Purpose: The purpose of this study was to evaluate the sensitivity and specificity of magnetic resonance imaging (MRI) in the diagnosis of medial and lateral meniscal injury in operated and non-operated knees.

Material and methods: This prospective longitudinal study was conducted between January 1st 1995 and December 31st 1997. Each patient had a physical examination, a standard radiography study, and an MRI and an arthroscopy. The MRI was performed with two machines running at 1.5 Tesla and 0.5 Tesla. Spin-echo T2 slices were obtained in the saggital plane, and spin-echo fat saturation slices in the frontal and axial planes. Arthroscopy was performed and/or controlled by the same operator.

Results: The protocol included 132 patients. Ten patients (ten knees) had another arthroscopy after arthroscopic meniscectomy. Mean time between the physical examination and MRI was 57 days, it was 69 days between MRI and arthroscopy. Sensitivity, specificity, positive predictive value, negative predictive value and precision were, respectively, 94.8%, 61.%, 86.7%, 81.5% and 85.6% for the medial meniscus and 73%, 93.3%, 76% 93.5% and 90.1% for the lateral meniscus.

Discussion: Arthroscopy remains the gold standard for prospective comparative studies (Reigher 1986, Jackson 1988, Kelly 1991). MRI is the examination of choice for the diagnosis of meniscal injury in non-operated knees (Polly 1988). Most false positives concern injuries located on the posterior part of the medial meniscus. For radiologists (Mink 1988), these false negatives would result from poor analysis by the arthroscopic surgeon (Quinn 1991). Arthroscopists point out that radiologists overestimate injuries of the posterior segment (Barronian 1989, Fischer 1991, Spiers 1994). These false negatives involve the lateral meniscus (Cheung 1997). Our study corroborates the results reported in the literature. For repeated arthroscopies, no conclusion can be drawn from the interval of confidence observed in a population of ten knees. Nevertheless, in these knees, the MRI provided a good means of identifying recurrent lesions of both the medial and lateral menisci.

Conclusion: The physical examination provide a strong clue to meniscal injury in non-operated knees, so it is not necessary to perform an MRI before arthroscopy In all other cases, particularly there is recurrence, MRI can identify injury to the medial or lateral menisci with good sensitivity and specificity. In most cases, if the MRI is negative, it is not necessary to perform arthroscopy to search for a meniscal injury.