AIM To assess the correlation between CT-FFR hemodynamic indicators and coronary artery stenosis, and develop a prediction model for diagnosing coronary artery stenosis.

METHODS Clinical and imaging data from 109 patients with coronary artery disease, treated at the Air Force 986 Hospital between 2023 and 2024, were analyzed. Patients were categorized into two groups based on CT-FFR values: CT-FFR >0.8 (negative group, n=40) and CT-FFR ≤ 0.8 (positive group, n=69). Comparative analysis of coronary artery imaging features was conducted, and a diagnostic model was formulated.

RESULTS A total of 109 patients with 201 coronary arteries were enrolled. AI-based measurements of the crime lesion site CT-FFR values, calcium scores, crime lesion type, length, volume, and minimum lumen area (MLA) revealed statistically significant differences in calcium score, crime lesion type, lesion length, MLA, and stenosis degree (P<0.05, P<0.01). ROC analysis demonstrated AUC values for stenosis degree, lesion length, MLA, calcium score, lesion type, lesion volume, and combined indicators as 0.585, 0.679, 0.282, 0.723, 0.689, 0.672, and 0.823, respectively. The combined indicators yielded the highest AUC value, providing optimal sensitivity and specificity.

CONCLUSION Coronary artery stenosis has certain diagnostic value in predicting coronary ischemia, but its diagnostic performance is generally poor. However, the combined index of coronary artery stenosis and multiple plaque characteristics has better diagnostic performance than a single index, and it is the best hemodynamic index for predicting ischemic stenosis.