Reporting knee meniscal tears: technical aspects, typical pitfalls and how to avoid them

There are limited possibilities to improve image quality for many of the mature MRI techniques in clinical practice today, and, therefore, the importance of radiofrequency coil design should not be underestimated [11]. Several different types of coils may be used for imaging the knee, including receive-only coils with one or multiple elements, and transmit/receive coils.

Flex coils are non-dedicated surface coils and their use results in several limitations, including a small field-of-view and heterogeneity of the signal intensity through the entire joint. Therefore, nowadays, the flex coils are not indicated for meniscal evaluation.

Quadrature coils or circularly polarized coils can be used as both transmit and receive coils and enable an optimal signal-to-noise ratio through the entire knee joint. The use of these dedicated high-resolution orthopaedic coils with different numbers of integrated elements is mandatory in order to ensure high homogeneity of the signal and high-resolution images (Fig. 1). As higher magnetic field systems (3.0 T, 4.0 T and 7.0 T) are used more and more often in clinical settings and research areas, the limits of performance in pulse sequence acquisition efficiency is approaching the limits [11]. Optimization of RF coils is seen as one of the solutions for further improvement [11]. Recently, dedicated 28-element RF coils with high acceleration factors have been developed for imaging at 7.0 T with very high-resolution (Fig. 2) [12]. However, all these developments imply significant costs and the decision makers must consider the benefits of the investment based on an analysis of effectiveness in relation to cost. The authors, however, strongly recommend the use of dedicated knee coils.

The clinical imaging protocol of the knee includes different MRI sequences with high-spatial resolution in all orientations: sagittal, coronal, and axial (Table 1). The spatial resolution should be maximized by using small field-of-view, thin slices, and high matrix size. The recommended field-of-view is 16 cm, but it also depends on the joint size and on the type of the radiofrequency coil used. Usually, the slice thickness is 3 mm or less, even with standard two-dimensional fast spin echo sequences. For axial images, the thinner the slice thickness, the better possible evaluation of the meniscal roots (Fig. 3). However, the increase in resolution decreases the signal-to-noise ratio. Although this is a limiting factor, it usually may not be a problem when dedicated knee coils are used at high magnetic field strengths (3.0 T and 7.0 T). Most clinical MR scanners on the market (1.5–3.0 T) provide excellent image quality even when high resolution images (in-plane and through-plane) are the desired goal.

Although MRI examination of the knee is relatively fast, it is not uncommon for the protocols to be “optimized” in clinical practice to shorten the acquisition time even more. However, technical improvements should be used to increase image quality and diagnosis accuracy, rather than to shorten the protocols.

Several meniscal lesions are best evaluated on sagittal images, including meniscal avulsion, radial tears (“ghost” meniscus sign), tears of the anterior horns, and posterior meniscocapsular separation (Figs. 4 and 5). Moreover, as many as 82 % of the meniscal tears are identified on sagittal images only [13]. The diagnosis mainly relies on the identification of an intrameniscal linear signal change and its contact with the meniscal surfaces. This includes that acquisition parameters are tailored such that potential meniscus tear can be detected, e.g. by having long echo time (TE) acquisitions for better fluid sensitivity. The number of the sagittal images in which the linear signal change contacts the meniscal surface may be an important factor in diagnostic accuracy and diagnostic confidence. The positive predictive value for diagnosing meniscal tears increases from 53 %, when the diagnosis is based on only one sagittal image in which the linear change extends to one of the meniscal surfaces, to 90 %, when the longitudinal linear signal contacts the meniscal surface in two consecutive sagittal images [5]. Therefore, in order to avoid false-positive reports, it is recommended that linear signal changes that contact the meniscal surface solely on one image are reported as possible tears [5, 10]. The false-positive diagnosis is even more likely in cases in which the intrameniscal signal contacts only the superior meniscal surface on sagittal images [5]. A correct evaluation of some particular types of tears (bucket-handle tears and radial tears) is difficult when it is based solely on sagittal images [13, 14]. In these cases, the images obtained in coronal and axial orientations play an important role.

Coronal images provide valuable information regarding meniscal shape and its attachments. Some types of meniscal tears are better characterized on coronal images combined with sagittal images when compared to sagittal images alone [13]. Other types of meniscal tears (e.g. small radial tears) may be seen only in the coronal plane [13]. The horizontal tears that are seen on sagittal images may not always be accurately described with regard to the relation with the meniscal surface. In those cases, the evaluation of the coronal images allows a better evaluation of the lesion, especially when located in the body of the meniscus [13]. The coronal plane is particularly useful in diagnosing bucket-handle tears, detached meniscal fragments, and meniscal extrusions beyond the tibial plateau (Fig. 6). Meniscal root tears are also highly accurately diagnosed on coronal images [15].

Axial images are always part of clinical knee protocols and are routinely used for evaluation of the patellofemoral joint and ligaments. In the evaluation of menisci, it has been shown that diagnostic accuracy and diagnostic confidence of axial images are lower only when compared to sagittal and coronal images [16]. However, axial images are complementary to sagittal and coronal planes and may better show the whole extent of a meniscal tear (Fig. 7) [16]. It was shown that radial tears, complex tears (i.e. flap tears), and displaced fragments may be better visualized and localized in axial planes when high-resolution images are obtained (Fig. 8) [16‐18].

The routinely used MRI sequences are two-dimensional (2D), fast spin-echo (FSE) sequences (proton-density (PD)-weighted, T1—weighted sequences, and T2- and T1—weighted sequences with and without fat suppression). When comparing with spin-echo (SE) sequences, FSE sequences facilitate the examination in a shorter overall time without statistically significant differences in meniscal tear detection rate [19]. By using high-magnetic field strength with dedicated coils, and parallel imaging techniques, the scan time may be decreased even more. Some particular types of meniscal tears have been proven to be better evaluated on specific sequences (e.g. coronal T2-weighted images showed a 96 % diagnostic accuracy compared with 85 % accuracy of PD-weighted coronal images for diagnosis of medial meniscal roots tears) [15].

Most authors still prefer standard two-dimensional over three-dimensional fast spin echo sequences and over gradient echo sequences, because the former provide classic, well-known T1-, T2-, and intermediate-weighted contrast information and enable detection of subtle abnormalities of the meniscus and ligaments. This is in contradiction to the fact that with three-dimensional (3D) gradient-echo (GRE) and 3D fast spin-echo FSE sequences, thinner sections with decreased partial volume averaging and isotropic voxel imaging with multiplanar reformations (that potentially reduce the examination time) are possible. Indeed, several studies have shown that 3D isotropic FSE sequences can provide rapid knee joint evaluation. By using 3D fat-saturated PD FSE sequences at 3.0 T with isotropic resolution between 0.4 mm to 0.7 mm with acquisition times between 5 and 10 minutes, the authors found similar sensitivities and specificities as the routine MR protocol for detecting knee joint lesions [20‐22]. However, the diagnostic performance of 3D sequences is not significantly increased in the overall evaluation of the meniscal tears [20, 23‐25]. The exception might be the diagnosis of meniscal roots tears that can benefit from the better visualization of small structures by using 3D FSE images [24, 26]. Other authors considered that the image quality of the 3D sequences is lower compared with 2D FSE sequences, and the diagnostic confidence of the radiologists is still higher with the 2D sequences regarding the meniscal evaluation [24]. This difference might be the result of the decreased visualization of low contrast structures, such as meniscus, with the 3D sequences [24].

Recently, synthetic-echo time post-processing techniques for generating images with variable T2-weighted contrast demonstrated high sensitivity and specificity in evaluation of abnormalities of menisci [27]. Commercially available synthetic imaging techniques are currently being developed by some vendors, i.e. for brain imaging, and it is only a matter of time before we will see further clinical trials in the musculoskeletal system as well [27]. MR fingerprinting may also be a future development that could dramatically change the way we assess meniscus. Both quantitative and qualitative evaluation could be possible from only one acquisition, as well as reconstruction of images with different contrast information [28]. However, MR fingerprinting and synthetic imaging has yet to be given research status, although they will be available soon by at least one vendor. Currently, synthetic techniques for meniscal imaging should not yet be used routinely as there is only very little evidence.

Another important development that will soon likely find its way into clinical routine (3–5 years) is the acceleration of 2D sequences by multi-slice acquisition techniques. The potential reduction of acquisition time is in the range of twofold to threefold [29, 30]. Since the original sequences and their contrast remain unchanged, the acceptance of these new sequences and their adoption to clinical imaging protocols will likely be very high amongst musculoskeletal radiologists.

Over the last decade, the expectations in terms of diagnosis accuracy has increased with the deployment of higher magnetic field strength. In many places, 3.0 T systems have replaced the 1.5 T equipment as standard of care. Musculoskeletal imaging is one of the fields that gains significantly from higher magnetic field strength when appropriate adaptation of pulse sequence parameters and high performance dedicated coils are used. The 3.0 T scanners have shown a significant improvement in visualization and evaluation of small structures such as ligaments, nerves, and articular cartilage compared to 1.5 T systems [31‐33]. In addition, 4.0 T and 7.0 T MR units are increasingly available. Until now, more than 65 installed MR systems worldwide operating at 7.0 T have been established as platforms for clinically oriented research. The introduction of the 7.0 T scanner has pushed the possibilities of morphological, functional, metabolic and diffusion-weighted imaging of the tendons, cartilage, and trabecular bone significantly forward (Fig. 9) [34‐37].

Although, there are data that show that higher magnetic field strength improves the diagnostic performance for anterior cruciate ligament (ACL) tears, cartilage and subchondral bone, the evidence regarding the added clinical value of higher fields MRI for meniscal tear evaluation is limited and is still a subject of debate [6]. Some authors consider that there is no significant difference for meniscal tear accuracy between magnetic field strengths ranging from 1.5 T to 7.0 T [6, 38‐40]. However, as a result of an increased spatial resolution and increased signal-to-noise ratio, the detection and characterization of small tears (e.g. of the free margin of the meniscus), small radial tears, or tears of the meniscal roots are better assessed when higher magnetic fields are used. Therefore, we consider that higher magnetic field strengths may improve the diagnosis accuracy and diagnosis confidence, and we recommend that knee examinations at 3.0 T scanners should be used, at this moment, as the new standard in clinical practice.

Whereas most radiologists and clinicians are well aware of classic signs for meniscal tears (in either the anterior horn, body or posterior horn), typical pitfalls in the diagnosis involve the meniscal roots and inter-meniscal connections, as well as the ligamentous attachment of the medial and lateral meniscus. Knowledge of the normal anatomy is key for diagnosis. The subsequent paragraphs provide a comprehensive overview of typical pitfalls; classic signs are briefly reviewed.

The anterior and posterior horns of the menisci are firmly attached to the tibial plateau through the insertional ligaments known as the meniscal root ligaments. They have an important role in knee biomechanics and kinematics by securing the meniscus in place and acting as anchors to the bone for both menisci [41]. Although meniscal root tears are less common than other types of meniscal tears, they can have a significant clinical impact. Cartilage defects and extrusion of the meniscus with respect to the tibial margin have been linked with tears of the meniscal roots [41‐46]. There are three types of meniscal root lesions: avulsion injuries of the attachments, radial tears, and degenerative changes. Usually, they are located within few millimetres from the bony insertion [47]. An accurate diagnosis of such lesions as well as of associated injuries are mandatory for the treatment decision, i.e. nonoperative versus operative, in order to avoid a poor clinical outcome and long-term prognosis. Meniscal root tears are, in many cases, unrecognized and neglected in the MRI reports. In the literature, it is reported that, e.g. one-third of the tears of the posterior medial root are missed [48, 49]. Visualization of the meniscal roots on MRI is challenging, but increased awareness of the normal anatomy and the imaging signs of meniscal roots tears may improve the diagnosis accuracy (Table 2). It is important to keep in mind that the anatomy of the meniscal roots is not uniform and that there are slight differences in terms of the exact attachment site as well as MR appearance.

An accurate description of meniscal tears is of utmost importance. A comprehensive report includes the exact localization, orientation, and extension of the meniscal tear. For the latter, it should always be noted to which surface the tear extends to (femoral surface or tibial surface). A clear description may obviate unnecessary surgery or may enable a better surgical planning.

The traditional description of a meniscal tear is that of a linear increased signal intensity within the meniscus. According to the orientation, a meniscal tear can be vertical, horizontal, or complex (Table 3). An horizontal tear is defined as a linear signal abnormality involving the surface of the meniscus in a horizontal orientation of less than 30° relative to the adjacent tibial plateau (Fig. 20) [5]. Vertical tears are subdivided into radial and longitudinal tears. Radial tears are perpendicular to the long axis of the meniscus and begin in the free edge of the meniscus (Fig. 4) [5, 75]. Vertical longitudinal tears are parallel to the long axis of the meniscus, away from the free edge (Fig. 5) [5, 75]. A complex tear refers to a combination of more orientations (e.g. parrot beak tear). Tears with displaced fragments such as bucket-handle tear, flap meniscus tear, or free meniscus fragment are also classified as complex tears (Figs. 21 and 22).

Meniscal grading, e.g. grade I, II, or III, which has been widely used in the past, is now considered obsolete in clinical routine imaging [79]. As with many classifications in musculoskeletal imaging, multiple versions of the same classification system were used, mixed up, or inconsistently used. This often caused misunderstanding, i.e. the referring physician could not distinguish between important and non-important lesions. Thus, a standard knee report may better distinguish between stable and unstable tears (Table 3), if there is a need for a classification system at all. Therefore, the orientation, the extension and involvement of meniscal surfaces (none, one, or both) are important features that should be recognized and described in the report. Stable tears (Fig. 20) have a potential for healing conservatively, whereas unstable tears often require surgery [1, 76]. It should be noted that the natural healing of meniscal tears might cause problems in the evaluation of follow-up MR images. Spontaneous healing is often associated with the presence of haemorrhage around the meniscal lesion [80]. Meniscal healing should not be misinterpreted as early re-tear.

The evaluation of post-operative meniscus can be challenging, e.g. when information about the type of prior surgery is not provided. The type of surgery defines the post-operative meniscus appearance. The two typical surgical approaches are meniscus preserving versus non-preserving therapies. The former includes meniscal suture, glueing, needle trephination or synovial abrasion. Non-preserving therapies include partial or total meniscectomy. Accordingly, normal intrasubstance changes in the operated meniscus after preservation surgery must be differentiated from a re-tear, which is usually characterized by a linear abnormal signal intensity that is more hyperintense than seen in healing meniscus. Intra-articular contrast application may help to differentiate re-tears (where the contrast is leaking into the cleft) from healing (where granulation and scar tissue fills up the cleft and where the contrast has no space to leak in). Missing parts of the meniscus may disclose non-preserving therapy. Sharp or truncated meniscal edges and loss of substance indicate meniscectomy.

A special category is the surgical repair of root tears. Usually, they challenge the surgeon in that the meniscus must be re-attached to bone. Surgeons may use arthroscopically assisted bone suture anchors (all-inside technique) or an intraosseous suture technique (“pullout technique”) [81]. To avoid misinterpretation of post-operative MR images, details about the type of surgery and the procedure are mandatory.