Background: Oscillating x-ray attenuation in the lungs provides an opportunity to evaluate pulmonary perfusion without contrast. Recent intensity-based methods have been compared to pulmonary scintigraphy and CT angiography but lack rigorous phantom studies.
Purpose: A new method to quantify the periodic signal amplitude was employed using spectral analysis. Performance was characterized using a water phantom capable of creating an oscillating x-ray attenuation at physiologic amplitudes. Feasibility in detecting abnormal perfusion was performed on a volunteer with pulmonary vascular disease and compared to pulmonary angiography, the clinical gold standard.
Methods: For each fluoroscopic acquisition, the normalized temporal signal from each pixel was decomposed into its frequency components using Fourier transformation, and the spectral amplitude, defined as the x-ray pulsatility index (XPI), was determined at the desired frequency using a band-pass filter. XPI was displayed as a pixel-wise parametric colormap. Based on XPI maps generated using two human volunteers, a water bath phantom was constructed with a fluctuating fluid height and a 1 cm diameter pulsatility defect. Contrast-to-noise (CNR) of the defect was measured using fluoroscopy images acquired at variable fluid height fluctuation (0.1-1.9 mm) and oscillation frequency (30-60 bpm). Various sampling frame rates (3-30 fps) and acquisition durations (1.8-8 s) using truncated datasets were reconstructed from full datasets. Fluoroscopic images were obtained in a patient just prior to pulmonary angiography in the same projection.
Results: XPI maps in human subjects showed high signal to background contrast with high central XPI measuring up to 0.5. Phantom experiments revealed CNR was linearly correlated to fluid height change (r2 = 0.998). CNR is proportional to increasing sampling frame rate and increasing acquisition duration as expected with Fourier analysis. XPI map displayed multifocal perfusion defects in good agreement with pulmonary angiography.
Conclusion: Spectral analysis is an accurate and sensitive method to detect small changes in periodic x-ray attenuation using a short fluoroscopic acquisition. This method demonstrated good agreement to pulmonary angiography and shows promise for clinical imaging of pulmonary perfusion using standard fluoroscopic methods.
Keywords: dynamic x-ray imaging; fluoroscopy; lung perfusion; pulmonary perfusion; spectral analysis.
© 2024 American Association of Physicists in Medicine.