INTRODUCTION: Quantitative Magnetic Resonance Imaging (MRI) sequences such as T2 and T2* mapping analyse biochemical changes of cartilage including the deterioration of the Extra Cellular Matrix (ECM) and changes in water content for early diagnosis of hip OA prior to serious cartilage degradation. Longitudinal studies and clinical assessments with T2 and T2* mapping rely on a high reproducibility for accurate diagnosis. We present a statistical analysis of T2 and T2* scan reproducibility using hip MR images. An automated cartilage segmentation process for true fast imaging with steady-state precession (FISP) hip MR images is used to retrieve biochemical information from co-registered T2 and T2* mapping images. METHODS: An anonymized dataset of unilateral hip images from 20 asymptomatic volunteers (M:F 10:10, aged 22-47 years, mass 50-90kg, height 163-189cm), was scanned with 7 T2 and 9 T2* scan repetitions. The subjects were classified as asymptomatic after an unbiased clinical investigation, individual scoring and structural MR examination, and had no previous hip surgery. A 3T MRI system (Magnetom Prisma, Siemens Healthcare, HIRF, Australia) was used where the asymptomatic hip for each subject was imaged with a 4-channel body matrix coil. Scans included: an isotropic T2 weighted 3D axial FISP Water Excitation scan (TR/TE: 10.18/4.3ms; Voxel-size (VS): 0.625×0.625×0.65mm; Field of view (FOV): 192mm; acquisition-time (AT): 155017.042500), sagittal T2 mapping sequences (TR/TE: 2080/18-90 ms; VS: 0.78×0.78×4mm; FOV: 200mm; AT: 94532.7775) and sagittal T2* mapping (TR/TE: 873/3.82-19.1ms VS: 0.625×0.625×3mm; FOV: 200mm; AT: 144334.39). Sagittal acquisition provides ideal visibility of anterior lateral cartilage which is important for clinical observations. An automated method using [1] was applied for the segmentation of the acetabular and femoral articular cartilages from the 3D-FISP images using clinically defined labels by the Steadman Philippon Research Institute (SPRI) [2]. The 3D FISP images were co-registered with the T2 and T2* images for the extraction of biochemical information. Statistical analysis quantified the reproducibility of scans with focus on the whole hip joint, femoral and acetabular plates and individual cartilage regions. The inferior regions of both femoral and acetabular plates were omitted due to the variability of T2 and T2* relaxation times in these thin cartilage regions. Bland-Altman plots displayed deviation between two repetitions, box-plots illustrated the spread of relaxation times and paired t-tests quantified statistically significant differences between repetitions. RESULTS: An example segmentation result of the 3D-FISP scan, and co-registered T2 and T2* scans of subject 3 is shown in figure 1, with the 3D segmentation volumes in the lower part of the figure. Boxplots of median relaxation times for selected hip joint regions are shown in figure 2 for the T2 and T2^* data respectively. Bland-Altman plots of median data are in figure 3 for the T2 and T2* data respectively. Statistically significant differences (p<0.05) were calculated in T2 data repetition scans: acetabular plate (mean data: p=0.0138), Femoral Superior-Anterior region (mean data: p=0.0377 and median data: p=0.0183), and Acetabular Mid region (mean data: p=0.0085). No statistically significant difference between T2* repetitions. DISCUSSION: T2* mapping is a more reproducible quantitative scan and has a lower variance within the data in comparison to T2 mapping. The median data in both T2 and T2* mapping provided no statistically different results between repetitions of clinically important articular cartilage regions and therefore these scans hold promise for biochemical early diagnosis of cartilage degradation in OA patients. However, it should be noted that thin cartilage regions can cause mislabelling of boundary voxels between regions in the co-registration of the structural and quantitative scans. Partial volume errors can also occur which create inconsistencies within repeated segmentations and therefore reduced repeatability of the experiments. CONCLUSION: T2* mapping is more reliable for application within a clinical environment. In longitudinal studies and within patient follow-up T2* mapping will ensure reproducible results of the relaxation times for accurate diagnosis of cartilage biochemical composition. These results support the feasibility of automatic quantitative T2* mapping of hip articular cartilage using structural 3D-FISP sequences.