Validation of time-resolved Digital Subtraction Angiography using patient specific 3D printed neurovascular phantoms
MetadataShow full item record
Digital Subtraction Angiography (DSA) is a standard golden tool for interventional radiology to clearly visualize blood vessels in a bony or dense tissue environment. Time-resolved DSA, also known as color-coded DSA or Parametric Imaging (PI), is a semi-quantitative tool giving temporal blood flow information. It generates a Time Density Curve (TDC), to give PI parameters – Bolus Arrival Time (BAT), Time to Peak (TTP) and Mean Transit Time (MTT). It can be a useful tool for vascular disease diagnosis and post-treatment analysis in a clinical environment. PI was validated using 3 major steps: 1) validation at very low mA (tube current) and exposure levels, 2) comparing the parameters of PI with velocities obtained from Doppler ultrasound and 3) obtaining 4D PI maps using 2D PI maps co-registered on a 3D vascular geometry to overcome the problem of incorrect PI parameters at regions of vessel overlap. In order to validate PI, we 3D printed patient specific neurovascular phantoms of the complete Circle of Willis, from a patient CT volume. These phantoms were connected to a flow loop containing a peristaltic pump, pumping fluid at various flow rates, simulating physiological flow conditions. To obtain the DSA images, the phantoms were also connected to an automatic contrast injector injecting contrast boluses at 10 ml/sec. The water pumped through the phantoms was not recirculated to prevent contrast contamination. The DSA images were obtained using the Toshiba Infinix Angio C-arm. To obtain the Doppler Ultrasound velocities, the phantoms were embedded in Ballistic gel. Doppler Ultrasound velocities were obtained using the Toshiba Aplio XG Ultrasound system and blood mimicking fluid was pumped through the phantoms. The DSA images were obtained as in Step 1 and saved in ‘.raw’ format. The DSA sequence was input into an in-house developed LabVIEW software to obtain the 2D parametric maps. For the 4D PI maps, we made use of a Computed Tomography (CT) volume using only a single Cone Beam Computed Tomography (CBCT) acquisition and co-registering the 2D Parametric maps on the 3D vascular CT volume with regions of vessel overlap were spline interpolated from regions of no vessel overlap. An in-house developed LabVIEW software was used to obtain the 4D PI maps. Results: The results of the PI maps obtained for each of the steps (1-2) were compared to a PI map at optimal conditions. For Step 1 at low Signal to Noise Ratio (SNR) at low x-ray exposures, the results showed minimal standard variation in SNR in the bigger arteries (high SNR), however there was a greater standard variation in SNR in the smaller arteries due to low SNR. For Step 2, for an increase in Doppler ultrasound velocity there was a decrease in the PI parameters. This is in good correlation to the fact that an increase in blood velocity will decrease the time taken to flow through the blood vessels. For Step 3, the 4D PI maps obtained overcame the problem of vessel overlap in standard 2D PI maps. At the regions of vessel overlap, interpolation helped in getting rid of higher PI Parameter values due to higher contrast density.