Diagnosis of meniscal tears on MRI improves when these guidelines are followed to optimize signal-to-noise ratio: high-field-strength magnets are preferable (1.5 T and stronger); a high-resolution surface coil should be used; the field of view should only encompass the necessary structures and routinely be 16 cm or less; image slices should not be too thick (3–4 mm); and the matrix size should be at least 256 × 192 or higher [15]. Some authors have suggested that many tears are best seen on sequences with a low echo time, namely, T1-weighted and proton density sequences [15]. However, it is the authors’ experience that tears are often better seen on the T2 images. This is probably related to fluid being more pronounced on the T2 sequences and subsequently accentuates fluid-filled defects in the meniscal tear.

A normal meniscus is low signal on all sequences. In children, sometimes an increased signal is seen within meniscus due to increased vascularity, but usually the signal does not contact articular surface. It is important to know the age of the patient when interpreting the MRI. Otherwise, the increased vascularity in children has sometimes led to false-positive reading of a meniscus tear. Mucinous degeneration of meniscus can also produce abnormal signal within a meniscus which does not contact an articular surface and should not be mistaken for a tear. This has also been described as grade 2 signal [12, 16, 17] (Fig. 3.3).

The most common criterion for diagnosing meniscus tear on MRI is an increased signal extending in a line or band to the articular surface. Another finding is the abnormal size or shape of the meniscus, which would indicate damaged surfaces [12, 16, 17]. Studies by Kaplan and colleagues [18] and De Smet and coworkers [19] showed that the abnormal signal must unequivocally contact the surface of the meniscus. Although it may seem easy to read a meniscus as torn from an MRI, it is actually sometimes difficult to interpret whether the abnormal signal comes close to or actually touches the surface.

To provide a greater degree of accuracy, De Smet advocated the “two-slice-touch rule.” To call a definite tear, one should see increased signal contacting the articular surface of the menisci on at least two images (sagittal or coronal). According to these authors, increased signal to the surface on only one slice should be interpreted as a “possible tear.” Cases of only one abnormal slice correlated to tears at arthroscopy 55 % of the time for the medial meniscus and 30 % for the lateral [19] and only 43 and 18 % in a follow-up study 13 years later [20].

Accuracy of diagnosing meniscus tear with these criteria has been good. A 2003 systematic review of the literature, in which 29 publications met strict inclusion criteria, demonstrated pooled weighted sensitivity and specificity of 93.3 % and 88.4 % for the medial meniscus and 79.3 % and 95.7 % for the lateral meniscus, respectively [21]. Subsequent studies show improvement in these numbers [20, 22].

Most meniscal tears are visible and best seen on sagittal images. This is because most tears occur in the posterior horns [12]. However, one study showed that some tears are better seen or only seen on coronal images [23]. More recently, few reports suggest very thin axial images through the menisci to improve accuracy and tear description [24]. Fortunately, the MRI scan includes images in all three planes, and all three should be looked at carefully. Specific tear orientations vary and may only be clear in one of the three planes.

Whether a torn meniscus is reparable depends on the type or pattern of tear, its location, and the quality of the meniscal tissue. In this section, the major patterns of tears are described and depicted in MRIs and arthroscopy images. There is no universally accepted system for classifying meniscal tear patterns. A classification system developed by the International Society of Arthroscopy, Knee Surgery, and Orthopedic Sports Medicine [25, 26] is used here, and associated names when they are in common use are provided. These patterns include longitudinal-vertical, bucket handle, horizontal, radial, vertical flap, horizontal flap, and complex [25, 26].