three founders, Incepto has gathered a team consisting of experts from the three fields: medicine, technology and industry.
Incepto’s mission is twofold: to help physicians to identify, access and use the best-performing solutions; and to co-create new applications with them, tailored to their specific needs. Incepto is both, a distributor and producer of clinically-relevant solutions.
As a producer of solutions, Incepto has already signed three contracts with major institutions and leading clinicians in Europe on the following topics:
1. Automate the production of knee x-ray and MRI reports.
2. Assist in emergency therapeutic decision for bowel occlusion
3. Automate and standardise the surveillance of vascular aneurysm before and after endovascular treatment
As a distributor of solutions, Incepto has agreements with four leading start-ups (mammography – ScreenPoint; thorax imaging – Qure.ai; neuro MR imaging – Icometrix; lung screening – Aidence) and more than 38 requests for installation after just 2-month activation.
This unique “alliance” will lay ground for a new interdisciplinary way of working, and will also be one of humans and machines, of human intelligence and artificial intelligence, that will help physicians in their daily mission: help saving lives.
Contact Person: Igor Dniestrowski
Incepto will be part of the ECR 2019 Artificial Intelligence exhibition in hall X1.
This article was provided by Incepto and edited by the ESR office.
