Back to the list
Congress: ECR25
Poster Number: C-28044
Type: Poster: EPOS Radiologist (scientific)
Authorblock: S. Sauranen, T. Mäkelä, T. Kaasalainen, M. Kortesniemi; Helsinki/FI
Disclosures:
Sara Sauranen: Nothing to disclose
Teemu Mäkelä: Nothing to disclose
Touko Kaasalainen: Nothing to disclose
Mika Kortesniemi: Nothing to disclose
Keywords: Artificial Intelligence, Computer applications, Radiation physics, CT, Physics, Technology assessment, Quality assurance
Methods and materials

A DECT phantom (Multi-Energy CT Phantom, Sun Nuclear Corporation, Melbourne, USA) was scanned using dual source CT (Somatom Force, Siemens Healthineers, Erlangen, Germany) with 90 kVp/Sn150 kVp and automatic tube current modulation at seven dose levels and with three repeats, resulting in 21 DECT images in total. The targeted dose levels were 1.8 (scanner minimum), 2.0, 4.0, 8.0, 16, 24 and 32 mGy; the realized doses were slightly lower. Aside from the dose levels, the DECT quality control scan protocol resembled clinical abdominal scan protocols. The DECT phantom comprised a water-equivalent body corresponding to average patient thickness as well as slots for inserts of different materials such as iodine and calcium, providing the possibility of including clinically relevant targets in the analysis. From the acquired DECT data, iodine maps and virtual monoenergetic images at 40 and 60 keV were reconstructed using the Siemens syngo.via post-processing software (Siemens Healthineers, Erlangen, Germany; version VB60A_HF05). AI noise reduction was applied to the data at different stages of the post-processing: either to the original X-ray tube A and tube B data prior to the DECT map calculation or directly to the DECT maps after their calculation. The three reconstruction schemes are shown in Figure 1.  

Fig 1: Reconstruction of the normal and AI noise-reduced DECT map versions.
The algorithm used for noise reduction was ClariCT.AI (ClariPi, Seoul, South Korea), a vendor-independent technology based on deep learning [4]. The contrast and noise of phantom targets of different sizes and materials were extracted from the noise-reduced versions and the version without additional noise reduction using a custom DECT QA tool and compared [5]. Contrast was defined as the mean CT number of the target and noise as the CT number standard deviation of the target. The inserts used in the evaluations had iodine concentrations of 2 to 15 mg/ml and diameters of 2 to 10 mm at 5 mg/ml, as well as calcium concentrations of 50 to 300 mg/ml. In addition, background regions-of-interest (ROIs) were defined for noise calculations.  Based on the evaluation of autocorrelation between adjacent slices, only every 7th slice of each repeated measurement was included in the comparison to achieve sufficient sample independence. The contrasts were extracted from the calculated maps and the noise from the subtraction images of two repeated measurements. Hence, all three repeated measurements were used for the contrast comparisons, while for the noise comparisons, only one-third of that (a subtraction of two repeated measurements) was available.
GALLERY