NOTE! This site uses cookies and similar technologies.

If you not change browser settings, you agree to it. Learn more

I understand

Learn more about cookies at :

June 17, 2019 | Electromagnetism used to study the mechanical properties of steel

List acier cnd 250Researchers at List, a CEA Tech institute, have developed simulation models and tools that leverage electromagnetism to characterize steel. These models and tools will be integrated into the CIVA platform so that they can be used to develop industrial non-destructive testing processes.

The mechanical properties of steel are closely related to the steel’s microstructure. The same is true of steel’s magnetic behavior. Modifications to the microstructure can affect the material’s mechanical and magnetic properties.

So, studying steel’s electromagnetic behavior can provide indirect insights into its mechanical properties—but only if other factors that cause variations in the magnetic properties of the metals that make up the steel are taken into account.

List, that earned the Carnot seal in 2006, developed a simulation model of an industrial non-destructive testing process that uses macroscopic measurements of steel’s electromagnetic signature. The idea is to take into account the related sources of variability and, especially, the types of probes used and the material’s heterogeneity. The method leverages an original semi-analytical formulation of the problem and non-linear models to rapidly arrive at a robust solution.

These tools will ultimately be integrated into the CIVA platform so that they can be transferred to industrial users. Whether it is to monitor structural health or control quality during manufacturing, inspecting the microstructure of steel is a major challenge in the nuclear, automotive, and other industries.

Read article at



June 6, 2019 | Automating toxic chemical identification

CEA List Identification automatisee de toxiques chimiques 250List1, a CEA Tech institute, partnered with the French National Center for Scientific Research (CNRS) to develop an algorithm to automatically identify toxic chemicals and explosives in real time. The research was part of a French government CBRN-E2 counterterrorism program.

The Police Nationale, one of France’s national police forces, turned to List, a CEA Tech institute, to develop a software application to automatically identify toxic chemicals and explosives in real time. The software leverages a support vector machine (SVM), a sophisticated tool used in artificial intelligence techniques just like neural networks.

Unlike neural networks, however, SVMs are optimal classifiers. In other words, they can determine with a high degree of confidence whether or not a chemical compound is present in a mix.

Any toxic compounds present have to be identified one after the other. So, the researchers developed a smart technique to preselect areas of interest in the FT-IR3 spectra of compounds included in a database of threats. These areas of interest are then inspected within the spectrum of an unknown sample. The information provided to the SVM is pre-calculated and simplified using conventional signal analysis methods.

Another original aspect of the software is that the SVM learns from theoretical spectra, eliminating the need for it to learn from huge databases. A tool developed by List can automatically generate, from just a few real spectra, theoretical spectra of chemical mixes from the thousands of molecules.

Blind tests revealed excellent selectivity and sensitivity (to within 4% or 5% for chemical mixtures in powder form). These methods, called Peak Correlation Classification (PCC), were patented by List.

[1] List earned the prestigious Institut Carnot seal in 2006 (Institut Carnot TN@UPSaclay)
[2] Chemical, biological, radioactive, nuclear, and explosives
[3] FT-IR, or Fourier Transform Infrared Spectroscopy

Read article at



May 23, 2019 | Medical scans: reducing doses without compromising on diagnostic quality

scan doseo 250List*, a CEA Tech institute, developed a mathematical model to improve medical imaging protocols so that the lowest possible dose of ionizing radiation can be used without negatively affecting the reliability of the diagnosis.

The level of exposure to radiation during medical scans has been on the rise for a number of years, creating increased health risks for patients. Researchers at List, a CEA Tech institute, have developed new indicators to keep the dose of radiation delivered to a patient during a scan to a strict minimum. The first is used to estimate the dose received by the patient as reliably as possible; the second evaluates the quality of the images produced.

The researchers began by developing and validating a model of the scanner at the DOSEO platform so that they could evaluate a breakdown of the distribution of the dose of radiation received by the patient's body during a scan depending on the equipment's characteristics (geometry, movement of the radiation head, etc.). A Monte Carlo method is used to accurately simulate the trajectory and interaction of particles in matter.

Next, a mathematical observer model was used on a torso phantom filled with water and inserts of various shapes and sizes to assess the image quality. A detectability indicator that reflects the capacity to detect or discern a lesion was used for the assessment. The results were compared to an analysis completed by radiologists assisting with the project so that the model could be validated. The model has already been used to correlate an artificial lesion detectability rate with the dose of radiation received by the patient.

Ultimately these advances could provide radiologists with medical imaging methods that successfully limit radiation doses without compromising patient diagnoses.

*List earned the prestigious Institut Carnot seal in 2006 (Institut Carnot TN@UPSaclay).

Read article at





May 9, 2019 | The “smart eye” for construction sites goes public

Arcure bourse 250Arcure, founded in 2009, develops 3D vision systems for industrial and construction vehicles with the support of List, a CEA Tech institute. The company recently completed its initial public offering. We spoke to List* artificial intelligence, language, and vision expert Patrick Sayd, who develops the image-analysis-based pedestrian-detection technology Arcure uses in its products.

1) Tell us more about Arcure.

Arcure was founded in 2009 to commercialize List's image-processing technologies on the industrial vehicle safety market. Our original research focused on the detection of vulnerable "targets" like pedestrians and cyclists for automotive applications. Arcure's founders came up with the idea of developing similar solutions to make industrial vehicles and robots safer and more autonomous. When Arcure was founded, we immediately set up a joint lab to adapt the technologies we had previously developed for the automotive industry to the industrial vehicle industry and create an ongoing technology-transfer process that we are still using ten years later.

Our effective cooperation produced Arcure's Blaxtair smart pedestrian detection system, which secures the "danger zone" around industrial vehicles in a variety of situations, including the most extreme, from construction sites and mines to logistics facilities and factories.

The system, installed on board the vehicle, includes a smart unit equipped with two cameras and a computer with image-analysis software that scans the environment to detect potential collisions with vulnerable targets and alerts the driver or operator in real time. The system's stereoscopic vision makes it possible to generate a 3D reconstruction of the environment around the vehicle, differentiate between pedestrians and other obstacles, and accurately locate pedestrians to within just a few centimeters.

Arcure recently added a new product to its lineup. The Omega 3D vision sensor, built on Blaxtair technology, was developed specifically for manufacturers and integrators of smart systems for Industry 4.0.

2) What makes these solutions different from those developed for the automotive industry?

We made the technologies List initially developed for the automotive industry more robust so that they would be able to analyze more complex environments than roads, which, despite the fact that they are populated by many moving users, are fairly well-organized. On a construction site, the vehicle's surrounding environment contains many obstacles of different kinds, such as people, equipment, and materials. The terrain can also vary substantially, from factories to mines. Finally, the vehicles themselves are more complicated than road vehicles, with the ability to rotate or tip, for example.

So, given these complexities, the system has to be able to identify dangerous situations robustly enough to prevent collisions, but without triggering nuisance alerts. It also has to factor in the vehicle's overall volume so that it can focus in on areas of immediate danger.

Blaxtair has carved out a position as the number-one pedestrian detection technology. Global market leaders that either use the technology or integrate it into the vehicles they manufacture—effectively responding to their need to ensure safety around their machines—have rolled out the technology in more than 30 countries. Blaxtair is even available as an optional feature from Jungheinrich, the world's third-leading forklift manufacturer.

3) Now that the company has gone public, where will it be seeking growth in the future?

Now that Arcure has gone public, its management would like to speed up development, penetrate key markets like Germany and the United States, and build partnerships with the world's leading equipment manufacturers. The company also plans to release Omega, its new high-performance 3D vision system for industrial automation professionals.

Arcure reported revenue of €7.4 million in 2018, with 54% of the company's sales outside France, and revenue growth of 50% from the previous year. Today, the company has set the ambitious revenue target of €60 million by 2023.

Finally, Arcure is pursuing R&D to support the rollout of the technology, first by optimizing the hardware and algorithms to bring costs down and, second, by improving Blaxtair technology's performance so that larger and more complex environments can be analyzed with the smallest possible number of sensors.

*List earned the prestigious Institut Carnot seal in 2006 (Institut Carnot TN@UPSaclay).

Read article at



April 16, 2019 | AI lightens the burden of paperwork on hospital administrators

docte gestio 250DocteGestio, a major private-sector health and human services company in France, and List, a CEA Tech institute and member of the Carnot Network, initiated the DIM-IA project in late 2018 with the goal of leveraging artificial intelligence to lighten doctors’ workloads.

With 7.7 million hospital admissions per year in France and an average of three medical procedures per patient, the doctors in charge of hospital medical records departments have their work cut out for them: They must reread and code each document generated by each patient consultation. And if the medical records are not coded, patients’ electronic files cannot be updated and the national health insurance program cannot reimburse the costs of their care.

Artificial intelligence to support doctors

Powerful semantic analysis algorithms leveraging natural language processing can be adapted to medical language so that all, or at least some, of this coding work can be automated. DocteGestio and List are trying to do just that. The first use case their joint project is addressing will be to automate the coding of hospital records.

The partners developed a demonstrator using the databases of hospitals managed by DocteGestio and List’s semantic analysis platform. They were able to confirm the benefits of an AI-based coding solution for medical professionals. The final prototype is now being tested by a cohort of medical records department heads at DocteGestio hospitals. The tests will be used to gather feedback and integrate any changes needed before the solution can be scaled up.

List Director Jean-Noël Patillon said, “Our partnership with DocteGestio has given us an opportunity to bring our knowledge of artificial intelligence to the healthcare sector. This kind of technology has the power to revolutionize hospitals by freeing professionals from lower-added-value tasks. Ultimately, the DIM-IA project will allow doctors to complete their administrative tasks more efficiently and substantially increase overall productivity.”

The demonstrator was presented to France’s national health insurance administration (CNAM), which has since expressed an interest in the project, in particular to support the nation’s shared health record initiative. For shared health records to work, access to the contents of patient records needs to be easier, and semantic analysis technologies could help achieve this.

More time for high-added-value activities

Beyond the first use case the partners are investigating, DIM-IA is aimed at freeing medical professionals’ time for higher-added-value activities (research, quality assurance, safety, improvements to the organization) at both hospitals and the agencies that oversee them.

And, in addition to lowering the costs associated with collecting and verifying patient information, the solution will also make the medical billing chain more robust and reliable.

DocteGestio CEO Bernard Bensaid said, “For us to be engaged in this project with a high-level scientific partner like List is major. It will position our company in the healthcare R&D space and create synergies with what is happening in AI R&D. We are already using collective intelligence. Today, we want to bring artificial intelligence to healthcare in France. I feel it has the capacity to substantially improve the data processing chain and lower government spending.”