It is known that ultrashort echo time (UTE) magnetic resonance imaging (MRI) sequences can detect signals from water protons but not collagen protons in short T2 species such as cortical bone and tendons. However, whether collagen protons are visible with the zero echo time (ZTE) MRI sequence is still unclear. In this study, we investigated the potential of the ZTE MRI sequence on a clinical 3T scanner to directly image collagen protons via D₂O exchange and freeze-drying experiments. ZTE and UTE MRI sequences were employed to image fully hydrated bovine cortical bone (n = 10) and human patellar tendon (n = 1) specimens. Then, each specimen was kept in a 30 mL syringe filled with D₂O solution for two days. Fresh D₂O was flushed every 2 h to reach a more complete D₂O–H₂O exchange. Later, the samples were lyophilized for over 40 h and then sealed in tubes. Finally, the samples were brought to room temperature and visualized using the identical 3D ZTE and UTE sequences. All hydrated bone and tendon specimens showed high signals with ZTE and UTE sequences. However, all specimens showed zero signal after the D₂O exchange and freeze-drying procedures. Therefore, similar to UTE imaging, the signal source in ZTE imaging is water. The ZTE sequence cannot directly detect signals from collagen protons in bone and tendons.

Osteoarthritis (OA) is the most prevalent degenerative joint disorder worldwide, causing significant declines in quality of life. The osteochondral junction (OCJ), a critical structural interface between deep cartilage and subchondral bone, plays an essential role in OA progression but is challenging to assess using conventional magnetic resonance imaging (MRI) due to its short T2 relaxation times. This study aimed to evaluate the utility of ultrashort echo time (UTE) MRI biomarkers, including macromolecular fraction (MMF), magnetization transfer ratio (MTR), and T2*, for in vivo quantification of OCJ changes in knee OA for the first time. Forty-five patients (mean age: 53.8 ± 17.0 years, 50% female) were imaged using 3D UTE-MRI sequences on a 3T clinical MRI scanner. Patients were stratified into two OA groups based on radiographic Kellgren–Lawrence (KL) scores: normal/subtle (KL = 0–1) (n = 21) and mild to moderate (KL = 2–3) (n = 24). Quantitative analysis revealed significantly lower MMF (15.8 ± 1.4% vs. 13.6 ± 1.2%, p < 0.001) and MTR (42.5 ± 2.5% vs. 38.2 ± 2.3%, p < 0.001) in the higher KL 2–3 group, alongside a higher trend in T2* values (19.7 ± 2.6 ms vs. 21.6 ± 3.8 ms, p = 0.06). Moreover, MMF and MTR were significantly negatively correlated with KL grades (r = −0.66 and −0.59; p < 0.001, respectively), while T2* showed a weaker positive correlation (r = 0.26, p = 0.08). Receiver operating characteristic (ROC) analysis demonstrated superior diagnostic accuracy for MMF (AUC = 0.88) and MTR (AUC = 0.86) compared to T2* (AUC = 0.64). These findings highlight UTE-MT techniques (i.e., MMF and MTR) as promising imaging tools for detecting OCJ degeneration in knee OA, with potential implications for earlier and more accurate diagnosis and disease monitoring.