Diagnostic Accuracy of a Novel On-site Virtual Fractional Flow Reserve Parallel Computing System.

Area of Science:

• Cardiovascular imaging and intervention
• Computational fluid dynamics in medicine
• Diagnostic accuracy of non-invasive cardiac imaging

Background:

• Fractional flow reserve (FFR) is the gold standard for assessing lesion-specific ischemia in coronary artery disease.
• Invasive FFR requires cardiac catheterization, which carries risks and costs.
• Non-invasive methods to predict FFR are highly desirable for improved patient management.

Purpose of the Study:

• To evaluate the diagnostic accuracy of a novel on-site virtual fractional flow reserve (vFFR) derived from coronary computed tomography angiography (CTA).
• To compare the performance of vFFR against invasive FFR in detecting significant coronary stenosis.

Main Methods:

• 100 vessels from 57 patients with CTA and invasive FFR data were analyzed.
• A fully automated algorithm performed coronary lumen segmentation and 3D reconstruction.
• On-site vFFR was calculated using parallel computing, compared to invasive FFR at cutoffs of ≤0.8 and ≤0.75.

Main Results:

• The automated vFFR calculation took an average of 12 minutes per patient.
• vFFR showed strong correlation (r=0.75) and minimal bias with invasive FFR.
• Diagnostic performance included high sensitivity (up to 95%) and specificity (up to 80%), with AUCs of 0.88 and 0.94 for vFFR versus 0.61 and 0.62 for CTA stenosis alone.

Conclusions:

• The novel, fully automated, on-site vFFR technology demonstrates excellent diagnostic performance.
• vFFR accurately detects lesion-specific ischemia, offering a promising non-invasive alternative.
• This technology has the potential to improve the diagnosis and management of coronary artery disease.