Introducing RapidCHECK
As the name implies, the idea of RapidCHECK is to have a fast and thus necessarily automated analysis of CT images. Although the software works with three different types of phantom (the IQphan Phantom, the CT ACR 464 Phantom and the Advanced Electron Density (AED) Phantom, our only application is AED.
In principle, the AED is similar to our CIRS 062 with the important difference that the AED rods have coding holes on one end where the code identifies the material of the rods. RapidCHECK read the code of each rod, identifies the material, evaluates the HU in a certain 3D volume of interest, and constructs HU-to-Electron-Density and HU-to-Mass-Density curves1.
(120kV curve in the "high" range up to titanium)
If, during the analysis, one or more rods were not identified correctly, the data table gives the possibility to "help the software". This happened in location 1 and 3:
(120kV "high" results. The ruler icon means that density values are taken from a calibration certificate.)
Location numbering is according to the following rules: 1 = center, 2 = halfway down, 3-10: inner circle clockwise, 11-16: outer circle clockwise:
(Location numbering in the AED phantom.)
How We Use the Software
The software is intended to be used primarily on the CT scanner (goOpenPro) and later also on the TrueBeams (CBCT). After setting up the institution and the CT machine, for each kV setting we defined a QA Task in the "low" density range (up to Cortical Bone) and one in the "high" density range (up to Titanium):
In the "high" setup, the Solid Water rod in location 13 is replaced with the Titanium rod. The scanner protocols also slightly differ: the iMAR metal artefact reduction is active in the "high" scans.
Two setups with eight energies each seems to be a lot of work, both on the scanner side and during the analysis. But it is not: two scan protocols are used on the goOpenPro, each with the predefined eight spiral scans. The user aquires a topogram, sets the start and stop positions, copy this setting to the other seven scans, and presses Go. After eight scans, the room is entered once, one rod is swapped, and the procedure is repeated.
When the scanner has finished its reconstructions, all image series are in the moment exported to Eclipse. From there, the images are exported as files, copied on a USB stick, transferred to the RapidCHECK laptop (see Part 2), and analysed.
There are several options for creating output in RapidCHECK: PDF report, CSV export, XML export. For simple QA requirements, we generate PDFs (here for the "low" scans of April 14, 2026: 70kV, 80kV, 90kV, 100kV, 110kV, 120kV, 130kV, 140kV).
Rod Identification
On the tip of each rod, several (between 3 and 13) holes are drilled (1 cm deep), which both define a coordinate system and a bit pattern, see below:
The resulting code (here: 163) is used to identify the rod material. Column 1 in the Rod Material Editor lists most of the codes:
(Rod Material Editor. The "measured" values are from the phantom certificate.)
After the first few scans on the goOpenPro, we observed that the rods are not always identified correctly by the software. The rods which fail more often are either at certain locations (1, 2), have low density (LN300 Lung) or high density (108 - Cortical Bone). This can be corrected, but it requires the user to be confident about the true material used at a certain location.
We adapted our paper form which we used to document the rod configuration of the CIRS pantom for the AED phantom:
(Simple sheet for the documentation of the rod configuration.)
The reason why the identification sometimes fails is unclear. There are certain possibilities to "fine-tune" the analysis geometry via Image Registration, such as small rotations in the phantom setup, but also the exact slice which is used for Rod Identification. Sometimes this helps, more often it doesn't.
Maybe the ring artefacts play a major role in "disturbing" the analysis, especially in location 1, on the phantom axis. We also tried several slice thicknesses (1mm, 2mm), without effect. It sure is no simple image processing task to locate the coding holes, especially if the material density is low, like in the LN300 Lung rod: the denisty difference between the rod material itself and the holes drilled in it is minimal.
In any case, this is not a major problem, since most of the time, only 3 or 4 rods are affected, and the material assignment can be corrected.
We won't discuss the various trending possibilities, because the number of scans performed so far is too small.
Conclusion
RapidCHECK seems to be a valuable tool for the fast analysis of AED scans. We therefore plan to implement a scheme of constancy checks for the goOpenPro, and currently discuss the frequency of checks. Especially if several kV settings are in clinical use, the time saving potential is significant.
Notes
1 Since we completely gave up AAA some time ago, we are only looking at Mass Density.