To date, only very few T2 mapping studies on knee cartilage have been performed with in situ loading, mainly because such experiments are strongly hampered by subject motion. The purpose of this work is to investigate the response of knee cartilage T2 to in situ mechanical loading in vivo for the patellofemoral and tibiofemoral joint compartments, using prospective motion correction to mitigate artifacts from load-induced subject motion.All experiments were performed on a Magnetom Trio 3T system (Siemens Healthcare, Germany), using an 8-channel multipurpose coil (NORAS MRI products, Germany) for signal reception. Knee loading was realized with an MR-compatible pneumatic loading device enabling accurate load adjustment in the range 0-50 kg. Prospective motion correction was performed with a moiré phase tracking (MPT) system (Metria Innovation Inc., Milwaukee, US), consisting of a single in-bore camera and a tracking marker, which was taped to the kneecap^1,2. For MRI of the patellofemoral joint under loading, the subject was positioned on the scanner bed with a knee flexion angle of 40°-50° (Fig. 1). For MRI of the tibiofemoral cartilage, the leg was loaded in nearly full extension. T2 mapping was conducted in healthy subjects with a 2D multiple spin-echo sequence (TE = [13.8, 27.6, 41.4, 55.2, 69.0, 82.8] ms, 11 slices, slice thickness = 3 mm, in-plane resolution = 0.6 mm). Slice positions were updated once per TR prior to the excitation pulse. Some T2 mapping experiments were performed with fat suppression to reduce residual motion artifacts from bright subcutaneous fat and bone signal. T2 maps were calculated from the acquired data using the Matlab-based StimFit algorithm, accounting for stimulated echo contributions arising from imperfect refocusing pulse profiles³.Fig. 2 shows T2 maps of the patellofemoral cartilage in a healthy subject under different loading conditions (0/20/40 kg), acquired with and without motion correction and with and without fat suppression. Images acquired without motion correction are not only strongly impaired by load-induced subject motion during acquisition, but also by position offsets increasing with the load. In contrast, the artifact level only mildly increases with the load when prospective motion correction is used and can be further reduced through fat suppression. Consistently higher T2 values were observed in the maps acquired with fat suppression compared to those acquired without fat suppression. The maps show a substantial load-induced T2 decrease in the superficial cartilage layers (Fig. 3). In particular, there is a thin line of strongly decreased T2 relaxation in the maps acquired with loading, marking the boundary between patellar and femoral cartilage. On the other hand, there is a deep cartilage layer with very short T2 values (T2 < 10 ms) close to the bone which gets thinner with increasing loads. Such load-induced T2 changes could not be observed in the T2 maps acquired from the tibiofemoral joint (Fig. 4).Motion was observed to be particularly severe for load experiments on the patellofemoral joint, which required knee flexion. However, the inter-scan position locking capability of our prospective motion correction implementation¹ enables a direct comparison of T2 maps acquired with different loads. Interestingly, substantial load-induced T2 changes could only be observed in the patellofemoral joint, but not in the tibiofemoral joint as observed by others^4–6, suggesting that changes in the tibiofemoral joint are more subtle and can only be unequivocally detected when results are averaged over larger subject cohorts. The observed T2 decrease in superficial cartilage is in line with results from the tibiofemoral joint by Souza et al.⁶ and with in-situ NMR microscopy measurements on knee cartilage plugs of pigs with concurrently acquired polarization light microscopic (PLM) reference images⁷. The PLM images in that work show that articular cartilage is composed of a thin cartilage layer with tangential fiber orientation at the surface. By applying mechanical loading to this cartilage structure, the anisotropy of this superficial cartilage layer is increased. Larger fiber anisotropy gives rise to increased residual dipolar coupling and thus enhances T2 relaxation. Therefore mechanical loading leads to decreased T2 relaxation values at the cartilage surface, which explains the thin boundary layer of very short T2 between femoral and patellar cartilage observed in the presented work. Souza et al. further reported a load-induced T2 increase in the deep cartilage layers, explaining this phenomenon with a fluid transport from superficial to deep cartilage layers⁶. However, our results suggest that this apparent T2 increase might rather be due to compression of the short T2 cartilage layers adjacent to the bone, reducing their influence on the overall T2 of deep cartilage.