X-ray medical imaging. Image quality

This article looks at the complete X-ray radiography chain and the quality of the final image. Components other than the detector contribute to image quality: mainly the source and the anti-scatter grid. And quality must be measured not by considering a noise image (which a detector's QED does), but by observing patient X-rays: the intrinsic characteristics of the pathologies the radiologist is looking for in the image, the patient's bodily characteristics, and the observer's mental decision-making process all come into play. The first part of this article looks at the modeling of all these parameters, in order to arrive at a figure of merit that can be used to estimate the quality of a radiology system as a whole (considered from the point of view of image quality). Noted d ′, this merit factor is called the "detectivity" index. It can be calculated on the basis of objective elements describing the X-ray chain (detector QED, X-ray source characteristics, etc.), but it also makes assumptions about the observer's performance: as there is no universal model to describe it, the article distinguishes between different practical cases (ideal observer, semi-ideal, etc.). An extension of the concepts of DQE, MTF and NPS is also deduced, which quantify the performance of a complete radiology system, but without taking observer or pathology into account.

The remainder of this article is devoted to three technological breakthroughs currently under consideration, which some experts believe will significantly advance X-ray imaging from the 2020s onwards: spectroscopic imaging by integration, then by photon counting, and phase contrast imaging. Spectroscopic imaging by integration (or multi-energy detection) has been known and implemented since the 1990s. As for counting and phase contrast, there are already practical applications in 2015 (in some prototype CT scanners for multi-energy counting, in synchrotron light sources for phase contrast), so these are not exploratory research topics. However, the difficulties encountered in adapting them to conventional radiology are considerable, and these technologies should be considered as being at the applied research stage.