The usefulness of the O-MAR algorithm in MRI skull base assessment to manage cochlear implant-related artifacts| ACTA Otorhinolaryngologica Italica

Abstract

Objective. To assess artifact size and MRI visibility when applying the “Orthopedic-Metal Artifact Reduction” (O-MAR) algorithm for cochlear implant (CI) scanning.
Methods. Two volunteers were submitted to 1.5 T MRI with an Ultra 3D CI receiver stimulator placed on their head. Four angular CI orientations were adopted: 90, 120, 135 and 160 degrees. Volunteers were scanned in each condition using T1w and T2w TSE sequences, as well as O-MAR sequences, in both axial and coronal planes. Quantitative comparisons were made of signal void and penumbra extent. Additionally, qualitative evaluations of global image quality, MRI readability with respect to 12 anatomical structures and visibility through the penumbra were undertaken.
Results. After application of the O-MAR protocol, the radius of the signal void reduced from 50.76 mm to 45.43 mm and from 49.22 mm to 40.15 mm on T1w and T2w TSE axial sequences, respectively (p < 0.05). Qualitatively, sequences acquired with O-MAR produced better outcomes in terms of image quality and anatomical depiction. Despite the area of the penumbra being increased for the O-MAR protocol, visibility through penumbra was improved.
Conclusions. Application of O-MAR may provide a complementary strategy to those already in use to obtain diagnostically useful MRI examinations in the presence of a CI, especially in case of skull base diseases requiring MRI monitoring.

Introduction

Historically, access to magnetic resonance imaging (MRI) has been limited in cochlear implant (CI) recipients due to the interaction of the external magnetic field with the CI magnet and the metallic components of implants. In the last decades, advances in CI technology have improved MRI safety and have overcome complications such as magnet displacement, demagnetisation and patient discomfort ¹. The novelties in MRI usability in CI recipients have led to the broadening of CI selection criteria to patients with skull base and intracranial pathologies which require MRI monitoring. The current literature has demonstrated growing interest about CI rehabilitation in skull base tumours, such as vestibular schwannoma (both sporadic and in NF2 patients), where the aim to rehabilitate hearing needs to take into consideration the issue of artifacts in MRI follow-up ². In this respect, the presence of a CI may reduce the diagnostic power of MRI in detecting disease evolution or recurrences. Image artifacts occur in the vicinity of the CI’s metallic components due to their magnetic susceptibility leading to their magnetisation when exposed to the MR environment. In these conditions, the devices exert their own magnetic fields, distorting the external magnetic field. This results in signal void areas and distortion signals, which cause concerns especially when attempting visualisation of near-by brain structures ³. Among the strategies employed in the endeavour to improve MRI image quality, tailored positioning of the CI in relation to the anatomic site to be explored has been described ³. In addition to this, the use of metal artifact reduction sequences (MARS) has been proposed. MARS is a general term, which refers to techniques intended to reduce the size and intensity of artifacts generated around metal components ⁴. Dedicated MR protocols have been developed especially in the orthopaedic field, with the aim of improving the image quality for patients with metallic implants ⁵. The application of these advanced techniques in CI recipients represents an emerging research topic, in which any potential benefit needs to be investigated. The current article aims to evaluate the effect of the Philips’ “Orthopedic-Metal Artifact Reduction” (O-MAR) algorithm on cochlear implant MRI artifact in terms of artifact size and MRI visibility.

Materials and methods

We undertook a preclinical study on two healthy adult male volunteers who provided written informed consent. The investigation was performed in the radiology department of our tertiary otologic referral centre after obtaining approval from our institutional review board and the medical ethics committee. An Advanced Bionics (AB) HiRes™ Ultra 3D CIs with Slim J electrode array was supplied for research purposes. For this investigation, a thin single-layer medical gauze was tied to the volunteers’ head. The implant package, with its magnet system in place, was then placed onto the head of each volunteer and held in place by a medical patch. No bandage or polyethene block were used to restrain the implant package. The CI was evaluated for different positions, each defined by a nasion-outer ear canal angle. Four angular orientations were adopted: 90, 120, 135 and 160 degrees. The distance from the centre of the magnet to the outer ear canal was 9 cm. For each of the four CI angular positions, both volunteers were scanned with the different MRI sequences.

Imaging study protocol

All imaging was performed using the Ingenia™ (Philips Medical Systems, Best, Netherlands) 1.5 T MRI scanner. The examination protocol was chosen in accordance with our institutional protocol (without gadolinium) for the investigation of brain pathologies. The MRI assessment involved planar T1 and T2 weighted (w) turbo spin echo sequences (TSE) with and without the application of O-MAR ⁶. Detailed MRI scanning parameters are reported in Table I.

Imaging analyses

In line with the recent study of Amin et al. ⁷, we analysed two specific features of the artefact: the signal void and the ‘‘penumbra.’’ The signal void was identified as the complete obscuration of image, whereas the penumbra was defined as the area of partial visibility. The evaluation included both quantitative and qualitative analyses.

Quantitative analysis

The signal void and penumbra sizes were measured for each sequence in the axial plane, using the IntelliSpace™ Portal (Philips Medical Systems, Best, Netherlands) certified reporting station. The maximum signal void radius was calculated by matching a circle to the visible edges of the complete obscuration within the brain. This was according to the method previously described by Canzi et al. ⁸. For measurement of the penumbra, a circle was matched to the visible edges of the partial visibility area. The radius of the signal void circle was subtracted from the radius of the penumbra circle to calculate the extent, or width, of the penumbra. The assessment of signal void radius and penumbra widths were repeated for each CI position in both of the volunteers. Mean sizes and standard deviations (SD) of the signal void and the penumbra were calculated.

Qualitative analysis

The MRI scans were submitted to two experienced neuroradiologists and one experienced otoneurosurgeon who independently evaluated them. A 4-point scale (0 = completely unusable, 1 = visible but not suitable for diagnostic purposes due to artifact, 2 = obscured by artifact but adequate for diagnostic purposes, 3 = high-quality view of the anatomic structure) was adopted to describe the following ipsilateral brain structures, with respect to the CI side: hypophysis, cochlea, semi-circular canals, vestibulum, internal auditory canal, cerebellopontine angle, brainstem, anterior part of the cerebellum, posterior part of the cerebellum, middle cerebellar pedunculus, cerebellar vermis and the occipital lobe. When unpaired median structures were described (e.g., hypophysis, brainstem, cerebellar vermis), the ipsilateral side of each structure was evaluated. The assessment was repeated for each MRI sequence across the two volunteers. Finally, MRI findings were analysed in order to investigate the following questions:

1. Does use of the O-MAR protocol improve the global image quality?
2. What are the benefits of the O-MAR protocol for each MRI sequence?
3. How does the anatomical visibility change when O-MAR is used for each CI position?
4. Does the use of O-MAR change the global visibility through the penumbra?

For assessment of penumbra, we adopted the scoring protocol introduced by Amin et al. ⁷: 1) severe obscuration; 2) moderate obscuration; 3) mild obscuration; 4) no obscuration. This evaluation was repeated for each sequence and CI position in both the volunteers comparing traditional and O-MAR sequences.

Statistical analysis

Statistical analysis was performed with R software (R version 3.1.3, R Development Core Team, R Foundation for Statistical Computing, Wien, Austria). Descriptive statistics, including the mean, standard deviation, median, minimum and maximum values, were calculated for all measures. The Shapiro-Wilk test was applied to assess for data normality. The Analysis of Variance (ANOVA) test was used to determine whether significant differences existed among the measures. The Tukey test was used as a post hoc measure. Significance for all statistical tests was predetermined at p < 0.05. Inter-rater reliability was calculated with Fleiss’ kappa.

Results

Quantitative analysis

Round-shaped artifacts were found in both traditional and O-MAR sequences acquired for axial planes. Employment of the O-MAR sequence significantly reduced the extent of the signal void area (Fig. 1). The maximum signal void radius was 50.76 mm (SD 2.33) and 49.22 mm (SD 7.38) on T1w and T2w TSE axial sequences, respectively. After application of the O-MAR protocol, the radius of the signal void reduced significantly to 45.43 mm (SD 2.18) and 40.15 mm (SD 4.86) on T1w and T2w TSE axial sequences, respectively (p < 0.05). On the other hand, the size of the penumbra increased after activation of the O-MAR protocol (Fig. 2). The penumbra width was 10.52 mm (SD 3.24) and 13.3 mm (SD 3.77) on T1w TSE sequences acquired without and with O-MAR, respectively (p = 0.05). For planar T2w TSE sequences, the size of the penumbra increased significantly from 8.73 mm (SD 5.3) to 11.43 mm (SD 4.79) after application of O-MAR (p < 0.05).

Qualitative analysis

1. Does the use of the O-MAR protocol improve the global image quality? When global image quality with and without application of the O-MAR protocol was compared, O-MAR sequences resulted in significantly better qualitative outcomes: mean image quality score was 2.64 (SD 0.93) with O-MAR vs 2.44 (SD 0.76) without O-MAR, (p < 0.0001).
2. What are the benefits of the O-MAR protocol according to each MRI sequence? When planar T1w and T2w TSE sequences were globally evaluated, application of the O-MAR protocol significantly improved visibility for both T1w and T2w. In the case of T1w sequences, the mean image quality scores were 2.37 (SD 0.98) and 2.57 (SD 0.80) without and with O-MAR, respectively, (p < 0.0001). For T2w sequences, global mean scores were 2.51 (SD 0.88) without O-MAR and 2.70 (SD 0.70) with O-MAR, (p < 0.0001). The sequences acquired in the coronal plane with O-MAR had better quality compared to those acquired in the axial plane. This applied to both T1w and T2w acquisitions (Fig. 3). In addition, the greatest improvement of image for O-MAR over traditional sequences was reported for acquisitions in coronal planes. Table II shows the difference in image quality score for sequences acquired with and without O-MAR.
3. How does anatomical visibility change with the use of O-MAR for each CI position? When each anatomical structure was analysed for the four different CI positions, visibility was significantly modified (p < 0.05) when the O-MAR protocol was used in the following cases:
   1. 90°: occipital lobe, semicircular canals, anterior part of cerebellum, posterior part of cerebellum (Fig. 4a);
   2. 120°: occipital lobe, semicircular canals, vestibulum, anterior part of cerebellum, posterior part of cerebellum, cerebellar vermis (Fig. 4b);
   3. 135°: occipital lobe, semicircular canals, anterior part of cerebellum, posterior part of cerebellum, cerebellar vermis (Fig. 4c);
   4. 160°: occipital lobe, semicircular canals, vestibulum, brainstem, anterior part of cerebellum, cerebellar vermis, middle cerebellar peduncle (Fig. 4d). Inter-rater reliability agreement ranged from 0.75 to 0.92 across all image evaluations and was consistent with a “high positive correlation” among the three investigators.
4. Does the use of O-MAR change the global visibility through the penumbra? The use of O-MAR significantly improved visibility through the penumbra (p < 0.0001), producing mean scores of 1.29 (SD 0.51) and 2.17 (SD 0.74) without and with O-MAR, respectively. Planar sequences acquired in the coronal plane with the O-MAR returned the best quality outcome in both T1w and T2w acquisitions: mean 2.36 (SD 0.68) for T1w sequences and mean 2.61 (SD 0.55) for T2w. The greatest advantages for penumbra visibility with O-MAR were found in coronal acquisitions compared to axial scans in both T1w and T2w sequences. Table III reports the difference in penumbra visibility scores in sequences acquired with and without O-MAR. Inter-rater reliability agreement ranged from 0.71 to 0.74 across all image evaluations and was consistent with a “high positive correlation” among the three investigators.

Discussion

A deterioration of MRI legibility in the presence of magnetic and metallic CI components occurs due to a combination of signal voids, image distortion, and signal inhomogeneity ⁹. Despite the recent introduction of adaptive CI magnet systems, designed to reduce the limits of MR compatibility, the artifact size generated in presence of different magnetic systems has been found to be similar ¹⁰. Several strategies are commonly used to reduce the severity of artifacts ^8,11-15, including MRI algorithm manipulation ^8,14,15. Simple concessions, aiming to improve image quality, have been traditionally adopted in MR acquisition techniques. These include: reduction of field strength, decreasing slice thickness, using fast imaging sequencies, adoption of fast spin echo sequences instead of gradient echo sequences, increasing the frequency encoding bandwidth or orienting the long axis of the metal along the frequency encoding direction ^4,16. In recent years, reduction of metallic artefacts has been attempted with the introduction of dedicated MRI protocols. MARS have been developed and applied predominantly in the orthopaedic and neurosurgical fields, demonstrating their clinical superiority for the depiction of periprosthetic tissues ^17-19 compared to conventional imaging sequences. On the other hand, to the best of our knowledge, the application of these advanced techniques for artifact reduction in MRI with CI have been reported only twice in the literature ^7,8. This is probably due to new CI magnet systems that have overcome the historical CI-MR incompatibility only recently. Amin et al. conducted a clinical study on 8 CI recipients implanted with different device models, matching a standard T1w sequence with a gadolinium and SEMAC-VAT sequence. They demonstrated that the application of SEMAC-VAT improved the visibility of the ipsilateral hemisphere ⁷. In contrast, Canzi et al. conducted an investigation on head specimens, reporting the improvement of global image quality score in planar axial T1w and T2w TSE sequences with the application of the O-MAR protocol ⁸. However, the visibility scores referred to each anatomic structure were not analytically compared between traditional and O-MAR sequences. In addition, only traditional surgical CI positioning was evaluated.

In the current study, the impact of O-MAR on MR image quality was investigated according to different sequences and acquisition planes, considering 12 intracranial structures and in relation to 4 different angular CI positions. The results of Canzi et al. ⁸ in their ex-vivo study were confirmed: a significant reduction in artifact size on both T1w and T2w TSE sequences was measured after activation of O-MAR protocol. Moreover, the current qualitative findings appear to corroborate the encouraging data in the published literature, demonstrating superiority for artifact reduction techniques in the depiction of intracranial structures over traditional sequences. To our knowledge, this is the first study in CI recipients in which the artifact reduction protocol has been applied for different planes of acquisition. The scans acquired in the coronal plane with the application of O-MAR showed the best quality outcome both for T1w and T2w acquisitions. This is in alignment with some clinical studies where traditional sequences reported better visibility in coronal plane views compared to axial ones ^7,13,19. Our image quality analysis concentrated on 12 intracranial structures, which were chosen by summarising the data from earlier studies ^7,8, concentrating on the posterior cranial fossa and on the anatomic structures of major otoneurosurgical interest. The improved visualisation of ipsilateral structures we found with the application of O-MAR is consistent with the data of Amin et al. ⁷. However, the number of anatomic areas considered in the study of Amin et al. was limited: internal auditory canal, cerebellopontine angle, cerebellar hemisphere and brainstem. Additionally, our analysis was provided by three independent observers with high positive correlation between them, whereas in the study of Amin et al. ⁷ the analysis had been conducted by consensus.

As a further novelty, we examined the advantages of MARS in visualisation of intracranial structures with reference to different CI positioning. According with a recent study of Canzi et al. ²⁰ focused on traditional MRI sequences, four rotational CI orientations were investigated. That work demonstrated, in accordance with literature ¹³, that the surgical location of the magnet represents a key point for the visual assessment of intracranial structures. Moreover, the present study points out that the use of advanced artifact reduction algorithms could make a further contribution to artifact reduction. More specifically, the occipital lobe, the semicircular canals and the anterior part of cerebellum had their visibility significantly improved for all CI orientations with the application of O-MAR. Interestingly, the highest number of structures for which visibility was improved with O-MAR was for the 160° orientation. This can be explained by the fact the brain structures we analysed mainly belonged to the posterior cranial fossa, which had been reported to have the worse visibility at a 160° CI position ²⁰.

We strongly believe the combination of customised CI placement with MARS application would have significant practical implications in challenging cases. For example, specific CI positionings have been demonstrated to allow the MRI postsurgical follow up after vestibular schwannoma resection and cochlear implantation ^2,13. In our opinion the employment of O-MAR can represent an additional tool for tumour follow-up. In accordance with Amin et al. ⁷, we introduced the evaluation of penumbra visibility in order to emphasise this aspect of the artifact. Our findings confirmed the trend already reported by Amin et al. ⁷, who demonstrated that the application of MARS induced an increase in penumbra size but also an improvement in visibility through it.

On the other hand, a few remarks about the application of the MARS in clinical practice deserve to be discussed. First of all, nowadays this technology is widely adopted especially in the orthopaedic and neurosurgical fields, although it may not be available in every clinical setting. Secondly, MARS are only applicable to certain sequences. For example, the O-MAR protocol can be used with T1w, T2w, PDw and short tau inversion recovery (STIR) contrast ⁶. However, this algorithm cannot be employed in the context of 3D fast spin echo, steady state free procession or diffusion weighted sequences, which are commonly used in otological field. Third, some MARS require additional scan time, representing a limit also in consideration of “conditional” MR safety recommendations made by certain CI manufacturers ²¹. However, this does not apply for all the existing artifact reduction protocols. For example, the O-MAR method does not increase scan time. Fourth, the commercial costs of this technology have to be considered, although it should be noted that the application of these algorithms is not limited to CI. Moreover, we might speculate the application of MARS could prevent in selected cases the additional costs of a surgical procedure for CI magnet removal.

Considering the present study, some limitations should be kept in mind. The investigations concerned only adult males, and other skull sizes were not investigated. In children, or women with smaller heads, a bigger artifact with reduced visibility of the intracranial structures might be expected. Furthermore, these results relate to the sole magnetic field strength of 1.5T. In addition to this, our analyses are limited to a single CI model and to only one MARS.

Conclusions

Preliminary evidence about the adoption of metal-artifact reduction algorithms when conducting MRI imaging with a CI are encouraging. However, confirmation from further studies is mandatory to corroborate the existing data. In addition, it can be expected that MARS would be effective not only with CI, but also in the presence of other neural implants. We believe these protocols may represent an additional strategy to be combined with others already in use, especially with reference to MRI surveillance of the main neurological diseases in neural implants recipients. In our view, the application of MARS could make a difference in particular circumstances to obtain diagnostically useful information with respect to specific structures or pathologies in cerebral MRI.

Conflict of interest statement

The authors state that this work has received funding from Advanced Bionics AG [Advanced Bionics, Stäfa, Switzerland] in the amount of 17,000€. Additionally, the authors confirm that the funders had no role in our study design, data acquisition, analysis and interpretation, and writing of the manuscript.

This does not alter the authors’ adherence to Acta Otorhinolaryngologica Italica policies on sharing data and materials.

Funding

The authors thank Advance Bionics AG (Stäfa, Switzerland) for loan of the cochlear implant and funding the study. Additionally, the authors confirm that the funders had no role in their study design, data acquisition, analysis and interpretation, and writing of the manuscript.

Author contributions

CP: conceptualisation; CP, CE, SA, LE, SA, MD, NM, MS: material preparation, data collection and analysis; BM, CL: supervision; CP, CE: writing. All authors have read and agreed to the final version of the manuscript.

Ethical consideration

This study was approved by the Institutional Ethics Committee of the IRCCS Policlinico San Matteo, Pavia (IRB# 20210027855). The research was conducted ethically, with all study procedures being performed in accordance with the requirements of the World Medical Association’s Declaration of Helsinki. Written informed consent was obtained from each volunteer for study participation and data publication.

Figures and tables

Figure 1.Assessment of signal voids. (A) Axial T1w TSE sequence without O-MAR, (B) with O-MAR.

Figure 2.Assessment of penumbra.(A) Axial T1w TSE sequence without O-MAR. (B) with O-MAR; blue arrow: width of penumbra.

Figure 3.Image quality. (A) mean sequence-specific scores and confidence intervals referred to T1w TSE images (dark bar refers to traditional sequences, light bar to O-MAR sequences, the mean score referring to O-MAR sequences is surrounded by a rectangle). (B) mean sequence-specific scores and confidence intervals referred to T2w TSE images (dark bar refers to traditional sequences, light bar to O-MAR sequences, the mean score referring to O-MAR sequences is surrounded by a rectangle). (C) axial T1w TSE sequences without (c1) and with O-MAR (c2) and coronal T1w TSE sequences without (c3) and with O-MAR (c4). (D) axial T2w TSE sequences without (d1) and with O-MAR (d2) and coronal T2w TSE sequences without (d3) and with O-MAR (d4).

Figure 4.Visual assessment of intracranial structures for: (A) 90° CI positioning. (B) 120° CI positioning. (C) 135° CI positioning. (D) 160° CI positioning. Dark bars refer to traditional sequences, light bars to O-MAR sequences. Image quality was evaluated merging data from both T1w and T2w sequences.

References

1. Srinivasan R, So CW, Amin N. A review of the safety of MRI in cochlear implant patients with retained magnets. Clin Radiol. 2019; 74:972.e9-972.e16. DOI
2. Sudhoff H, Gehl HB, Scholtz LU. MRI observation after intralabyrinthine and vestibular schwannoma resection and cochlear implantation. Front Neurol. 2020; 11:759. DOI
3. Canzi P, Luzzi S, Carlotto E. Customized cochlear implant positioning in a patient with a low-grade glioma: towards the best MRI artifact management. Otol Neurotol. 2022; 43:E628-E634. DOI
4. Jungmann PM, Agten CA, Pfirrmann CW. Advances in MRI around metal. J Magn Reson Imaging. 2017; 46:972-991. DOI
5. Khodarahmi I, Isaac A, Fishman EK. Metal about the hip and artifact reduction techniques: from basic concepts to advanced imaging. Semin Musculoskelet Radiol. 2019; 23:E68-E81. DOI
6. Hey S, Hoogenraad D, Elanchezhian V. Orthopedic metal artifact reduction. Distortion correction in the presence of a orthopedic implant.Publisher Full Text
7. Amin N, Pai I, Touska P. Utilization of SEMAC-VAT MRI for improved visualization of posterior fossa structures in patients with cochlear implants. Otol Neurotol. 2021; 42:E451-E458. DOI
8. Canzi P, Aprile F, Simoncelli A. MRI-induced artifact by a cochlear implant with a novel magnet system: an experimental cadaver study. Eur Arch Otorhinolaryngol. 2021; 278:3753-3762. DOI
9. Edmonson HA, Carlson ML, Patton AC. MR imaging and cochlear implants with retained internal magnets: reducing artifacts near highly inhomogeneous magnetic fields. Radiographics. 2018; 38:94-106. DOI
10. Todt I, Guerkov R, Gehl HB. Comparison of cochlear implant magnets and their MRI artifact size. Biomed Res Int. 2020; 2020:5086291. DOI
11. Wackym PA, Michel MA, Prost RW. Effect of magnetic resonance imaging on internal magnet strength in Med-El Combi 40+ cochlear implants. Laryngoscope. 2004; 114:1355-1361. DOI
12. Wagner F, Wimmer W, Leidolt L. Significant artifact reduction at 1.5T and 3T MRI by the use of a cochlear implant with removable magnet: an experimental human cadaver study. PLoS One. 2015; 10:E0132483. DOI
13. Todt I, Rademacher G, Mittmann P. MRI artifacts and cochlear implant positioning at 3 T in vivo. Otol Neurotol. 2015; 36:972-976. DOI
14. Sharon JD, Northcutt BG, Aygun N. Magnetic resonance imaging at 1.5 tesla with a cochlear implant magnet in place: image quality and usability. Otol Neurotol. 2016; 37:1284-1290. DOI
15. Majdani O, Rau TS, Götz F. Artifacts caused by cochlear implants with non-removable magnets in 3T MRI: phantom and cadaveric studies. Eur Arch Otorhinolaryngol. 2009; 266:1885-1890. DOI
16. Kolind SH, MacKay AL, Munk PL. Quantitative evaluation of metal artifact reduction techniques. J Magn Reson Imaging. 2004; 20:487-495. DOI
17. Sutter R, Ulbrich EJ, Jellus V. Reduction of metal artifacts in patients with total hip arthroplasty with slice-encoding metal artifact correction and view-angle tilting MR imaging. Radiology. 2012; 265:204-214. DOI
18. Jungmann PM, Ganter C, Schaeffeler CJ. View-angle tilting and slice-encoding metal artifact correction for artifact reduction in MRI: experimental sequence optimization for orthopaedic tumor endoprostheses and clinical application. PLoS One. 2015; 10:E0124922. DOI
19. Lee YH, Hahn S, Kim E. Fat-suppressed MR imaging of the spine for metal artifact reduction at 3T: comparison of STIR and slice encoding for metal artifact correction fat-suppressed T2-weighted images. Magn Reson Med Sci. 2016; 15:371-378. DOI
20. Canzi P, Magnetto M, Simoncelli A. The role of cochlear implant positioning on MR imaging quality: a preclinical in vivo study with a novel implant magnet system. Eur Arch Otorhinolaryngol. 2022; 279:2889-2898. DOI
21. Advanced Bionics. MRI Safety Information for the HiRes Ultra 3D cochlear implant.Publisher Full Text