This study examined a clinically relevant question—How do metallic dental restorative materials impact MR image uniformity? Practically, the situation represents patients with dental restorations who will undergo MRI of the face, jaws, and orbits. Although the susceptibility artefacts caused within the vicinity of the material are recognized, the full extent of these artefacts and their influence on image homogeneity at more distant sites is less well appreciated.

In this study, we investigated changes to image uniformity with different surface coils (Brain vs. HN), scanning planes (axial vs. coronal vs. sagittal) and with two MR sequences (SE and GRE). The study also examined the impact of these parameters on image uniformity in the presence of commonly used dental metallic materials (Au, Ag, Al, Au–Ag–Pd, Ti, and Co–Cr).

In the absence of metals, the two scanning sequences (SE and GRE) did not affect image uniformity irrespective of the coil type used or the selection of the primary scan plane. Interestingly, the SE sequence elicited high uniformity in the presence of Co–Cr and Ti. Furthermore, we found that the two coils (Brain and HN) did not disproportionately affect image uniformity on axial or coronal images irrespective of scanning sequence, with or without Co–Cr and Ti. The only exception was with the use of a rain coil which yielded high uniformity on the coronal images near Co–Cr. On the other hand, the HN coil elicited high uniformity on sagittal images irrespective of the scanning sequence in the presence of Co–Cr and Ti. The selected scanning plane had no effect on image uniformity irrespective of the scanning sequence or coil type in the no metal condition. Lastly, in the presence of Co–Cr and Ti image,non-uniformity was increased, whereas all the other metallic materials did not affect image uniformity.

Several methods have been developed to measure image uniformity. For instance, the American College of Radiology offers an alternative, precise calculation method [12]. In addition to the 17-point calculation method using peak deviation non-uniformity employed in this study, the NEMA method provides approaches to produce greyscale uniformity maps and determine the absolute averaged deviation uniformity. Although each method has its own advantages and drawbacks, we elected to employ the peak deviation non-uniformity methodology utilizing 17 points because the NEMA guidelines make this particularly suitable for evaluating image uniformity in surface coils [11].

Another point of discussion is the selection of the MR surface coils. Our results showed that the HN coil elicited high uniformity on sagittal images irrespective of the scanning sequence in the presence of Co–Cr and Ti or not, although the two coils did not affect image uniformity on axial or coronal images irrespective of the scanning sequence when Co–Cr and Ti were present or not with very few exceptions. It seems suitable to obtain a uniform image for this larger region because the HN coil is intended to cover a comprehensive range from the top of the head to the upper chest area. However, the Brain coil still has relatively high image uniformity.

Although the three scanning planes showed similar uniformity patterns irrespective of scanning sequences or coil type in the absence of metallic materials, it was impossible to compare NUI statistically between the three scanning planes because the cross-section through the center of the metallic materials differed among them. However, in the presence of Co–Cr and Ti, the increase in NUI value was greatest in the axial plane, followed by the sagittal and coronal planes irrespective of coil type or scanning sequence. In this study, the coronal planes were considered to be the most metal-insensitive plane.

In this study, two commonly used scanning sequences (SE and GRE) were employed. The primary disadvantage of GRE compared with SE is its susceptibility to the inhomogeneity of the magnetic field, particularly when metallic materials are present and exhibit high magnetic susceptibility (Table 1) [13]. In this study, the two scanning sequences did not affect image uniformity in any combination of coil type or scanning plane, and irrespective of the presence or absence of metallic materials. However, GRE elicited low uniformity in the presence of metallic materials with larger magnetic susceptibilities such as Co–Cr and Ti. This is practically relevant. For clinical diagnostic imaging, the presence of metal and its magnetic susceptibility should be ascertained, and the appropriate sequence can then be selected to minimize undesirable impact on MR image uniformity.

In clinical practice, a patient’s body, particularly in the oral and maxillofacial region, may contain metals and/or alloys (e.g., dental fillings, crowns and bridges, orthodontic appliances, dental implants and TMJ replacements). Consequently, it is imperative to assess the impacts of prevalent metallic materials on image uniformity. The present study investigated the use of noble metals, including Au and Ag, as well as noble metal alloys, such as Au–Ag–Pd. Base metals such as Ti (and occasionally Al), which are commonly used in medical and dental procedures (e.g., pins, screws and implants, etc.), were also investigated [14]. Finally, Co–Cr alloy, which are commonly used in medical and dental prostheses (e.g., artificial joints and dental implants), was also studied [15]. When any metallic material is set in a strong magnetic field, the material will be magnetized. The precise degree of the magnetization is referred to as magnetic susceptibility (Table 1). Previous studies have pointed out that metallic materials can cause large metallic artefacts on MRI images; however, whether they would also impair overall image uniformity was not understood [16]. However, Co–Cr elicited the lowest uniformity, followed by Ti with a larger magnetic susceptibility but not for any other metallic materials. The findings indicate that local non-uniformity originating from metallic materials other than Co–Cr and Ti did not significantly affect the overall uniformity in our phantom study. However, our results strongly indicate that the GRE sequence should be avoided in the presence of Co–Cr and Ti.

In this study, a 15 cm3 cubic phantom made of glass (SiO2) filled with 5% CuSO4 solution was used, in accordance with ASTM recommendations (e.g., well-known T1/T2 relaxation times for CuSO4, no artefacts or distortions in glass, etc.) [9]. Although the phantom size was a little smaller than a human head, the size and location of the metallic material (1 cm3 in size) were comparable to commonly encountered clinical situations (e.g., a tooth crown in the molar area). However, the phantom-based approach does not account for patient tissue-induced inhomogeneity. Second, patient-specific situations are more complex, for example, patients usually have more dental fillings, crowns, and bridges of various types which may result in different patterns of artefacts and non-uniformity compared with those in this study. Nevertheless, the current results are informative and provide a baseline to examine this issue in general. Perhaps patient scenarios likely to be affected include those with multiple contiguous implants and restorations, full arch prosthesis, etc. Future studies with a larger assortment of metallic materials are needed to better reflect the numerous relevant clinical situations.

The results of this study indicate that non-uniformity was exceptionally high in areas close to metallic materials with high magnetic susceptibility (e.g., Co–Cr and Ti, etc.). This non-uniformity may have little impact on evaluating the presence or gross extensions of a tumor in the oral and maxillofacial region. However, when assessing changes in inflammatory diseases, such as osteomyelitis, mucositis, etc., SI plays an important role as a diagnostic indicator. The use of SE sequence and HN coil selection are recommended when Co–Cr or Ti is in the vicinity of the lesion. In addition, coronal or sagittal imaging may be recommended under these conditions but imaging in the axial plane will likely be non-diagnostic. With the GRE sequence, Co–Cr caused high NUI values as far as 6.6 cm from the center of the material. However, the impact of this inhomogeneity on clinical diagnosis remains to be determined.