Purpose: To quantify the severity of motion artifacts in 2D pulmonary angiography using conventional digital subtraction angiography (DSA) and dual-energy (DE) subtraction angiography in a free-breathing animal model.
Methods: Real-time dual-energy pulmonary angiography was performed in three free-breathing porcine subjects on an interventional C-arm system equipped with 30 Hz fast-kV switching capability. Tissue-subtracted dual-energy images were generated by weighted log subtraction of low and high kV image pairs. The relative size and intensity of respiratory-induced motion artifacts over a single respiratory cycle were quantified and compared between dual-energy and conventional DSA images. Severity of motion artifacts was measured from image profiles placed across the diaphragm boundary. Deformable image registration was implemented to correct inter-frame respiratory motion in the 1/30 s period between low and high kV images. Additionally, dual-energy images were generated after application of a machine-learning based denoising algorithm. Dual-energy images with one or both algorithms applied were subjected to the same artifact analysis and compared with regular dual-energy.
Results: For dual-energy acquisitions, the areas of diaphragm motion artifact profiles were reduced by 93.2% to 96.8% compared to DSA. Application of deformable registration further reduced artifact areas (98.1% to 98.7%). Artifact profile intensity was reduced by 32.2% to 74.6% using dual-energy, and by 80.3% to 90.3% for registered dual-energy. Noise reduction in dual-energy images across all porcine subjects ranged from 58.5% to 67.4%.
Conclusion: Motion artifacts in dual-energy and conventional DSA on an interventional C-arm have been evaluated in an animal model of free-breathing pulmonary angiography. Dual-energy imaging with 30 Hz kV switching enables a large reduction in diaphragm motion artifacts seen in DSA images. Application of a deformable registration algorithm further reduced residual inter-frame motion artifacts seen in dual-energy subtraction images.
Funding Support, Disclosures, and Conflict of Interest: Financial support was provided by NIH Grant No. R21 EB023008 and funding received from Siemens Healthineers.