Optimization of Coronary Artery Calcium Scoring Accuracy in Low-Dose Chest Computed Tomography Using Fast Non-Local Means Denoising with High-Pitch Acquisition and Tin Filtration

초록

This study evaluated quantitative accuracy of the coronary artery calcium score (CACS) and the associated radiation dose reduction achieved by applying the fast non-local means (FNLM) algorithm to non-electrocardiography (ECG)-gated, low-dose chest computed tomography (CT) images acquired with a high-pitch scan and tin filter. Thirty patients underwent standard-dose CACS and low-dose chest CT were retrospectively analyzed. The processed low-dose images using the FNLM algorithm demonstrated a 4.2-5.0% mean CACS decrease and <= 16.4% median value increase relative to the standard-dose CACS CT but without statistical significance (p > 0.05). Notably, the quantitative error progressively decreased with increasing algorithm strength. The Pearson correlation coefficient reached 0.949 at Stage 3, indicating robust agreement with the standard-dose CACS CT. When stratified by patient heart rate, the high heart rate cohort exhibited the largest scoring errors but without statistical significance (p > 0.05). Importantly, the FNLM-processed protocol substantially reduced radiation dose, decreasing the mean volume CT dose index by 71.2% and dose-length product by 56.5% (p < 0.05). FNLM proves to be an effective post-processing technique that preserves CACS accuracy in non-ECG-gated, low-dose chest CT, thereby offering a clinically viable alternative imaging protocol for patients requiring routine screening.

키워드