Intra-patient and inter-observer image quality analysis in liver MRI study with gadoxetic acid using two different multi-arterial phase techniques

Abstract

Purpose

To evaluate intra-patient and interobserver agreement in patients who underwent liver MRI with gadoxetic acid using two different multi-arterial phase (AP) techniques.

Methods

A total of 154 prospectively enrolled patients underwent clinical gadoxetic acid-enhanced liver MRI twice within 12 months, using two different multi-arterial algorithms: CAIPIRINHA-VIBE and TWIST-VIBE. For every patient, breath-holding time, body mass index, sex, age were recorded. The phase without contrast media and the APs were independently evaluated by two radiologists who quantified Gibbs artefacts, noise, respiratory motion artefacts, and general image quality. Presence or absence of Gibbs artefacts and noise was compared by the McNemar’s test. Respiratory motion artefacts and image quality scores were compared using Wilcoxon signed rank test. Interobserver agreement was assessed by Cohen kappa statistics.

Results

Compared with TWIST-VIBE, CAIPIRINHA-VIBE images had better scores for every parameter except higher noise score. Triple APs were always acquired with TWIST-VIBE but failed in 37% using CAIPIRINHA-VIBE: 11% have only one AP, 26% have two. Breath-holding time was the only parameter that influenced the success of multi-arterial techniques. TWIST-VIBE images had worst score for Gibbs and respiratory motion artefacts but lower noise score.

Conclusion

CAIPIRINHA-VIBE images were always diagnostic, but with a failure of triple-AP in 37%. TWIST-VIBE was successful in obtaining three APs in all patients. Breath-holding time is the only parameter which can influence the preliminary choice between CAIPIRINHA-VIBE and TWIST-VIBE algorithm.

Advances in knowledge

If the patient is expected to perform good breath-holds, TWIST-VIBE is preferable; otherwise, CAIPIRINHA-VIBE is more appropriate.

Introduction

Gadoxetic acid is widely used for liver imaging due to its hepatobiliary phase (HBP) imaging,^1–3 which may confer a diagnostic advantage. In particular, the diagnostic superiority of gadoxetic acid-enhanced MRI has been proven for the detection and characterization of small hepatocellular carcinomas (HCC),⁴ colorectal liver metastases,⁵ but also for differentiating between HCC and other hepatic lesions or hypervascular pseudolesions in the cirrhotic liver.¹ Thus, recent guidelines, including Liver Imaging Reporting and Data System (LI-RADS), the Asia–Pacific clinical practice guidelines and the Korean Liver Cancer Study Group guidelines, have included gadoxetic acid-enhanced MRI for the primary diagnosis of HCC alongside MRI performed with extracellular contrast media and CT.^6–9

Nonetheless, arterial phase (AP) images in gadoxetic acid–enhanced MRI can be suboptimal. Firstly, the conventional gadoxetic acid dosage (0.1 mL/kg) is equivalent to a quarter of the gadolinium concentration of extracellular contrast agents and to half the volume, due to its greater relaxivity.¹⁰ Moreover, gadoxetic acid has a shorter late AP window if compared to that of an extracellular contrast media. Additionally, the intravenous administration of gadoxetic acid can cause transient severe motion (TSM) in the AP, which further compromises image quality.

To date, the use of advanced MR sequences with reduced scan times have been proposed to reduce respiratory motion artefacts and obtain high-quality AP images.^11–14 Some studies also showed that a saline dilution of gadoxetic acid at 1:1 and a slower injection rate of 1 mL/s could be effective for acquiring images with fewer AP artefacts.^15–17

Nevertheless, achieving satisfactory or optimal AP images in gadoxetic acid-enhanced MRI remains a challenge in clinical practice. As a result, multi-AP imaging has emerged as an additional approach to address suboptimal enhancement.^11,18 Despite this, there has been only few studies assessing the differences among various techniques for multi-AP imaging.

Hence, the aim of our study was to evaluate the technical performance of two different multi-AP techniques, CAIPIRINHA-VIBE (Controlled Aliasing In Parallel Imaging Results In Higher Acceleration-Volumetric interpolated breath-hold examination; Siemens Healthcare, Erlangen, Germany) and TWIST-VIBE (Time-resolved angiography with interleaved stochastic trajectories; Siemens Healthcare, Erlangen, Germany), in patients who underwent liver MRI with gadoxetic acid.

Methods

Study population

This prospective study was approved by our institutional review board and informed consent was obtained from all patients.

The patient inclusion criteria were as follows: (1) patients aged more than 18 years, (2) underwent MRI twice within 12 months using two different multi-arterial algorithms, CAIPIRINHA-VIBE and TWIST-VIBE. Half of the patients underwent CAIPIRINHA-VIBE algorithm followed by TWIST-VIBE, while the other half had TWIST-VIBE first, followed by CAIPIRINHA-VIBE to ensure that any potential impact on breath-hold, that could be influenced by possible worsening of patient’s performance status, would have a similar impact on both acquisition sequences. All patients had clinical indications to execute gadoxetic acid-enhanced liver MRI, that is, suspect HCC, characterization of incidental focal liver lesion, suspected liver metastases, and follow-up of hepatic lesions.

For every patient, breath-holding time, body mass index (BMI), sex, and age were recorded before the second MRI.

Protocol

All MRI exams were carried out on a 1.5 T scanner (Magnetom Aera, Siemens Healthcare, Erlangen, Germany) with an 18-channel phased array coil. Patients were examined in supine position. Images were obtained in the axial plane, with a rectangular field of view (FOV 380-440mm), before and after intravenous injection of 0.025 mmol/kg gadoxetic acid (rate of 1 mL/s), followed by a 30-mL saline flush at the same injection rate, according to the MR protocol as in Table 1. AP acquisition commences 8 s after manual triggering (to allow for breathing instructions) on the descending aorta.

Table 1.

MRI protocol.

Without contrast media	Breath-hold dual T1w GRE, in-phase and out-of-phase sequences, axial, section thickness of 3 mm
After injection of gadoxetic acid	CAIPIRINHA-VIBE (TR 3.88 ms, TE 1.89 ms, FOV 380 mm, slice thickness 3 mm) or TWIST-VIBE sequences (TR 6.65 ms, TE 2.39 ms, FOV 380-440mm, slice thickness 3 mm) until 3’ delay, axial, section thickness of 3 mm
	DWI using breath-hold STIR EPI sequences (b value 0-50-400-800 s/mm2), 5 mm continuous sectioning ADC calculations with monoexponential data fitting
	Axial respiratory-triggered single-shot TSE T2w sequence, SPAIR fat suppressed, section thickness of 3 mm Axial respiratory-triggered single-shot TSE T2w sequence, not fat suppressed, section thickness of 3 mm
	GRE T1w VIBE HD sequence in the hepatobiliary phase acquired at 5’, 15’ and 20 minutes after contrast agent administration, flip angle 10° and 30°, axial and coronal (only 20’), section thickness of 3 mm

Abbreviations: GRE = gradient echo, CAIPIRINHA-VIBE = Controlled Aliasing In Parallel Imaging Results In Higher Acceleration—Volumetric interpolated breath-hold examination, TWIST-VIBE = Time-resolved angiography with interleaved stochastic trajectories—Volumetric interpolated breath-hold examination, DWI = Diffusion-Weighted Imaging, STIR EPI = Short-TI Inversion Recovery Echo Planar Imaging, ADC = Apparent diffusion coefficient, TSE = Turbo spin echo, SPAIR = SPectral Attenuated Inversion Recovery.

Multi-AP techniques

CAIPIRINHA-VIBE is an advanced 3D acceleration imaging technique that takes into account the receiver coil array's sensitivity profile whilst sampling k-space in a parallel imaging manner. In contrast to conventional parallel imaging techniques, CAIPIRINHA mitigates aliasing artefacts and noise while maintaining a high signal-to-noise ratio, ultimately enabling a higher acceleration (Figure 1).

Figure 1.

Arterial phase ((A) early; (B) intermediate; (C) late [22s calculated with care bolus]) acquired with CAIPIRINHA-VIBE in a 58-year-old patient. Acquisition commences 8 s after manual triggering (to allow for breathing instructions) on the descending aorta; each phase is approximately 5 s.

TWIST-VIBE is a radial k-space undersampling technique. The k-space is divided into two regions: the first contains the low frequencies that provides contrast information, which is densely sampled, whilst the second contains higher frequencies and is only partially sampled in each acquisition. To fill in the missing data in the high-frequency region, data from multiple acquisitions are copied into the raw data to calculate the whole image data (Figure 2).

Figure 2.

Arterial phase ((A) early; (B) intermediate; (C) late) acquired with TWIST-VIBE in a 64-year-old patient. Acquisition commences 8 s after manual triggering (to allow for breathing instructions) on the descending aorta; the first phase lasts approximately 7 s and the following two phases are approximately 5 s.

Image analysis

The images were evaluated independently and randomly by 2 abdominal radiologists with 10 and 12 years of experience in abdominal imaging, who were blinded to which technique was used, to the results of different AP acquisition in each individual and to the history of the patient.

Single hepatic AP and each subphase of dual and triple AP s were reviewed for every patient regarding respiratory motion artefact and rated using a 5-point Likert scale (1: no artefacts; 2: mild artefacts without diagnostic impairment; 3: moderate artefacts without diagnostic impairment; 4: severe artefacts with diagnostic impairment; 5: non diagnostic).

The presence or absence of noise and Gibbs artefacts were then reviewed. Noise was defined as an undesirable background interference or disturbance that affects image quality. Gibbs artefact, also known as truncation, ringing, or spectral leakage artefacts, typically appears as multiple fine parallel lines immediately adjacent to high-contrast interfaces (Figure 3).¹⁹

Figure 3.

Late arterial phase acquired with TWIST-VIBE affected by Gibbs artefacts.

Lastly, the general quality of images was assessed using a 5-point Likert scale (1: optimal exam quality; 2: good exam quality; 3: no effect on diagnostic exam quality; 4: image degraded but interpretable; 5: non diagnostic).

Statistical analysis

For patient characteristics (age, sex, BMI, breath-holding time), continuous variables are presented as mean ± standard deviation (SD) and categorical variables are presented as counts (percentage).

The McNemar test was used for comparison of the presence or absence of Gibbs artefacts and the presence or absence of noise (dichotomous variables). Respiratory motion artefacts and quality scores (ordinal variables) were compared using Wilcoxon signed rank test.

Binary logistic regression was used to analyse which variables influenced the feasibility of three AP s. Inter-reader agreement for image analyses were evaluated using Cohen Kappa coefficient. The kappa values were interpreted as follows: 0 = no agreement, 0-0.20 = poor agreement, 0.21-0.40 = fair agreement, 0.41-0.60 = moderate agreement, 0.61-0.80 = good agreement and 0.81-1.00 = very good agreement. For all the tests but the inter-observer agreement test, p < 0.05 was considered statistically significant.

All statistical analyses were performed using the IBM SPSS Statistics software for Apple Mac, version 28 (IBM Corp.).

Results

Patient demographics

We enrolled 154 patients who underwent liver MRI twice using two different multi-arterial algorithms: CAIPIRINHA-VIBE or TWIST-VIBE.

A total of 83 males and 71 women (mean age of 65 years (SD 13) with mean BMI of 26.1 (SD 4.42) and breath hold time of 17.68 s (SD 1.16) were enrolled. Patient demographic of the final included cohort is summarized in Table 2.

Table 2.

Patient demographic.

Characteristics (N = 154)
Age (mean)	65 years (SD 13)
Sex (female/male)
Male	83	53.9%
Female	71	46.1%
BMI (mean)	26.1 (SD 4.42)
Breath-holding time	17.68s (SD 1.16)

Number of AP s acquired

By using CAIPIRINHA-VIBE algorithm, 97 patients had three APs (63%), 40 patients (26%) had two and 17 patients (11%) had only one AP; therefore, only 17 patients did not obtain a multi arterial study.

Whereas, by using TWIST-VIBE algorithm all patients (100%) obtained 3 APs.

CAIPIRINHA-VIBE vs TWIST-VIBE

Images obtained with CAIPIRINHA-VIBE technique had always better breath-holding scores (P < .0001), quality score (P < .0001) and less Gibbs artefacts (P < .0001), while worst noise score. TWIST-VIBE images had lower score for Gibbs and breath artefacts, but better noise score compared to CAIPIRINHA-VIBE images with triple AP: in detail, 23.3% of patients in the phase without contrast media had noise (vs. 45.6% in CAIPIRINHA-VIBE, P < .0001), 17.8% in the first AP (vs. 28.9%, P < .004), 17.8% in the second AP (vs 30%, P < .03) and 16.7% in the third AP (vs 31.1%, P < .0009) (Figures 4 and 5).

Figure 4.

Intraindividual comparison. CAIPIRINHA-VIBE: (A) pre-contrast acquisition; (B-D) arterial phase; (E) portal venous phase. TWIST-VIBE: (F) pre-contrast acquisition; (G-I) arterial phase; (J) portal venous phase.

Figure 5.

Intraindividual comparison. CAIPIRINHA-VIBE ((A) pre-contrast acquisition; (B) arterial phase; (C) portal venous phase) is characterized by more noise but less artefacts whereas TWIST-VIBE (D) pre-contrast acquisition; (E) arterial phase; (F) portal venous phase) has less noise but more artefacts.

Influential factors for feasibility of multi-arterial study

We analysed which factor could influence the feasibility of multi-arterial study and we found that sex (P = .92), age (P = .76), and BMI (P = .83) did not have any significant difference, whilst the breath-holding time was recognized as the only parameter that could influence the performance of multi-arterial technique (P < .01).

Then, we studied if we could individuate a breath-holding threshold value that allowed deciding when to choose multi-arterial techniques while using CAIPIRINHA algorithm, but we did not find a statistically significant one.

Interobserver agreement

Inter-observer agreement was moderate for Gibbs artefacts (κ: 0.56) and the presence/absence of noise (κ: 0.58), while good for breath hold and quality score (κ: 0.71 and 0.78, respectively).

Discussion

Our study shows that images obtained using the CAIPIRINHA-VIBE technique always had better breath scores, quality scores and fewer Gibbs artefacts; rarely, it was possible to acquire only one AP image. On the other hand, TWIST-VIBE images had better noise score and it always allowed the acquisition of 3 APs. Breath-holding time was the only parameter that could influence the performance of multi-arterial technique.

Arterial phase imaging plays a crucial role in determining the LI-RADS score for focal liver lesions observed in patients with liver cirrhosis or chronic liver diseases.²⁰ Recently, the use of multi-AP imaging has emerged as a strategy to improve AP images while reducing the effect of TSM caused by gadoxetic acid. CAIPIRINHA-VIBE technique, a form of parallel imaging that skips specific K-space lines diagonally, offers rapid acquisitions and generally maintains high overall image quality, also in multi-arterial gadoxetic acid–enhanced liver MRI.^21,22 On the other hand, TWIST-VIBE utilizes radial K-space undersampling, enabling the acquisition of multi-AP images with both high temporal and spatial resolutions.²³ Both techniques present advantages over conventional single-phase acquisitions.

In their study, Ikram et al²⁴ investigated whether the use of a triple-phase arterial imaging enhances the detection of AP hyperintense lesions at 3 T compared to single-phase imaging in patients at risk for HCC. They found that triple-phase imaging offers better AP imaging for hepatic lesions, concurrently improving lesion contrast-to-noise ratio compared to single-phase imaging.

Pietryga et al¹¹ confirmed that an increase in temporal resolution achieved by the use of multiple arterial acquisitions could lead to optimal phase timing, even if employing a fix delay.^25,26 In their study, triple-phase arterial CAIPIRINHA imaging provided at least one satisfactory AP acquisition in 81% of patients analysed.

This holds significant clinical relevance, given that adequate AP imaging is crucial for the characterization of liver lesions.²⁷ Our findings support the potential advantages of employing different multiple-acquisition sequences based on patient's ability to hold their breath. Since both CT and MRI exhibit decreased sensitivity for lesions <2 cm,^28,29 it becomes crucial to prevent any additional reduction in imaging quality, such as that caused by breathing artefacts.

There are some limitations to this study. First, we did not conduct further assessment on the diagnostic accuracy and conspicuity of lesions in the different AP acquisitions. Therefore, validation of our results and their clinical impact should be investigated in future studies. Secondly, even though we conducted direct intraindividual comparisons, the two MRI scans for each patient were acquired within a 12-month timeframe, which is a relatively long interval between consecutive examinations. This introduces the possibility of variations in patients' condition that might influence the image quality. Nevertheless, conducting repeated MRIs with administration of contrast media in a short period of time is impractical and unethical in clinical practice.

In conclusion, TWIST-VIBE reliably offered a well-timed hepatic AP with better noise score in gadoxetic acid MRI. However, APs acquired with CAIPIRINHA-VIBE were most effective in reducing respiratory motion artefact, even though it was not always possible to obtain 3 APs. The way TWIST-VIBE acquisition of K-space is vulnerable to motion artefact as peripheral k-space data is shared through all subphases. Thus, the use CAIPIRINHA-VIBE can be considered for patients who may struggle with maintaining good breath-holds such as patients with severe systemic conditions and others who cannot follow breath-holding commands. For patients capable of holding their breath, TWIST-VIBE is preferable.

Funding

No funding was received for this study.

Conflicts of interest

D.Z. is an employee of Siemens Healthcare. The other authors have declared that no competing interests exist.