Analysis of MR Imaging with FSE and DESS for the Diagnosis of Meniscal Tears in 316 Patients

Lee, Jae Gue; Ryu, Kyung Nam; Hong, Hyun Pyo

Abstract

PURPOSE: To evaluate the accuracy of a magnetic resonance(MR) imaging strategy that primarily uses fast spinecho(SE) sequences for the diagnosis of meniscal tears.
MATERIALS AND METHODS
The original clinical interpretations of MR images in 316 patients who underwent imaging for suspected internal derangement of a knee joint were correlated with results from subsequent arthroscopy (mean interval : 48.9 days). In all patients, MR examinations included double-echo fast SE T2- weighted sagittal and coronal imaging and double-echo steady state (DESS) sequence sagittal imaging. In 199 patients fat-suppressed conventional SE T1-weighted sagittal imaging was used. In cases in which interpretation was erroneous, imaging findings and arthroscopy reports were reviewed.
RESULTS
For ISO confirmed tears of the medial meniscus, sensitivity, specificity, and accuracy were 94 %, 93 %, and 94 %, respectively, while respective values for 147 confirmed tears of the lateral meniscus were 85%, 97 %, and 91%. These values are within the ranges recently reported for imaging strategies relying predominantly on conventional SE sequences. Of the 12 false-positive tears of the medial meniscus, five menisci showed a high signal contacting the surface on only one image and seven, that in all cases were located in the periphery of the posterior horn, showed such signal on more than one image. Of the six false-positive tears of the lateral meniscus, three menisci showed a high signal contacting the surface on only one image. Of the nine false-negative tears of the medial meniscus, eight menisci showed an abnormal signal that did not demonstrate definitive contact with the surface. Of the 22 false-negative tears of the lateral meniscus, 18 menisci showed this same type of signal.
CONCLUSION
Fast SE imaging of the knee can be an alternative to conventional SE imaging for the detection of meniscal tears. Most errors in our series were due to either an abnormal signal that failed to show definitive contact with the surface, a high signal which contacted the surface on only one image, or a signal of this type that was located in peripheral posterior horn of the medial meniscus, on more than one image.