At ECR 2026, Getting up to speed: emerging technologies in cross-sectional imaging covered three advances that are reshaping what CT, MR and PET/CT can deliver in routine practice. The talks moved from detector physics to reconstruction software and whole-body sensitivity, with a consistent focus on clinically usable gains in image quality, speed and dose efficiency.
Photon-Counting CT Brings Native Spectral Detail
Victor Mergen (Zürich, Switzerland) described photon-counting detector CT as a detector-level change that improves dose efficiency, spectral capability and spatial resolution. He explained the signal chain directly: “The incident photon is transformed into an electrical signal in a direct conversion process.” Because pulse height reflects photon energy, thresholds can suppress electronic noise, and he noted that “only pulses with a height larger than a threshold energy are counted… thereby eliminating the electronic noise.” With multiple thresholds, photons can be sorted into energy bins, making spectral information intrinsic to the scan.
He linked these technical properties to clinical use. Ultra-high spatial and temporal resolution can support cardiovascular imaging, including reduced calcium blooming and improved evaluation in heavily calcified vessels and stents. Spectral output enables virtual monoenergetic imaging, which can be tuned for different goals, including contrast optimisation at lower keV and artefact reduction at higher keV. He also outlined thoracic and abdominal opportunities tied to low electronic noise, potential dose reduction and improved depiction of subtle findings, while keeping expectations cautious on how much earlier detection will translate into downstream outcomes. He highlighted metal artefact reduction through monoenergetic selection and described future directions such as K-edge imaging and novel contrast media, but stressed the timeline for translation: “the integration of such novel contrast media into routine clinical practice will still take many years.”
Related Read: Deploying Radiology AI That Works, Helps and Endures
Deep Learning Reconstruction Compresses MR Scan Time
Martin John Graves (Cambridge, United Kingdom) positioned deep learning reconstruction as the latest acceleration step after parallel imaging and compressed sensing, aimed at reducing acquisition time without losing diagnostic content. He set the priority clearly: “we’re going to sort of primarily emphasise the reduction in acquisition time here, the speed ups that we can obtain with this method.” Building on the basics of undersampling artefacts and iterative reconstruction, he explained how neural networks are trained on pairs of lower quality and higher quality images, then applied to suppress noise and artefacts while preserving acquired information through a data consistency step.
Across examples, he highlighted two linked outcomes: faster scans and improved apparent image quality, including an increase in effective signal-to-noise and sharper spatial definition in some settings. He also underscored that vendor implementations vary and that these tools sit inside regulated clinical products, so radiologists need to understand how reconstructions behave across protocols. He concluded with a balance of benefit and caution. In his words, “the deep learning is really giving us an improved effective signal to noise ratio,” but he warned that “we do need to be very vigilant,” emphasising ongoing validation to ensure reconstructions do not obscure relevant findings.
Long-Axial Field-of-View PET/CT Enables Low Dose and New Protocols
Clemens Mingels (Sacramento, United States) described long-axial field-of-view PET/CT as systems with extended axial coverage and much higher sensitivity than conventional scanners. He offered a practical definition: “any scanner over 1 metre axial field of view would be classified as a long axial field of view PET CT scanner,” with total-body systems extending beyond 188 cm, enabling imaging from vertex to feet in 95% of the population. The clinical implications, he argued, start with injected activity. “Dose reduction is a main clinical application of long axial field of view PET CT,” supported by higher sensitivity and a more homogeneous sensitivity profile when more of the patient sits within the detector field.
He mapped that capability onto clinical scenarios where dose and motion matter, including paediatrics and pregnancy and patients who cannot tolerate long imaging times. He also discussed ultra-short protocols enabled by sensitivity, alongside the limits that appear as dose decreases, since background noise rises and can affect lesion quantification, particularly in delayed imaging. Beyond routine staging, he highlighted opportunities in delayed studies to leverage clearance and contrast, and in dynamic imaging with kinetic modelling when the whole body remains in view during tracer passage. He closed with an implementation reminder: “it’s not a one man show.”
The session showed three different routes to similar goals: more actionable information with less burden on patients and services. Photon-counting CT turns spectral data and ultra-high resolution into native output while pointing to longer-term potential in K-edge applications. Deep learning reconstruction in MR shifts acceleration into software, improving speed and effective image quality while requiring careful clinical oversight. Long-axial field-of-view PET/CT converts sensitivity and coverage into lower-dose options, shorter and delayed protocols and new dynamic approaches, with quantification limits that need attention as injected activity drops.
Source & Image Credit: ECR 2026
