Scanning and image reconstruction
Image quality phantoms were scanned using HRCBCT and 3DRA in the same field of view (FOV; 8 inches) and same size matrix for evaluation of a feasible approach.
Both systems were equipped with the same angiography system with a flat panel detector (Allura Clarity FD 20/20,
Philips Healthcare,
Best,
Netherlands),
and the acquisition protocols and postprocessing algorithms were embedded in the system (Fig.1).
The HRCBCT imaging was conducted at a rate of 30 frames/s,
total rotation time of 20 s,
focus size of 0.4 mm,
620 frames total,
240° angle using a 1,024×1,024 pixel matrix detector with an 8-in FOV,
and a fixed 80 kV tube voltage.
The conventional 3DRA imaging was conducted at a rate of 30 frames/s,
total rotation time of 4.1 s,
focus size of 0.7 mm,
120 frames total,
240° angle using a 1,024×1,024 pixel matrix detector with an 8-in FOV and an 88 kV tube voltage.
Then,
HRCBCT images were reconstructed using stent mode as an optimal kernel,
voxel size of 5123 and maximum spatial resolution matrix,
and 3DRA images were reconstructed in very sharp mode as an optimal kernel,
voxel size of 5123,
and maximum spatial resolution matrix using a 3D workstation (Xtravision; Philips Healthcare).
Finally,
these images were rendered with a slice thickness of 5.0 mm using Ziostation2 (Ziosoft,
Japan) at the medical imaging workstation for evaluation of the following four image qualities.
Modulation transfer function
Modulation transfer function (MTF) was measured using a laboratory-made 20 cm diameter water phantom with a 0.2 mmφ copper wire at the center.
Each signal value of rendered images was measured by ImageJ,
which is a Java-based application for analyzing images,
and the MTF was obtained by the wire method (Fig.
2).
Noise power spectrum
A Catphan phantom (Phantom Laboratory,
Salem,
NY) was used for measurement of the noise power spectrum (NPS).
The phantom consisted of various cylindrical sections (modules),
each of which was designed for a specific test.
For this study,
module CTP712 (uniformity section) was used.
Each signal value of rendered images at the phantom center was measured,
and the NPS was obtained by a one-dimensional method using numerical slit scanning (slit method) with five slit traces using a 1 × 30 pixel virtual slit by ImageJ (Fig.
3).
Signal-to-noise ratio
The signal-to-noise ratio (SNR) was calculated from the MTF and NPS.
We defined it as SNR=(MTF2⁄NPS)1/2.
Contrast intensity
Contrast intensity was measured using a laboratory-made 20 cm diameter water phantom that has 2 mm diameter silicon tubes filled with diluted contrast medium (Omnipaque™ 300; Iohexol ,
300 mgI/mL) placed at the center of the phantom (Fig.
4).
The dilution factors of the tubes were: 1 (100% undiluted contrast medium),
1/2 (50% concentration),
1/3(33%),
1/5 (20%),
1/7 (14.3%),
1/10(10%).
Signal values of the tubes in rendered images were measured using Ziostation2.