This single center prospective study included eight patients with clinically suspected coronary artery disease (mean age 54 ± 8 [SD]; range 44-65; 7 males) who underwent stress-rest dynamic CT perfusion scanning between December 2021 and January 2023 (Figure 1). Stress dynamic CTP was performed as part of the comprehensive cardiac CT protocol shown in Figure 2. Dynamic stress CTP was performed with an intravenous adenosine triphosphate load of 0.14 mg/kg/min for 3 minutes. In case of inadequate loading, the dose was increased by 0.02 mg/kg/min every minute up to 0.18 mg/kg/min. A non-ionic contrast agent (Iomeron 350; Eisai, Tokyo, Japan) was injected at a fractional iodine dose of 25 mg iodine/kg/s for 10 s. Then, saline chaser was injected at the same rate. Five seconds after the start of contrast injection, the prospective ECG dynamic mode was used for 25 s to collect data at each heartbeat, targeting the 300 ms phase of the R-R interval. Adenosine triphosphate administration was stopped immediately after the end of imaging. The scan parameters for CTP were as follows: tube voltage, 80 kVp; tube current, 120 mA; gantry rotation speed, 0.275 s/rotation; detector collimation, 320 × 0.50 mm. Scan range was adjusted to include the entire left ventricle. Subsequently, rest CTP and delayed-enhancement CT were performed after stress dynamic CTP, respectively. All images were reconstructed using a deep learning-based algorithm (Advanced intelligent Clear-IQ Engine [AiCE]; Canon Medical Systems) with a new dynamic perfusion image noise reduction filter (4-dimensional similarity filter).^3-5 The reconstructed images (slice thickness of 1 mm and section spacing of 1 mm) were transferred to a workstation (Ziostation2 plus; Ziosoft, Tokyo, Japan and VITREA; Vital, Tokyo, Japan) for post-processing. The golden standard for the MBF values was calculated using the single compartment method corrected by the Renkin-Crone equation (MBF_ref), which was validated with ¹⁵O-water PET. Then, MBF was calculated using the same data with the maximal upslope and deconvolution methods and corrected with the Renkin-Crone formula as MBF_MU and MBF_dec, respectively.

 MBF calculations were performed on a per-segment basis using the American Heart Association 16-segment model. The values before and after the correction were analyzed using the Bland-Altman analysis.