Deep Detection, new IFAE spin-off that offers new industrial inspection solutions combining X-ray detection and artificial intelligence

On November 3, 2020, the Institut de Física d’Altes Energies (IFAE) became a partner of the company Deep Detection S.L. and the technology transfer from IFAE to the new spin-off of the center became effective. Deep Detection is composed of a scientific team made up of IFAE researchers Dr. Mokhtar Chmeissani and Dr. José Gabriel Macías, and a business team formed by David Ciudad (CEO), and aims to offer detection solutions to the industry thanks to a multispectral X-ray camera that includes a photon-counting technology capable of detecting hitherto undetectable foreign bodies, such as plastics, wood or bones.

Deep Detection is based on a technology developed at IFAE in the Medical Imaging group led by Dr. Mokhtar Chmeissani with the aim of creating medical and industrial applications derived from the knowledge obtained at IFAE in the development of detectors for particle physics experiments.

The developed system is made of chips specially designed to perform x-ray quality control tests integrated in a spectral camera for in-line scanning. The technology has been developed within the framework of a project Retos Coloaboración Collaboration (ERICA) of the MICINN and an AGAUR “Producte” grant (LINDA). The chip was assembled entirely in the IFAE’s microelectronics laboratory.

Chip developed at IFAE for X-ray quality control tests Credit: IFAE

“With the existing technology so far, we could either get a low-resolution x-ray image with good spectral information, or a high-resolution x-ray image with very little spectral information. At IFAE we have solved this problem. The technology we have developed allows us to use X-ray detectors with small pixels, to achieve a good spatial resolution, and generate, at the same time, images with different energy levels”, explains Mokhtar Chmeissani, an IFAE researcher.

“The spectral cameras we have developed allow us to obtain images of excellent quality, in multiple channels, of objects passing at 60 meters per second through a production line. The cameras can count up to 10 million photons per mm² every second and this allows us to obtain, in each channel, the image quality needed to discriminate and identify materials of very similar densities such as plastics in yogurts with fruit or cereals or uncalcified bones in filleted chicken, two of today’s most complex food security challenges”, adds José Gabriel Macias, an IFAE researcher.

The technology was submitted to the acceleration program The Collider (MWC) and was one of 15 selected for the validation phase of the business opportunity competing with more than 200 projects throughout the state. It was finally chosen as one of the 5 business projects for the venture building phase, with the consequent creation of the company Deep Detection S.L. on July 13, 2020, with an initial investment of € 50,000 from the MWC foundation, as a partner in the company.

Spectral images obtained with the camera developed at the IFAE Credit: IFAE

Applications in the food industry

X-ray inspection is used in the food and beverage industry to carry out quality controls and ensure their safety, specifically in the detection of foreign bodies such as metals, glass, wood, etc. All materials that have high densities are likely to be detected with current technology, but not when it comes to low density materials.

“The big challenge of X-ray inspection in the industry is the detection of plastics, and we can provide a solution for it. In fact any low density element is an industrial challenge and a current problem. Plastics in yoghurts, bones or amount of fat meat … they are problems that we can solve, “says David Ciudad, CEO of Deep Detection.

The key to the technology solution is in the combination of X-ray imaging and artificial intelligence, as Colin Burnham, COO of Deep detection, points out, “The benefits of Artificial Intelligence or deep learning can only be obtained if the quality of the captured data is high enough for the algorithms to differentiate key patterns. Conventional X-ray detectors cannot handle complex detection problems because the images they generate are too noisy.The team at IFAE has developed a solution from scratch to capture high-definition X-ray data, so we can apply the latest artificial intelligence strategies to improve the quality and safety of manufactured products.”