Detailed evaluation of the hepatic arteries is a necessary task before liver surgery or transcatheter arterial chemoembolization (TACE). Intra-arterial digital subtraction angiography (IADSA) is a useful technique for visualizing the hepatic arterial anatomy; however, as it is an intra-arterial procedure, it is invasive. Contrast-enhanced computed tomography (CT) and magnetic resonance angiography (MRA) are standard non-invasive procedures used to visualize the hepatic arteries [1] [2] . However, these techniques require the use of contrast agents that have the potential to cause severe adverse reactions such as anaphylactic shock, making them unsuitable for patients who have asthma, severe renal dysfunction, or a history of allergy to contrast agents. Furthermore, gadolinium-based MRA contrast agents can cause nephrogenic systemic fibrosis in patients with severe renal dysfunction [3] . Therefore, vascular imaging without contrast medium is safe and low cost modality. Non-contrast-en- hanced MRA techniques are suitable for visualization of the hepatic arteries. Time-spatial labeling inversion pulse (Time-SLIP) is a form of spin labeling that uses inversion pulses to suppress the background signal before data acquisition [4] . By changing the inversion time (TI), the inflowing blood signal can be controlled and the vessel to be visualized can be selected. Recent advances in magnetic imaging technology mean that it is now possible to use multiple inversion pulses. Selective visualization of the hepatic arteries by using respiratory- triggered three-dimensional (3D) true steady-state free-precession (SSFP) combined with Time-SLIP has been reported, and the optimal TI for balancing vessel-to-liver contrast and peripheral hepatic artery visualization is 1200 ms [5] [6] . We consider IADSA to be the gold standard technique for visualizing hepatic arteries. Therefore, here, we scored the visualization of hepatic arteries by using non-contrast-enhanced hepatic MRA by referencing the images of IADSA. The purpose of this study is to evaluate respiratory-triggered three-dimensional (3D) true steady-state free-procession (SSFP) projection magnetic resonance angiographic sequences with time- spatial labeling inversion pulse (Time-SLIP) for visualizing the hepatic arteries and to optimize the image acquisition protocol.

Images were acquired from all patients and no motion artifacts were seen. The total imaging time for each

Time-SLIP was approximately 10 min. Twenty four patients experienced TACE and we established the images of intra-arterial angiogram as the gold standard. After referencing the angiogram, the two radiologists evaluated the hepatic MRA images. More peripheral branches were identified in images captured by using tag pulse Pattern I than in images captured by using Pattern II (P < 0.05). The highest score of 5 (meaning that the fourth- order branches of the artery could be detected) was achieved in only one case, of visualization of the right hepatic artery with tag pulse Pattern I, although this patient did have ascites (Table 1 and Figure 3). In spite of better quality images of Pattern I, three left hepatic arteries (LHA) and two right hepatic arteries (RHA) were not visualized. Their average age was 80.0 years. Three of five patients breathed irregularly and did not synchronize the respiratory-triggered scan. Aging and irregular breathing might be defectiveness factor. There were no significant differences between the scores of images from patients with Child-Pugh classifications A and B for either hepatic artery (P > 0.05) (Table 2). Normal hepatic arterial anatomy was seen in 21 patients and anatomical variations were seen in four patients: two patients had a replaced right hepatic artery arising from the superior mesenteric artery, and two had an aberrant left hepatic artery arising from the left gastric artery. Both radiologists identified the aberrant arteries before angiography was performed (Figure 4 and Figure 5). The vascular variants were coincident with the results of angiography (IADSA).

Multi-detector-row 3D CT angiography provides excellent visualization of vascular details, and owing to its accessibility and brief scan time it is now a commonly used technique for assessing liver anatomy [7] [8] . Moreover, 320-row CT 4D DSA can accurately visualize normal anatomy and variation of hepatic artery, dynamically display tumor staining and reproducibly delineate the three-dimension relationship between tumor and blood vessels [9] . However, CT scans expose the patient to ionizing radiation and intravenous contrast agents that may cause serious adverse events such as anaphylactic shock. Contrast-enhanced MRA is a useful method for evaluating

LHA: left hepatic artery; RHA: right hepatic artery.

CPA: Child-Pugh classification A; CPB: Child-Pugh classification B.

vascular disease without the need to expose the patient to radiation [10] [11] . However, patients with moderate to severe renal dysfunction may develop a serious adverse reaction called nephrogenic systemic fibrosis in response to gadolinium-based MR contrast agents. This adverse event is rare but non-reversible. Other adverse events such as anaphylactic shock are sometimes fatal. Therefore, non-contrast techniques are required for visualizing the hepatic vessels in these patients.

Previous reports have described the selective visualization of hepatic arteries without the need for an exogenous contrast agent by using a combination of respiratory-triggered 3D true SSFP sequences and Time-SLIP [5] [6] . Recently, Shimada et al. demonstrated that fat suppression by means of short-tau inversion recovery provides better background signal suppression than chemical shift selective pulse [5] . The most important factor for clearly visualizing hepatic arteries with non-contrast-enhanced hepatic MRA is TI. A short TI (800 or 1000 ms) favorably suppresses the background signal, but the inflow signal is weak and therefore the images obtained are unsuitable for peripheral vessel visualization [5] . A longer TI (1400 ms) shows more peripheral arterial flow, but the vessel-to-liver contrast deteriorates because of background signal recovery. An intermediate TI of 1200 ms allows the visualization of more peripheral hepatic arteries and suppresses the flow signal of the splenic vein, superior mesenteric vein, and inferior vena cava [5] [6] . A TI of 1200 ms gives the optimal balance between vessel-to-liver contrast and peripheral hepatic artery visualization [5] , and therefore was the TI that we used here. We used short-tau inversion recovery not only to obtain complete fat suppression but also to improve image quality by suppressing signals from the surrounding tissues (liver parenchyma, gastrointestinal tract, and gall- bladder) [6] [12] [13] . We also examined different patterns of inversion pulse placement. Placement of the inversion pulses closer to the start point of the inflowing arterial signal provided better visualization of the peripheral vessels, which is why the hepatic arteries were better visualized with tag pulse Pattern I than with Pattern II. The main difference between the two tag pulse patterns was where the inflowing blood signal from the aorta, superior mesenteric artery, and several veins (such as the inferior vena cava, splenic vein, and portal vein) began. The superior border of tag pulse Pattern I was placed near the celiac artery, whereas that of Pattern II was placed at the descending aorta at a level of the heart inferior border (Figure 1). This means that the inflowing signal in Pattern I began at the celiac trunk, whereas that in Pattern II was nearer the descending aorta. This difference in where the inflow began influenced the visualization of the peripheral hepatic arteries (Figure 6). However, the disadvantage of tag pulse Pattern I was seen in the poor visualization of one case of aberrant left hepatic artery arising from the left gastric artery (Figure 4). The image was of poor quality because the branch of this vessel was outside the tag pulse pattern, so the background signal was not suppressed.

We expected the flow in the hepatic arteries of patients with liver dysfunction to be slow and poorly visualized because of intrahepatic vascular increased pressure and resistance [14] . However, the scores of images from patients with Child-Pugh classifications A and B were not significantly different (P > 0.05). Patients with Child-Pugh classification C were not enrolled in the study: these patients have substantial liver dysfunction and

are in overall poor health and are therefore not candidates for TACE. One important point to note regarding our results is that patients with ascites had high scores, indicating that ascites did not influence the visualization of hepatic arteries with Time-SLIP MRA.

This is the report to evaluate non-contrast-enhanced hepatic MRA by referencing angiography for the visualization of hepatic arteries. Non-contrast-enhanced hepatic MRA with Time-SLIP is a safe and effective method for visualizing hepatic arterial anatomy before angiography or surgery.