Click here to


Are you sure ?

Yes, do it No, cancel

Impact of Reconstruction Kernel and Iterative Reconstruction On Coronary Calcium Scores

M Barreto*, P Schoenhagen , F Dong , The Cleveland Clinic, Cleveland, OH


(Sunday, 7/29/2018) 3:00 PM - 3:30 PM

Room: Exhibit Hall | Forum 8

Purpose: Coronary calcium scores are used to predict risk of future major adverse cardiac event (MACE). This study investigates the impact of reducing dose with iterative reconstruction on coronary calcium scoring.

Methods: An anthropomorphic calcium scoring phantom with 9 Calcium Hydroxyapatite inserts of 3 densities (200, 400, 800 mg/cc) and 3 sizes (1 x 1, 3 x 3 and 5 x 5 mm; diameter x height) was scanned on a dual-source scanner (Siemens Definition Force, Forchheim, Germany). Scans were performed with a clinical calcium scoring protocol using 120 kV and fixed mAs/rotation of 100/180/240 in small/medium/large sized phantom and repeated with 25% and 50% lower mAs/rot. Reconstructed images were 3 mm thick with 50% overlap using vascular medium smooth filter (Bv40) and vendor recommended quantitative filter (Qr36) with FBP and iterative reconstruction (IR) ADMIRE of strength 3 and 5 (total 6 image sets). Agatston scores (AS) were measured from all images using commercial software (syngo Calcium Scoring, Siemens Healthineers). AS were also measured from 10 patients who received calcium-scoring scans for clinical purposes. Images were reconstructed using the same kernels as in the phantom scans. AS data were compared using percentage deviation (%dev) from Qr36 and non-parametric analysis (Wilcoxon signed rank test).

Results: There were no statistical differences for AS between FBP images for all dose levels. AS from IR was significantly lower compared to FBP (p<0.05). However, AS showed dependence on the reconstruction kernels: Bv40 kernel with IR underestimated AS more than Qr36 kernel with IR. Six of ten patients had zero AS. In 4 patients, Minimum %dev observe was in Qr36-IR-strength-3 (11%) and maximum %dev observed in Bv40-IR-strength-5 (50%).

Conclusion: AS are sensitive to the type of reconstruction kernel; quantitative kernels are more reliable and offer a potential for consistent calcium scoring with reduced dose and iterative reconstruction.

Funding Support, Disclosures, and Conflict of Interest: Frank Dong receives research support from Siemens Healthineers for an unrelated project.


Calcifications, Computed Radiography, Diagnostic Radiology


IM- CT: Coronary artery calcium scoring

Contact Email