Proton density-weighted MR imaging of the knee: fat suppression versus without fat suppression

To prospectively evaluate the diagnostic accuracy of proton density-weighted imaging with and without fat suppression for detecting meniscal tears. The study involved 48 patients who underwent arthroscopy less than 3 months after proton density-weighted imaging with and without fat suppression. Sagittal images were independently reviewed by two radiologists for the presence of meniscal tears. Medial and lateral menisci were separately analyzed in terms of anterior horn, body, and posterior horn. Interobserver agreement was assessed using κ coefficients. The McNemar test was used to determine any differences between the two methods in terms of sensitivity and specificity. Arthroscopy findings were used as the diagnostic reference standard. Arthroscopy revealed 71 tears involving 85 meniscal segments: 34 medial meniscal segments and 51 lateral meniscal segments. The sensitivity, specificity, and accuracy of each radiologist were 95% (81/85), 92% (186/203), and 93% (267/288), and 93% (79/85), 93% (189/203), and 93% (268/288) when using fat-suppressed proton density-weighted imaging, and 91% (77/85), 93% (189/203), and 92% (266/288), and 91% (77/85), 93% (188/203), and 92% (265/288) when using proton density-weighted imaging without fat suppression, respectively. Interobserver agreement for meniscal tears was very high with proton-weighted imaging with (κ = 0.87) or without (κ = 0.86) fat suppression. There were no significant differences for detection of medial meniscal tears when using proton density-weighted imaging with or without fat suppression for both readers (p > 0.05). Fat-suppressed proton density-weighted imaging can replace proton density-weighted imaging without fat suppression for the detection of meniscal tears. (orig.)