Introduction
Magnetic resonance imaging (MRI) plays a vital role in the diagnosis and characterization of brain tumours due to its high spatial resolution and optimal soft-tissue contrast [
1]. However, the key challenge in detecting tumours at an early stage by conventional MRI is its low sensitivity [
2]. Therefore, various contrast agents have been developed to improve the sensitivity of MRI [
3]. Gadolinium (Gd)-based contrast agents, including Gd-DTPA (Magnevist®), Gd-DO3A-butrol (Gadovist®) and Gd-EOB-DTPA (Primovist®), are the most commonly used contrast agents in clinical practice. Following intravenous administration, Gd-based contrast agents may increase MRI sensitivity for the detection of brain tumours [
2,
3]. However, Gd3 + chelates still have low toxicity. The risk of nephrogenic systemic fibrosis (NSF) for patients with impaired renal function and long-term adverse effects due to Gd brain deposition has raised concerns [
4‐
6], especially for patients with brain diseases that require longitudinal monitoring after treatment, such as low-grade gliomas or brain metastases. In diagnosing and treating most brain tumours, multiple gadolinium contrast agent-enhanced scans are required to determine the tumour grade, progression, and prognosis during follow-up. Furthermore, the effects of gadolinium deposits in the brain are still unknown; therefore, doses should be kept as low as possible to prevent gadolinium buildup [
7]. Thus, designing new Gd3 + chelates and using lower doses of contrast agents have always been popular research topics [
2,
8,
9].
The development of high- and ultrahigh-field MRI scanners (≥ 3.0 Tesla (T)) offers the possibility for using fewer contrast agents in the clinic [
9‐
11]. Many researchers have conducted comparative studies on injection doses at 1.5 T versus 3.0 T and 3.0 T versus 7.0 T and have shown the feasibility of injecting a reduced amount of contrast agents during higher-field MRI for brain tumours [
11,
12]. However, whole-body imaging at 7.0 T has not yet been approved by the Food and Drug Administration (FDA) and is not available in the clinic due to safety issues such as increased specific absorption rate (SAR) distributions in the body and technical issues such as B1 field inhomogeneity [
13,
14]. Recently, a 5.0 T clinical MRI scanner was developed that can be used to scan the whole body with good image homogeneity and contrast uniformity while avoiding issues such as a high SAR [
15]. Many basic MRI applications could benefit from the increased signal intensity (SI), contrast and spatial resolution of 5.0 T systems. However, the feasibility and image quality of lower-dose contrast-enhanced scanning with 5.0 T systems have not been investigated or compared with those of standard full-dose contrast-enhanced scanning with 3.0 T systems.
This study aimed to compare the enhancing effects of half-dose enhanced scanning at 5.0 T and full-dose enhanced scanning at 3.0 T in brain tumours. The enhancement effects were assessed using quantitative indices, including the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR), differences in SI before and after enhanced scans, as well as subjective image quality scores. Then, the feasibility of using half-dose contrast agents on 5.0 T MRI for brain tumour diagnosis was evaluated.
Discussion
The clinical routine for diagnosing brain tumours involves a full-dose contrast-enhanced MRI scan at 3.0 T. In this study, the Gd-based contrast enhancement of brain tumours using a half-dose at 5.0 T and a full-dose at 3.0 T were compared. Both quantitative and subjective evaluation results indicated that 5.0 T MRI with a half-dose of contrast enhancement may be a feasible option to meet the diagnostic requirements in the clinic.
The results indicated that the tumour-to-brain contrast, as reflected by the CNR of lesion/white matter and lesion/grey matter, was significantly greater with a half-dose at 5.0 T than with a full dose at 3.0 T. This finding was consistent with previous studies that compared lesion enhancement between high-field and low-field magnetic resonance imaging (MRI) systems [
11]. Moreover, we observed that the CNR increase in some patients and tumour lesions was even greater than the increase in magnetic field strength (i.e., more than 1.7-fold). This might be due to the following two reasons. On the one hand, the effectiveness of the T1-shortening effect of a Gd-based contrast agent increases nonlinearly with the field strength [
12]. In vitro experiments indicated that the r1 relativity of Gd-based contrast agents exhibited minimal variations across diverse field strengths [
20,
21], and the increased baseline tissue T1 relaxation times at higher fields amplify the relaxation-modifying effect of contrast agents [
12]. An increase in tissue T1 values with increasing field strength results in a corresponding increase in relative contrast enhancement. This is due to the combined effects of protein binding, which leads to increased field strength and solvent dependencies, ultimately resulting in notable changes in T1 relaxivity values at higher magnetic field strengths [
22]. Prior studies have also shown that the augmentation of channel head coils at elevated field strengths is advantageous for enhancing the signal-to-noise ratio and image resolution of voxels [
23,
24]. A 3D-GRE sequence was used in this study, and a similar TR and TE and the same flip angle were set for both 3.0 T and 5.0 T sequences. Therefore, when imaging with similar sequence parameters at both field strengths, grey and white matter may not relax completely and may exhibit a lower-than-expected SI increase at 5.0 T. Since the extensive invasion of glioma may impair the integrity of the blood‒brain barrier (BBB) [
25], tumour lesions with more severe BBB disruption would relax almost completely and exhibit high SI at 5.0 T. In addition, a previous study demonstrated that a 3D-GRE sequence was clinically more suitable for detecting brain tumours than other sequences [
26]. Hence, the augmented field strength and refined protocols employed in this study resulted in a substantial enhancement in the contrast between tumour tissue and brain tissue at 5.0 T.
In addition to the improved tumour-to-brain contrast, the SNR and CNR of grey matter, white matter, and tumour lesions were significantly greater with half-dose imaging at 5.0 T than with full-dose imaging at 3.0 T. This could be attributed to the greater magnetic field strength (5.0 T vs. 3.0 T) and the greater number of channels of the receiving coils (48 channels at 5.0 T vs. 24 channels at 3.0 T) [
27,
28]. Theoretically, a thinner slice or smaller voxel size would result in poorer (lower) SNR and CNR in the same magnetic field due to a decreased amount of aligning protons within the small voxel. The increased SNR and CNR obtained at 5.0 T even with a thinner slice and smaller voxel size indicates the extraordinary benefits of higher magnetic field strength, including clearer images at higher resolutions, which is beneficial for clinical applications [
12,
13]. Furthermore, the enhancement of SI at ultrahigh fields, together with the modification of transverse and longitudinal relaxation times, produces enhanced image contrasts that are useful in anatomical MRI applications. We also observed an increase in the grey matter/white matter contrast at 5.0 T versus 3.0 T, which was consistent with previous findings comparing 3.0 T with 1.5 T [
29]. A previous study revealed an increased CNR in T1w images at 3.0 T compared with 1.5 T [
29]; however, this increase was not found in a previous comparison study of 7.0 T versus 3.0 T [
11]. Since B1 field inhomogeneity might influence the CNR [
30], this finding at 7.0 T might be because the B1 field inhomogeneity at 7.0 T was more severe than that at 5.0 T and 3.0 T, thus leading to decreased grey matter/white matter contrast [
13]. A previous simulation study also revealed that the variation in B1 magnitude was nearly twofold greater at 7.0 T than at 4.0 T [
16]. Together with this previous simulation study [
31], our findings indicated that 5.0 T might have better B1 field uniformity than 7.0 T and similar uniformity to 3.0 T, thus leading to good grey matter/white matter contrast. Therefore, from a practical perspective, we can conclude that 5.0 T might be superior to 3.0 T in brain tumour imaging, not only because of the greater SNR of brain tissues but also because of better grey matter/white matter contrast than 3.0 T.
One limitation of this study is that only 12 subjects with 16 enhanced lesions were included, and some subjects had only mild BBB leakage. We hypothesized that patients with severe BBB disruption might benefit more from a half-dose at 5.0 T, but the CNR was only slightly improved in lesions with low-level BBB leakage. A previous dynamic contrast-enhanced imaging study suggested that low-level BBB leakage might induce systematic errors in the calculation of measured parameters [
32]. In some 3.0 T studies, a double dose is recommended for patients with brain metastases [
33]. Thus, our results suggested that the doses of contrast agent might need to be modified for patients with subtle BBB breakdown. In the future, more subjects with different types of brain tumours should be included to increase the generalization of our conclusions. In addition, only one contrast agent (Gadovist) and one type of T1-weighted sequence (GRE 3D) were examined in the current study; thus, our findings might not be applicable to other contrast agents and MRI sequences. However, a previous study suggested that Gadovist is a recommended contrast agent in routine MRI protocols for brain tumours [
34], the GRE sequence might be superior to fast spin‒echo sequences [
21]. Moreover, we only examined brain tumours, and future studies of tumours in various organs of the body, especially the abdominal area, should be conducted to investigate the full potential of low-dose contrast agent-enhanced MRI on 5.0 T systems. Another limitation of this study is that 3.0 T scanning was performed before 5.0 T scanning for all the subjects due to ethical considerations, and retention of the contrast agent in subsequent scans due to the leakage of contrast agent in tumours would be possible, despite the presence of at least a 24-h gap between the two contrast injection sessions. To rule out this limitation, a precontrast scan was performed to ensure little or no retention of the contrast agent, and the postcontrast images were subtracted from the precontrast images to calculate the SI difference. The precontrast images on the second scan (5.0 T ) showed no significant enhancement. The subtracted SI value was still significantly greater at 5.0 T, as shown in Table
3. Thus, we believe that the cumulative effect of contrast agents on tumour lesions is minimal. In the future, an earlier imaging session of 5.0T may be necessary to support the results of subsequent imaging studies.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.