To assess whether the radiomics signature of pericoronary adipose tissue (PCAT) from coronary computed tomography angiography (CCTA) can distinguish between perimenopausal women with coronary heart disease (CHD) and those without coronary artery disease (CAD). This single-center retrospective case–control study comprised 140 perimenopausal women with CHD presenting with chest pain who underwent CCTA within 48 h of admission. They were matched with 140 control patients presenting with chest pain but without CAD, based on age, risk factors, radiation dose and CT tube voltage. For all participants, PCAT around the proximal right coronary artery was segmented, from which radiomics features and the fat attenuation index (FAI) were extracted and analyzed. Subsequently, corresponding models were developed and internally validated using Bootstrap methods. Model performance was assessed through measures of identification, calibration, and clinical utility. Using logistic regression analysis, an integrated model that combines clinical features, fat attenuation index and radiomics parameters demonstrated enhanced discrimination ability for perimenopausal CHD (area under the curve [AUC]: 0.80, 95% confidence interval [CI]:0.740–0.845). This model outperformed both the combination of clinical features and PCAT attenuation (AUC 0.67, 95% CI 0.602–0.727) and the use of clinical features alone (AUC 0.66, 95% CI 0.603–0.732). Calibration curves for the three predictive models indicated satisfactory fit (all p > 0.05). Moreover, decision curve analysis demonstrated that the integrated model offered greater clinical benefit compared to the other two models. The CCTA-based radiomics signature derived from the PCAT model outperforms the FAI model in differentiating perimenopausal CHD patients from non-CAD individuals. Integrating PCAT radiomics with the FAI could enhance the diagnostic accuracy for perimenopausal CHD.