Log in / Register
A web-based framework for image quality assessment in photon-counting detector (PCD) computed tomography, built for both clinical and research use. Run standardized, reproducible evaluations across scanner configurations and acquisition protocols.
Register each scanner, schedule its tests, and let QAF track what is due, what has drifted and what has failed. Every result is judged against your acceptance criteria and stored permanently with the operator and date — so you can show how a system has performed, not just how it performs today.
Eight measurements, each resolved per energy bin
Centre and four peripheral regions placed automatically from a single traced circle, sized to the recommended area and kept clear of the phantom edge. Measured in Hounsfield units with per-energy-bin tolerances and a per-region diagnostic breakdown.
A structured visual check for ring, streak, shading and blooming artefacts. Each review records the image type examined, the findings, the reviewer and the timestamp, building a permanent log suitable for daily use.
Mean, standard deviation and signal-to-noise measured per region and per energy bin, averaged across the series with dispersion reported. Judged against a baseline tolerance or an absolute minimum; perfectly uniform regions are reported as undefined rather than zero.
A background region sets the reference level and the noise; any number of signal regions are then compared against it, per energy bin. Supports a baseline tolerance or an absolute minimum such as the common low-contrast threshold.
One-, two- and three-dimensional noise texture with peak and average frequency as summary metrics. Multiple regions averaged for stability, resolved per energy bin, with normalisation verified against a known analytical identity.
Derived from point sources, wires or a high-contrast edge. Reports full width at half maximum together with the spatial frequencies at 50%, 20% and 10% modulation, per energy bin and averaged, with baseline evaluation.
How a material's response varies across energy bins and across concentration, in Hounsfield units or linear attenuation. Air and water referencing is built in for systems that do not provide calibrated values.
Each tissue-surrogate insert measured against its known reference value, with the region sized automatically to the recommended fraction of the insert so the boundary is never included. Tolerances are set by the physicist.
Everyday utilities that support the analyses
Identify materials by comparing density predictions against ground-truth energy images. Reports accuracy, precision, recall and F-score per region, with an adjustable classification threshold.
Determine concentration from calibration samples of known composition. The calibration relationship is built automatically from measured regions, then applied to quantify unknown regions.
Frequency distribution of values within one or more regions, for a single image or averaged across the series, with adjustable binning, summary statistics and export.
Intensity along a drawn line, with adjustable width for smoothing. Used to examine edges, boundaries and gradients, with multiple profiles comparable on one plot.
Convert attenuation values to the Hounsfield scale from water and air references, per energy bin. Calibrate by drawing regions or by entering values directly; the result is shared across every module.
Mean, standard deviation, minimum and maximum for any drawn region, reported across every energy bin. Rectangle, circle, line and polygon regions, with sizes specified in millimetres.
What surrounds the analysis