Correlation Between T1 MRI and Arthroscopy in Adults with Chondromalacia

3 and resliced along coronal and axial aspects and interpolated again along each aspect to 0.5 mm x 0.54 mm 2 to match the resolution of T1 -weighted images. Patient motion during image acquisition was found to be problematic for T1 quantification, especially in axial views where non rigid body rotation of the leg was observed. An initial manual alignment of the axial T1 -weighted images to the resliced axial T1-weighted image was performed by in- plane rotation and translation after which a region of interest containing cartilage and bone was selected. Masked, axial T1 -weighted images were exported to 3DVIEWNIX (MIPG, University of Pennsylvania, Philadelphia, Pennsylvania) where T1 -weighted images were registered to T1-weighted images. The cartilage was semi-automatically segmented from the T1-weighted images using a LiveWire algorithm (21) and masks were applied to all T1 -weighted images. Masked, coregistered T1 -weighted images were fit pixelwise to the linearized, monoexponential signal decay equation ln(S) = -TSL/T1 + ln(S0). T1 relaxation maps were viewed by two of the authors, who characterized lesions as either diffuse or focal in patients only. While diffuse lesions were quantified by the average compartment T1 (Table 1), focal lesions were quantified by mean T1 in an ROI drawn at the site of the lesion and possibly extending across multiple slices. Data Analysis. The difference between subjects and patients was performed using multivariate ANOVA and bootstrap confidence interval tests. Correlation between arthroscopy and T1 MRI was determined by calculating mean compartment T1 or calculating T1 mean of a large focally elevated region in a compartment. Results: Median T1 relaxation times among symptomatic and asymptomatic subjects were significantly different (p < 0.001) and symptomatic T1 exceeded asymptomatic articular cartilage median T1 by 2.5 to 9.2 ms. Patellar T1 was 2.5-8.3 ms higher than the tibial compartment (p < 0.01). In 8 observations of mild (grade 1 and 2) osteoarthritis at arthroscopy, mean compartment T1 was elevated in 5, but in all cases, large foci of increased T1 were observed. In 6 cases of moderate or severe chondromalacia, compartment mean T1 was elevated. Discussion: Correlation between mean T1 for the entire compartment and arthroscopy was excellent in all cases where the patient had grade 3 or 4 chondromalacia at arthroscopy, however, in cases of mild chondromalacia, the correlation was much weaker when averaging over a compartment. It must be understood, however, that those facets labeled as diffuse grade I changes, demonstrated considerable heterogeneity at arthroscopy with normal and abnormal cartilage intermixed and that the grade is assigned on the basis of the highest grade of abnormality present, not on the basis of some type of average. Thus, a compartment average would underestimate the arthroscopic grade and the use of at least reasonably large foci of elevated T1 to compare to the arthroscopic grade is justified in that it better corresponds to how the arthroscopic grade is assigned. One potentially confounding variable is the presence of clinically occult osteoarthritic changes in asymptomatic subjects. This would artifactually elevate the mean T1 in the asymptomatic subjects, making it more difficult to discriminate between normal and abnormal values. Conclusion: In conclusion, these results demonstrate that T1 relaxation mapping correlates with chondral lesions identified by arthroscopy. High correlation between T1 MRI and chondral damage was observed for grades 3 and 4 chondromalacia but correlation was only modest in the case of grade 1 or 2 damage when averaging across compartments.