RATIONALE AND OBJECTIVES: Patients with a low Agatston score often present with clinical signs and symptoms suggestive of coronary artery disease, despite having minimal calcium deposits. This study aimed to compare the efficacy of low-dose non-contrast cardiac CT with coronary computed tomography angiography (CCTA) in pericoronary adipose tissue (PCAT) radiomics for predicting coronary artery plaques, using CCTA as the reference standard. MATERIALS AND METHODS: This retrospective study analyzed 459 patients with suspected coronary artery disease and a coronary artery calcium score <
100 Agatston units, who were treated between June 2021 and December 2023 at a tertiary hospital. Three predictive models for coronary artery plaques were developed: (1) a clinical factor model, (2) a hybrid model integrating clinical factors and CT PCAT radiomics, and (3) a hybrid model integrating clinical factors and CCTA PCAT radiomics. Multivariable logistic regression and receiver operating characteristic curve evaluations were performed to develop and validate predictive models. RESULTS: Both hybrid models showed significant correlations in the training set (r = 0.890, P <
0.002) and the validation set (r = 0.920, P <
0.002). The mean agreement in the training set is 0, with 3.42% (11/322) of the data points outside the 95% CI (-0.18-0.18, P <
0.002). The mean agreement in the validation set is -0.244, with 6.57% (9/137) of the data points outside the 95% CI (-0.443-0.045, P <
0.002). CONCLUSIONS: Non-contract CT PCAT radiomics showed comparable efficacy to CCTA PCAT radiomics in predicting coronary artery plaques among patients with low Agatston scores.