Program Information
Stability Investigation of a Gamma Fitting Algorithm for Angiographic Parametric Imaging at Low X-Ray Exposures Using a Patient Specific Neurovascular Phantom
A Balasubramoniam1,2*, D Bednarek2, , S Rudin1,2 , C Ionita1,2 , (1) Department of Biomedical Engineering, SUNY Buffalo,Buffalo, NY, (2) Toshiba Stroke and Vascular Research Center, SUNY Buffalo, Buffalo, NY
Presentations
MO-FG-CAMPUS-I-6 (Monday, July 13, 2015) 5:00 PM - 5:30 PM Room: Exhibit Hall
Purpose:
To analyze the stability of a Gamma fitting algorithm based on digital subtraction angiography sequences, at different exposure, hence noise levels.
Methods:
Starting with a patient CT volume, we built a 3D printed neurovascular phantom which contained a complete Circle of Willis with major arteries and five aneurysms. We placed the phantom in a 15 cm water bath and connected it to a flow loop containing a peristaltic pump which simulated physiological relevant flow conditions. We injected 10 ml contrast boluses using an automatic contrast injector at a rate of 10 ml/sec. Digital Subtraction Angiography images were acquired at 30 frames/sec and processed with a gamma fitting based algorithm, to yield parametric maps of: Mean transit time (MTT), Time-to-Peak (TTP), Bolus Arrival Time (BAT). Starting with the optimal exposure parameters selected by the x-ray system automatic exposure control, while keeping the same kV, we lowered the mA and exposure per frame in four steps until we reached the minimum value allowed by the system. We analyzed the variation of the MTT, TTP and BAT for various artery signal to noise ratios using four ROIs.
Results:
The peak (maximum opacification) SNRs for full dose and minimum dose were: 35 and 13 for 4 mm arteries and 18 and 7 for 2 mm arteries. The parameters standard deviation expressed as a percent fraction of the value measured at the full dose value, were: MTT=2.4%, BAT=0.5% and TTP=1.9% for 4 mm vessels and MTT=3.7%, BAT=5.4% and TTP=6.3% for 2 mm vessels.
Conclusion:
Despite a significant decrease of the peak SNR the algorithm performed very well displaying variations less than 6.3% of the ideal conditions. Partial Support: NIH grant R01EB002873 and Toshiba Medical Systems Corp.
Funding Support, Disclosures, and Conflict of Interest: Partial Support: NIH grant R01EB002873 and Toshiba Medical Systems Corp.
Contact Email: