Comporison of knee magnetic resonance imaging (MRI) and arthroscopic findings in the diagnosis of meniscal tear: ınvestigation of the causes of MRI positive and negative predictions

The aim of this study was to compare knee magnetic resonance imaging(MRI) with arthroscopic findings in the diagnosis of meniscal tear andinvestigating the causes of MRI?s positive and negative predictions.One hundred six patients who underwent knee MRI and arthroscopy formeniscal tear were included in this study. Their ages were between 19-76 years(mean 44.7). Forty-two of them were female and sixty-four of them were male.Fifty-nine patients were evaluated with 1.5 T MR System (GE Signa Excite,Wisconsin, WI, USA) and forty-seven patients were evaluated with 3 T MRSystem (Siemens Verio VB17, Erlangen, Germany). All MRI examinations wereinterpreted by two experienced musculoskeletal radiologists on the workstations.A definite meniscal tear was defined as visualization of an intrameniscal signalcontacting the free edge or articular surface of meniscus or the distortion ofmeniscus without meniscal surgery on two or more images or both sagittal andcoronal images. A possible meniscal tear was defined as visualization of anintrameniscal signal contacting the free edge or articular surface of meniscus ononly one MR image. Meniscal tears were localized as in meniscus anterior horn,body and posterior horn. A surgical classification was used for type of themeniscal tears. Meniscal tears were classified as being horizontal, flap, complex,longitudinal, radial or bucket handle tears. All of the MRI examinations wereassessed together by two radiologists in consensus and those were compared witharthroscopic diagnosis.96 of 106 cases at arthroscopy and 96 of 106 cases at MRI were detectedto have meniscal tears. At arthroscopy, 61 of meniscal tears were on medialmeniscus, 17 of meniscal tears were on lateral meniscus, 18 of meniscal tearswere on both medial and lateral meniscus. At MRI, 55 of meniscal tears were onmedial meniscus, 16 of meniscal tears were on lateral meniscus, 17 of meniscaltears were on both medial and lateral meniscus. At both arthroscopy (%30.2) andMRI (%44.3), the most common detected meniscal tears were complex tears onmedial meniscus posterior horn. Patient age, patient gender, MR magnet system,presence of knee joint effusion, trauma history, presence of bone contusion,presence of parameniscal cyst, presence of anterior cruciate ligament tears andpresence discoid meniscus were not statistically related to the diagnosis ofmeniscus tears. For diagnosis of meniscus tears in patients where the time intervalbetween MRI and arthroscopy was less than 30 days, MRI sensitivity was %94.1and accuracy was %91.1. For diagnosis of meniscus tears in patients where thetime interval between MRI and arthroscopy was more than 60 days, MRIsensitivity was %72.7 and accuracy was %73.9. In patients where the timeinterval between MRI and arthroscopy was more than 60 days, MRI accuracy fordiagnosis of meniscus tears in patients was found lower. When arthroscopy wasaccepted as the gold standard for diagnosing of meniscus tears; MRI sensitivitywas %88.5, specificity was %80, positive predictive value was %97.7, negativepredictive value was %92.1 and accuracy was %87.7. MRI was found to bestatistically related to the presence of meniscus tears (p=<0,001). Whenarthroscopy was accepted as the gold standard for diagnosis of meniscus tears,sensitivity, specificity and accuracy of MRI were found to be statisticallysignificantly high.The accuracy of MRI and arthroscopy for diagnosis of meniscus tears aresimilar and therefore MRI is an efficient imaging method for diagnosis ofmeniscus tears.Key words: Magnetic resonance imaging; meniscus tears; arthroscopy