Photon-Counting Chest CT at Radiography-Comparable Dose Levels: Impact on Opportunistic Visual and Semiautomated Coronary Calcium Quantification.

Abstract

Objectives: The introduction of photon-counting detector computed tomography (PCCT) has allowed for significant dose reductions compared to energy-integrating-detector CT, making it particularly relevant for applications such as lung cancer screening. Coronary artery calcification is an important incidental finding in lung cancer screening, warranting attention in this context. This study aims to assess the impact of dose reduction to levels comparable to that of a chest radiography on opportunistic evaluation of coronary artery calcification on PCCTs of the chest.

Materials and methods: Sixty-eight out of 115 patients with age >45 years and body mass index ≤30 kg/m2 undergoing noncontrast low- and chest-radiography-comparable-dose PCCT in the same session were included. Scans were performed at 100 kVp with image quality settings 12 (low-dose) and 2 (radiography-comparable-dose). Visual calcium scoring was conducted by 2 readers using 2 scoring approaches (CAD-RADS 2.0 and Shemesh). Semiautomated quantitative analysis was performed using commercially available software. Image quality was evaluated using 5-point Likert scales.

Results: Sixty-eight patients (65.9 ± 8.6 years; 49 men) were subjected to evaluation. CTDI was lower for radiography-dose scans (0.11 mGy vs 0.68 mGy; P < 0.001). Image quality was found to be inferior for radiography-dose scans (4.01 vs 2.03; P < 0.001). In both visual scoring approaches, coronary calcification was scored significantly lower in radiography-dose scans (P < 0.001 for both) with almost perfect reader agreement (CAD-RADS score Cohen's kappa =0.82; Shemesh score Cohen's kappa =0.81), most importantly reclassification from mild to absent occurred for CAD-RADS score in 31%/21% of cases and for Shemesh score in 23%/15% of cases (reader 1/reader 2). Semiautomated assessment showed no significant differences between low and radiography dose (P = 0.121). Strong correlation between scores (Pearson's r = 0.98, P < 0.001) with good agreement (Cohen's kappa =0.61) was found.

Conclusions: Coronary artery calcifications are underestimated on radiography-dose PCCT visually, whereas semiautomatic analysis provides more robust results. Visual underestimation of coronary artery calcification in low-dose imaging is further amplified with the additional dose reduction to radiography-comparable dose levels, indicating that while estimation of high cardiovascular risk is feasible, exclusion of such risk is not possible.