MR diagnosis of meniscal tears of the knee: analysis of error patterns

Abstract

Despite high accuracy of magnetic resonance imaging (MRI) for diagnosing meniscal tears, MR findings do not always agree with surgical findings. We performed a blinded, retrospective study to analyze the nature and frequency of errors in the MR diagnosis of meniscal tears. Medical records of 100 consecutive patients who underwent MR and arthroscopy of the knee at our institution were reviewed. Twelve patients underwent prior meniscal surgery. Twenty-three patients had 27 discrepancies between MR and surgical findings. These were independently reviewed by two additional musculoskeletal radiologists in a double blinded fashion. Original incorrect diagnoses were categorized as either unavoidable, interpretation error or equivocal for meniscal tear. MR accuracy was 88% for the medial and 85% for the lateral meniscus. Of 27 incorrect MR diagnoses, 12 (44%) were unavoidable, 10 (37%) equivocal and 5 (19%) interpretation errors. Of the 67 medial meniscal tears, 12 (18%) were missed. Eight (67%) of these 12 were categorized as equivocal, including three postoperative menisci. Of 30 lateral tears, 12 (40%) were missed, 7 (58%) of which were categorized as unavoidable. Of these 12, 11 (92%) showed fraying of the inner edge, which was shaved at arthroscopy (n = 8) or had stable tear treated conservatively (n = 3). There were three false-positive diagnoses, all occuring in the lateral meniscus, two of which were unavoidable and one interpretation error. Of all missed lateral meniscal tears, most are unavoidable and related to confusion between what represents fraying and what represents a tear. Unavoidable false-positive diagnoses are infrequent and may be related to incomplete arthroscopic evaluation. Subtle or equivocal findings still make MR diagnosis difficult, even for experienced radiologists.