This retrospective study was approved by our Institutional Committee on Human Investigation. Informed consent for 4D CT in addition to routine CT imaging was obtained from all patients prior to 4D CT scanning. From March 2018 to September 2019, 66 patients underwent IR treatment for such as balloon-occluded retrograde transvenous obliteration and percutaneous transhepatic sclerotherapy at our department. After excluding 27 patients with diagnoses other than gastrorenal shunt, 39 patients with gastric varices and/or portosystemic encephalopathy including gastrorenal shunt met the criteria for 4D CT prior to IR treatment. Of these, the following were also excluded: 10 patients who were scanned less than 3 months prior to IR treatment, 6 who had a history of varix rupture, and 5 with renal failure. A final total of 18 consecutive patients (9 men, 9 women; mean age 65.3 years; age range 41–83 years; mean weight 61.7 kg, weight range 46–83 kg) who had undergone 4D CT before IR procedures were enrolled in the study.
4D CT protocol
4D CT was acquired by a 320-detector-row CT scanner (Aquilion one, Canon Medical Systems, Otawara, Japan) equipped with 320 × 0.5-mm-wide detector rows covering a 16-cm-long volume per rotation. A bolus of contrast medium was injected by an automatic injector via the cubital vein, followed by a saline flush. To visualize the aorta at the level of the celiac artery bifurcation, a bolus-tracking scan was performed after intravenous injection of 10 mL of contrast medium (Oypalomin 300, Fuji Pharma, Tokyo, Japan) into the cubital vein to confirm enhancement of the aorta and celiac artery. After the bolus-tracking scan, 600 mgI/kg of contrast medium was injected at a rate of 3.7–4.8 mL/s over 20 s, followed by 20 mL of saline, during 4D CT scanning. When contrast enhancement of the aorta at the level of the celiac artery bifurcation reached > 100 HU, patients were asked to inhale and hold their breath and 4D CT scanning was initiated. Scan parameters were as follows: detector collimation 320 × 0.5 mm; matrix 512 × 512; slice thickness 0.5 mm; tube voltage 120 kVp; mean tube current 58.8 mA (range, 40–80 mA); gantry rotation time 0.5 s; interval 1 s; number of imaging volumes 18; total scan time 26 s. Temporal resolution was 1.0 s, corresponding to the interval time. The current was determined based on twice the CT dose index for commonly performed helical CT examinations such as portal phase imaging. In this protocol, a series of 18 dynamic volumes was obtained, consisting of 320 slices covering the portal circulation. To make it easier for patients to hold their breath during scanning, they were given oxygen via a mask beforehand to increase their oxygen saturation [22].
After 4D CT scanning, portal, venous, and late phase imaging was performed from the diaphragm to the pelvis. Portal and venous phase acquisitions were initiated 35 and 55 s after the start of 4D scanning. The late phase scan was initiated 180 s after injection of contrast medium. Scan parameters were as follows: detector collimation 320 × 0.5 mm; matrix 512 × 512; table feed 65.04 mm/s; slice thickness 0.5 mm; tube voltage 120 kVp; mean tube current 392.8 mA (range, 277.5–562.5 mA); gantry rotation time 0.5 s; interval 0.5 mm.
Post-processing of 4D CT images
The 18 phases of the 4D CT images were transferred to a workstation (PhyZiodynamics; Ziosoft, Tokyo, Japan) for post-processing. This commercially available software creates smooth video images using a motion coherence function that fills in interphase motion with interpolated data. Using this software, we generated two additional phases between the original phases to create interphase motion, thus creating partially artificially constructed smooth and clear 4D motion [23]. To display only the portal circulation, the abdominal arteries were removed from the cine images, and only portal vessels with enhancement > 70 HU were displayed. Finally, a total of 52 phases were reconstructed into three-dimensional volume-rendering and maximum intensity projection images. Multiple-frame dynamic cine loops from the anterior and inferior aspects were generated for evaluation. A radiographer generated dynamic cine loops of the maximum intensity projection and volume-rendering images of all patients in multiple views, which took 3–5 h to create, depending on the experience of the radiographer.
Evaluation of 4D CT images
Two radiologists, each of them with more than 15 years of experience in diagnostic and interventional radiology, independently evaluated all 4D CT examinations. A 2-point scale (visualized or not visualized) was used to assess visualization of flow in the portal vein, splenic vein, superior mesenteric vein, and gastrorenal shunt. They watched videos of the flow dynamics in the portal vein, splenic vein, and superior mesenteric vein and assessed flow direction of these vessels using a 2-point scale (hepatopetal flow, toward the liver or hepatofugal flow, away from the liver) by consensus.
Radiation doses of CT scans
Radiation doses were recorded in all 18 cases for the 4D CT scans and all other scans.
Validation study with US
In the first 7 patients, color Doppler US was performed as a preliminary examination to assess the direction of hepatopetal or hepatofugal flow in the portal, splenic, and superior mesenteric veins, using a LOGIQ S7 Expert machine (GE Healthcare, Tokyo, Japan) equipped with a C1-5-D convex probe. US was performed with the patient in the supine position and was scheduled on the same day as 4D CT.
Statistical analysis
No statistical analysis was performed due to complete concordance observed in the validation study.