Deep Detection Raises €1m Bringing Defect Detection X-ray Cameras To Market

Deep Detection, Barcelona, Spain, who build multi-spectral X-ray cameras for quality control and product safety in manufacturing has raised €1m in funding. The company intends to use the funds to accelerate industrialization of PhotonAI technology in key market sectors. Deep Detection is a spin-off from IFAE (Institut de Física d´Altes Energies). IFAE developed the deep science behind the PhotonAi camera series that led to patented solutions in photon counting X-ray detectors.

Founded in July 2020, Deep Detection is preparing to launch its first product, PhotonAi, an X-ray camera that fits into X-ray machines to inspect a product and ensure it is safe, contains no foreign materials and meets production standards. First commercial sales are anticipated in 2022 after completing a series of industrial PoCs with international corporations to validate the technology in each market sector.

Deep Detection builds its detection system on top of a proprietary ASIC, a microchip designed to bond with ‘high Z’ detectors that convert incoming X-rays into electrical pulses, which can be counted and measured for energy level. In this way, 20 million photons are binned every square millimeter every second during inspection. Onboard processing collates the data to produce a multilayer image in the X-ray band.

Photon counting fundamentally changes the way in which X-ray information is captured. For 120 years X-rays have been captured on photographic plates or photodiodes working in the visible spectrum. That means that incoming X-rays are converted to light and then measured indirectly. The light is generated by a scintillator which spreads the signal over many neighboring pixels, sacrificing both resolution and contrast. Secondly, the signal has to be integrated and treated as one energy band, hence, we mostly see X-ray in black and white.

An application specific integrated circuit is needed to count tens of millions of photons while operating alongside special algorithms to maximize the good count, minimize the lost counts and compensate for particular phenomena as charge sharing across pixels or pile up of photons. Only this combination of readout electronics and mathematical transformations can preserve the image resolution and open the door to spectral resolution. Colour is just energy, so spectral resolution means classifying the photon energy into ‘color’ bands, measured in KeV to build up multi-channel images, like RGB.

Deep Detection´s microelectronics and patented algorithms ensure this can happen in every pixel to produce multi-channel images at practical production speeds and computing power. Since every pixel is working simultaneously a complete camera can output a multi-channel image frames line by line which are perfectly registered down to the pixel.