Artificially produced cartilage lesions in small joints: detection with optimized MRI-sequences

The purpose of this study was to evaluate different magnetic resonance imaging (MRI) pulse sequences in the detection of artificial cartilage lesions in small joints using an animal model. A total of 32 artificial cartilage lesions were created in the knee joints of 20 rabbits. Twenty lesions were produced 4 weeks and 12 lesions 1 h before the MRI examination, performed in an extremity coil at 1.0 T. All joints were examined with five imaging sequences: (1) a conventional T1-weighted spin-echo (SE)-sequence (repetition time [TR] 600 ms, echo time [TE] 15 ms), (2) a T2-weighted turbo-SE-sequence (TR 2000 ms, TE 85 ms), (3) a two-dimensional (2D) gradient echo (GE)-sequence (TR 440 ms, TE 10 ms, flip angle 60 degrees) and (4,5) two three-dimensional (3D) GE-sequences (TR 40 ms, TE 7 ms, flip angle 40 degrees, with fat suppression (FS) and TR 30 ms, TE 9 ms, flip angle 40 degrees). Two examinations had to be excluded because of insufficient image quality and the remaining examinations were analyzed by two experienced radiologists. The MRI images were correlated with the pathologic findings and anatomical structures were scored according to a 5-level scale. Direct comparison of the pathological and MRI findings showed that 19 of the 30 artificially induced cartilage lesions were detected with the FS 3D GE-sequence, 13 with the 2D GE-sequence, 11 with the 3D GE-sequence, 3 with the T1-weighted SE-sequence, and 2 with the T1-weighted SE-sequence. The highest percentage of artificial cartilage lesions was demonstrated using a fat-suppressed 3D GE-sequence. However, the analysis of cartilage defects in small joints with optimized sequences as well as clinical routine hardware and software had limitations. Therefore a pilot study was performed analyzing newly developed high resolution FS 3D GE images obtained from 5 rabbit knees with 10 cartilage lesions at 1.5 T. Two sequences were used with 1.0 and 0.5 mm slice thickness, a matrix of 256 x 256 and 512 x 256 and a field of view of 12 x 6 and 8 x 6 cm. In this small subset detection rates were substantially higher than in the 30 rabbit knees examined before.