Use of MAVRIC-SL for metal artifact reduction in postoperative canine stifle MRI

Abstract

Importance

Cranial cruciate ligament rupture (CCLR) is a common cause of pelvic limb lameness in dogs. Surgical correction using metallic implants is commonly performed; however, using postoperative magnetic resonance imaging (MRI) for evaluation is limited owing to susceptibility artifacts.

Objective

This study aimed to determine whether a multiacquisition variable-resonance image combination-selective (MAVRIC-SL) sequence, which reduces susceptibility artifacts, is useful for postoperative evaluation.

Methods

MRI was performed on four dogs that underwent CCLR correction using metallic implants. Imaging was performed with MAVRIC-SL and two-dimensional fast spin echo (2D-FSE) sequences, including proton density (PD), PD fat-suppressed (F/S), and T2-weighted F/S. The artifact area for each sequence was quantitatively analyzed. A four-level qualitative image analysis was performed to assess image artifacts, cartilage delineation, and visualization of the cranial cruciate ligament and bone.

Results

The artifact area was significantly reduced with MAVRIC-SL than with 2D-FSE, regardless of the metal materials used. In addition, the amount of artifact, cartilage delineation, and visualization of the bone were significantly improved with MAVRIC-SL compared with 2D-FSE.

Conclusions and Relevance

In this preliminary study, the MAVRIC-SL sequence improves postoperative MRI evaluation by reducing susceptibility artifacts in dogs with CCLR treated with metallic implants.

INTRODUCTION

Cranial cruciate ligament rupture (CCLR) is a common cause of pelvic limb lameness in dogs [1]. The preferred treatment method is surgical correction with techniques such as tibial plateau leveling osteotomy (TPLO), tibial tuberosity advancement (TTA), and extracapsular stabilization [2]. Postoperative complication rates after TPLO and TTA have been reported to be as high as 20.0% and 26.5%, respectively [3]. Possible postoperative complications include tibial fracture, patellar ligament desmopathy, osteomyelitis, meniscal injuries, and implant loosening [4]. Furthermore, dogs treated for unilateral CCLR have a high risk of subsequent contralateral ligament rupture, and the operated stifle joint is prone to complications [5]. Postoperative magnetic resonance imaging (MRI) may be required in dogs with metallic implants, particularly when persistent or recurrent lameness, joint effusion, or pain suggests concomitant intra-articular pathology (e.g., meniscal injury, synovitis, or progressive osteoarthritic change) that cannot be fully characterized with radiography alone. Despite the challenge of metal artifacts, MRI provides a comprehensive assessment of intra-articular soft tissues and osteochondral structures, including menisci, articular cartilage, synovium, and subchondral bone, and has demonstrated diagnostic utility for clinically relevant stifle lesions in dogs [6].

Radiography plays an important role in the diagnosis of fractures and loosening [7]. However, radiography is limited in assessing intra-articular and soft tissue structures such as cruciate ligament remnants, menisci, and synovium, which frequently contribute to persistent lameness or joint effusion [6]. Ultrasonography provides a rapid and noninvasive method for evaluating the superficial periarticular soft tissues, including the patellar tendon and joint capsule, and may assist in detecting postoperative effusions or superficial infection [8]. Nevertheless, its diagnostic accuracy is highly operator-dependent and inadequate for deep intra-articular structures such as menisci [9]. Computed tomography (CT) offers high-resolution imaging of the osseous structure and implant geometry with reduced susceptibility to metallic artifacts compared with MRI [10,11]. Despite these advantages, CT provides limited soft tissue contrast and cannot directly assess the ligamentous or cartilaginous integrity [12].

MRI has become the gold standard modality for characterizing intraarticular and soft tissue abnormalities owing to its superior contrast resolution [10,13]. It enables direct evaluation of cruciate ligament fibers, meniscal tears, articular cartilage, synovial proliferation, and bone marrow changes that are not visible on other imaging modalities [14]. However, susceptibility artifacts observed on MRI can be induced by various factors such as hemosiderin from bleeding, metallic materials, or air/gas [15]. These artifacts result in geometric image distortions, areas of signal voids, and areas of high signal intensity surrounding the signal voids [2]. Susceptibility artifacts render an MRI scan nondiagnostic [16]; therefore, metal artifact reduction sequences (MARS), such as view angle tilting, slice encoding for metal artifact correction, and multiacquisition variable-resonance image combination (MAVRIC), have been developed and continue to be refined to mitigate susceptibility artifacts. In human medicine, MRI examinations utilizing optimized conventional pulse sequences and advanced metal artifact reduction techniques can significantly enhance diagnosis, prognosis, risk stratification, and surgical planning. In veterinary medicine, MARS, such as wavelet-transform-based reconstruction of artifacts (WARP)-turbo spin echo, have been demonstrated to reduce susceptibility artifacts; however, a previous study reported that WARP-turbo spin echo has a longer scan time [2]. Therefore, research on other sequences that reduce susceptibility artifacts is necessary in veterinary medicine.

This study aimed to assess the applicability of the MAVRIC-SL sequence for imaging dogs after CCLR correction using metallic implants. We hypothesized that MAVRIC-SL sequences would generate smaller susceptibility artifacts, leading to a more accurate evaluation of the bone-implant interface and surrounding soft tissues.

METHODS

Four dogs were registered as study participants by their owners. We included dogs that underwent CCLR reduction using metallic implants such as TPLO and extracapsular stabilization. All procedures were approved by the Institutional Animal Care and Use Committee of Chungbuk National University (CBNUA-25-0038-01).

All MRI examinations were performed using a 1.5T MRI scanner (SIGNA Creator; GE Healthcare, USA) with an 8-channel flex coil (GE Healthcare). All dogs were sedated with intravenous injections of butorphanol tartrate (0.2 mg/kg IV; Myung Moon Pharm., Korea) and midazolam (0.1 mg/kg IV; Bukwang Pharm., Korea), followed by anesthesia induction with propofol (4 mg/kg IV; Daewon Pharm., Korea). The endotracheal tube was inserted, and 1.0%–2.0% isoflurane (Terrell; Piramal Critical Care Inc., USA) was used to maintain an anesthetic flow rate of 1.0–1.5 L/min in oxygen. During anesthesia, all dogs were placed in dorsal recumbency, and a ventilator was used to maintain anesthesia. The MRI protocol comprised two-dimensional fast spin echo (2D-FSE) and MAVRIC-SL sequences. The 2D-FSE sequence included proton density (PD), PD fat-suppressed (F/S), and T2-weighted F/S (T2w F/S). Images were acquired using the PD, T2w F/S, PD F/S, and MAVRIC-SL sequences in the dorsal and sagittal; transverse and sagittal; sagittal; and dorsal, transverse, and sagittal planes, respectively. The imaging parameters are summarized in Supplementary Table 1.

All digital images were assessed using the Digital Imaging and Communications in Medicine (DICOM) viewer (RadiAnt DICOM Viewer software; Medixant Co., Poland). The MRI images were evaluated by one observer (SK) with three years of radiology experience for quantitative analysis of the artifact area. Artifact area was approximated per slice as (AP extent × CC extent) using the DICOM ruler tool [7], and slice-wise values were summed to obtain total artifact area per sequence, including the PD, PD F/S, T2w F/S, and MAVRIC-SL sequences. Thereafter, the artifact area was approximated by multiplying the AP and CC extensions, and all artifact areas from each slice were added. Two observers (SK and JP) with three years of radiology experience qualitatively evaluated the MR images. A four-level scale [7] was used to grade the amount of artifact, delineation of cartilage, and cruciate ligaments and bone visibility. For artifact assessment, grading was based on the extent to which susceptibility artifacts obscured the anatomical structures of interest: grade 1 indicated severe artifacts with markedly degraded diagnostic confidence; grade 2 indicated moderate artifacts with partial obscuration but preserved overall interpretability; grade 3 indicated mild artifacts with minimal impact on image interpretation; and grade 4 indicated negligible or no visible artifacts. Cartilage delineation was graded according to the clarity of cartilage margins, internal signal homogeneity, and the ability to distinguish cartilage from adjacent soft tissues and subchondral bone: grade 1 indicated poor visualization with indistinct cartilage margins and inability to reliably differentiate cartilage from surrounding tissues; grade 2 indicated partially defined margins with moderate blurring or signal heterogeneity but acceptable identification of cartilage structures; grade 3 indicated well-defined cartilage margins with minor blurring and good differentiation from adjacent tissues; and grade 4 indicated excellent visualization with sharply defined margins, homogeneous signal intensity, and clear separation from surrounding anatomical structures. Bone visibility was evaluated based on cortical definition, trabecular detail, and the ability to identify osseous structures despite the presence of metallic implants: grade 1 indicated severely impaired visualization with indistinct cortical margins and loss of trabecular detail; grade 2 indicated moderate degradation of bone visualization with partially identifiable cortical and trabecular structures; grade 3 indicated good visualization with clearly recognizable cortical margins and trabecular patterns, despite minor artifact-related distortion; and grade 4 indicated excellent visualization with sharply defined cortical contours and preserved trabecular architecture. However, because the cranial cruciate ligament had been surgically removed in the patients owing to CCLR, evaluation of the ligament was excluded from this study.

All analyses were performed using SPSS 21.0 (IBM SPSS Inc., USA). Normal distribution of the data was determined using the Kolmogorov–Smirnov test. The data were presented as the mean ± SD. The Wilcoxon signed-rank test was used to compare the quantitative data obtained from the 2D-FSE and MAVRIC-SL sequences. The Wilcoxon signed-rank test was performed to compare the artifact areas between the 2D-FSE sequences, including the sagittal PD, sagittal T2w F/S, and sagittal PD F/S sequences, and the MAVRIC-SL sequence. A two-tailed Wilcoxon signed-rank test was used to determine differences in image quality scores between the 2D-FSE and MAVRIC-SL sequences. The degree of agreement between the observers’ scores on qualitative image analysis was determined using Cohen’s kappa coefficient. The kappa coefficient values were evaluated as follows: < 0.20, slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement; and 0.81–1.0, near perfect agreement. A p value < 0.05 was considered statistically significant.

RESULTS

The mean age of the four dogs was 7.55 ± 3.0 years (range, 3–9.5 years), and their mean body weight was 32.3 ± 4.5 kg (range, 27–37 kg) at the time of the MRI scan. Three of the four dogs were spayed females, and one was a castrated male. The represented breed was the Labrador Retriever (n = 4). The presented clinical signs included pelvic limb lameness (2/4, 50%). The metallic implants were confirmed to be stainless steel (Table 1).

Table 1. Implant materials and surgical procedures in 4 dogs with cranial cruciate ligament rupture.

Case No.	Plate material	Screw material	Other implant	Procedure
1	Stainless steel	Stainless steel	NA	TPLO
2	NA	NA	Stainless steel wire	Extracapsular stabilization
3	Stainless steel	Stainless steel	NA	TPLO
4	Stainless steel	Stainless steel	NA	TPLO

NA, not applicable; TPLO, tibial plateau leveling osteotomy.

MRI revealed the following abnormalities: osteophyte of the adjacent stifle joint (4/4; 100%) (Fig. 1C and F), meniscal injury (4/4; 100%) (Fig. 1D and E), mineralization of the patellar ligament (1/4; 25%) (Fig. 2C), and visualization of the osteotomy line (1/4; 25%) (Fig. 3C and D). Among the abnormal findings, meniscal injury, visualization of the osteotomy line, and mineralization of the patellar ligament were only identified using the MAVRIC-SL sequence, likely because these findings were obscured by susceptibility artifacts on the 2D-FSE images. No abnormal findings in adjacent muscles or metal implants, such as dislocation, loosening, or breakage, were observed in any dog.

Fig. 1. Magnetic resonance imaging of the right stifle joint in case 1. Evaluation of the joint structures is limited by metal-induced artifacts in dorsal PD (A), sagittal PD (B), and transverse T2-weighted fat-suppressed (C) sequences. (D-F) Artifact reduction in multiacquisition variable-resonance image combination-selective sequences allows improved visualization of the stifle joint. Lateral meniscal extrusion is noted (thin white arrow), along with a meniscal tear (yellow arrow). An osteophyte at the lateral femoral condyle is identified (red arrows).

PD, proton density.

Fig. 2. Magnetic resonance imaging of the right stifle joint in case 2. Visualization of joint structures is limited by artifacts in the dorsal proton density (A) and transverse T2-weighted fat-suppressed (B) sequences. (C, D) Multiacquisition variable-resonance image combination-selective sequences allow improved assessment of the stifle joint. Mineralization of the patellar ligament is identified (yellow arrows).

Fig. 3. Magnetic resonance imaging of the left stifle joint in case 3. (A, B) PD sequences showing no visible osteotomy line owing to artifacts. (C, D) Multiacquisition variable-resonance image combination-selective sequences clearly depict the osteotomy line (yellow arrows).

The area of metal artifact measured on the MAVRIC-SL sequences was smaller than that measured on the 2D-FSE images for all patients. The mean artifact area on the Sag PD sequence was 230.34 ± 89.76 cm² (range, 111.62–321.00), Sag T2w F/S sequence was 261.87 ± 91.11 cm² (range, 168.35–353.47), Sag PD F/S sequence was 296.96 ± 116.58 cm² (range, 168.33–446.32), and MAVRIC-SL sequence was 58.21 ± 22.02 cm² (range, 43.27–93.78). The MAVRIC-SL sequences exhibited a significantly smaller artifact area compared with the Sag PD, Sag T2w F/S, and Sag PD F/S sequences (p = 0.0425) (Fig. 4).

Fig. 4. Metal artifact area for each individual sequence in dogs. It was confirmed that the metal artifact area was significantly reduced in the MAVRIC-SL sequence compared with the Sag PD, Sag T2w F/S, and Sag PD F/S sequences.

MAVRIC-SL, multiacquisition variable-resonance image combination-selective; Sag, sagittal; PD, proton density; T2w, T2 weighted; F/S, fat suppressed.

^*p < 0.05.

The mean score for the amount of artifact was 2.1 ± 0.5 (range, 1–3) for the PD sequence, 1.8 ± 0.75 (range, 1–3) for the T2w F/S sequence, 2.3 ± 0.51 (range, 2–3) for the PD F/S sequence, and 3.25 ± 0.46 (range, 3–4) for the MAVRIC-SL sequence. The amount of artifact was significantly improved in the MAVRIC-SL sequence compared with the PD, T2w F/S, and PD F/S sequences (p < 0.001, p = 0.008, and p = 0.008, respectively). The mean score for the cartilage delineation was 1.9 ± 0.44 (range, 1–3) for the PD sequence, 1.6 ± 0.61 (range, 1–3) for the T2w F/S sequence, 2 ± 0.53 (range, 1–3) for the PD F/S sequence, and 3.75 ± 0.46 (range, 3–4) for the MAVRIC-SL sequence. Cartilage delineation was significantly improved in the MAVRIC-SL sequence compared with the PD, T2w F/S, and PD F/S sequences (p < 0.001, p < 0.001, and p = 0.007, respectively). The mean score for the bone visualization was 1.6 ± 0.47 (range, 1–2) for the PD sequence, 1.3 ± 0.5 (range, 1–2) for the T2w F/S sequence, 1.6 ± 0.51 (range, 1–2) for the PD F/S sequence, and 3.3 ± 0.51 (range, 3–4) for the MAVRIC-SL sequence. Bone visualization was significantly improved in the MAVRIC-SL sequence compared with the PD, T2w F/S, and PD F/S sequences (p < 0.001, p < 0.001, and p = 0.007, respectively) (Fig. 5). Interobserver agreement for qualitative analysis was evaluated using Cohen’s kappa coefficient. The agreement between observers was moderate for the grading of the amount of artifact (κ = 0.482) and delineation of cartilage (κ = 0.452) and substantial for the evaluation of bone visibility (κ = 0.752).

Fig. 5. Image quality grading of amount of artifact, delineation of cartilage, and visualization of bone for each individual sequence in dogs. Subjective image quality was scored on a four-point scale (1 = poor, 4 = excellent). Artifact, cartilage delineation, and bone visualization scores were significantly higher with MAVRIC-SL.

MAVRIC-SL, multiacquisition variable-resonance image combination-selective; PD, proton density; T2w, T2 weighted; F/S, fat suppressed.

^**p < 0.01, ^***p < 0.001.

DISCUSSION

The MAVRIC-SL sequence demonstrated both quantitative and qualitative improvements in reducing susceptibility artifacts compared with conventional 2D-FSE sequences. Although artifact suppression was not uniform across the entire image, and some blurring affected the overall image quality, the sequence provided excellent soft tissue contrast and preserved the noninvasive advantages of MRI. These findings support the clinical feasibility of MAVRIC-SL for postoperative imaging after CCLR repair procedures, including TPLO and extracapsular stabilization. In this study, the artifact area was estimated using a simplified rectangular approximation based on the product of the AP extent and CC extents on each slice [7]. However, this approach may overestimate the true artifact area for irregularly shaped regions and has not been specifically validated for stifle MRI in veterinary patients. Therefore, the absolute artifact area values should be interpreted with caution. Nevertheless, because the same measurement method was consistently applied across all sequences, the relative comparison between MAVRIC-SL and 2D-FSE sequences remains valid. Future studies employing more precise segmentation-based methods may further improve the quantitative assessment of artifact burden.

MARS are excellent at reducing the occurrence of metal artifacts; however, their scan time is disadvantageously long because of the use of multiple spectral bins and phase coding added in the direction of slice selection [16]. In a previous study on metal artifact reduction, the acquisition time for MRI images using a 3T WARP-turbo spin echo, approximately 35 min per plane, was considered to be longer than desirable in clinical patients [2]. The MAVRIC-SL sequence uses 3D-FSE acquisition, requiring an increased number of excitations to acquire data from each excited slice, thus extending the scan time compared with 2D-FSE sequences [16]. However, in this study, the scan time was approximately 6 min per plane when the MAVRIC-SL was applied to the canine stifle joint. These results are believed to be related to the long echo train length (ETL) of MAVRIC-SL. Generally, ETL is inversely proportional to the image acquisition time [17]. In this study, the ETL was 19–23 when MAVRIC-SL was used; however, the ETL was 10 in the previous WARP-turbo spin echo. Although parameters related to signal-to-noise ratio were not measured in this study, no subjective differences in image quality were observed between the MAVRIC-SL and 2D-FSE sequences. Therefore, it is clinically applicable considering the relatively short scan duration compared with other reported MARS scan durations.

The abnormal MRI findings in this study included medial and lateral menisci, a tibial osteotomy line, and osteophytes. Among these, identification of the meniscus and tibial osteotomy lines was confirmed using only the MAVRIC-SL sequence. These structures are potential sites for postoperative complications after CCLR reduction [18]. Meniscal injuries affect the load distribution, lubrication, and stability of the tibiofemoral joint. Meniscal injuries can cause pain and joint effusions and may lead to osteoarthritis by predisposing the adjacent articular cartilage to increased axial and shear stress [4,19,20]. It is noteworthy that meniscal injury was observed in four dogs (4/4; 100%). Meniscal injury is a common concurrent lesion in dogs with cranial cruciate ligament disease, which may explain the high frequency observed in this cohort. One dog (1/4; 25%) had a lateral extrusion of the meniscus. Recently, there has been increasing interest in lateral extrusion of the meniscus in human medicine [21]. However, the exact cause of lateral meniscal extrusion remains unknown, and various causes, such as degenerative changes and complications after anterior cruciate ligament rupture repair surgery, are being considered [21]. If lateral extrusion is observed, further investigation may be required because there is no treatment other than surgical correction of the meniscus. Therefore, postoperative MRI is recommended to evaluate meniscal injuries and other anatomical structures. In addition, large-scale studies on lateral meniscal extrusion are necessary.

This study had several limitations. First, the metal artifact area was measured by a single observer during quantitative analysis. Second, the shapes and sizes of the metal implants varied. There are a few studies related to MAVRIC-SL sequences, and in veterinary medicine, clients may not be aware of the information regarding metal implants applied to their patients. Therefore, we aimed to confirm the results of the MAVRIC-SL sequences for metal implants of various shapes and sizes. Third, owing to the limited number of patients, detailed statistical analyses could not be performed, and statistically significant results could not be achieved. Fourth, the stifle joints were not visually evaluated using methods such as surgery or arthroscopy after MRI. Finally, the patients were mostly overrepresented by large breeds. Therefore, it is difficult to evaluate whether similar images can be obtained using the MAVRIC-SL sequence when scanning small dog breeds (< 5 kg).

In conclusion, the MAVRIC-SL sequence reduces magnetic susceptibility artifacts relative to 2D-FSE sequences. In addition, the MAVRIC-SL sequence could be a valuable tool for evaluating the surgical area in dogs with metallic implants after CCLR correction surgery. Future studies with a broader scope and prospective designs are necessary.

SUPPLEMENTARY MATERIAL

Supplementary Table 1

Magnetic resonance imaging parameters