Knee dGEMRIC at 7 T: comparison against 1.5 T and evaluation of T(1)-mapping methods

BACKGROUND: dGEMRIC (delayed Gadolinium Enhanced Magnetic Resonance Image of Cartilage) is a well-established technique for cartilage quality assessment in osteoarthritis at clinical field strengths. The method is robust, but requires injection of contrast agent and a cumbersome examination procedure. New non-contrast-agent-based techniques for cartilage quality assessment are currently being developed at 7 T. However, dGEMRIC remains an important reference technique during this development. The aim of this work was to compare T(1) mapping for dGEMRIC at 7 T and 1.5 T, and to evaluate three T(1)-mapping methods at 7 T. METHODS: The knee of 10 healthy volunteers and 9 patients with early signs of cartilage degradation were examined at 1.5 T and 7 T after a single (one) contrast agent injection (Gd-(DTPA)(2−)). Inversion recovery (IR) sequences were acquired at both field strengths, and at 7 T variable flip angle (VFA) and Look-Locker (LL) sequences were additionally acquired. T(1) maps were calculated and average T(1) values were estimated within superficial and deep regions-of-interest (ROIs) in the lateral and medial condyles, respectively. RESULTS: T(1) values were 1.8 (1.4–2.3) times longer at 7 T. A strong correlation was detected between 1.5 T and 7 T T(1) values (r = 0.80). For IR, an additional inversion time was required to avoid underestimation (bias±limits of agreement − 127 ± 234 ms) due to the longer T(1) values at 7 T. Out of the two 3D sequences tested, LL resulted in more accurate and precise T(1) estimation compared to VFA (average bias±limits of agreement LL: 12 ± 202 ms compared to VFA: 25 ± 622 ms). For both, B(1) correction improved agreement to IR. CONCLUSION: With an adapted sampling scheme, dGEMRIC T(1) mapping is feasible at 7 T and correlates well to 1.5 T. If 3D is to be used for T(1) mapping of the knee at 7 T, LL is preferred and VFA is not recommended. For VFA and LL, B(1) correction is necessary for accurate T(1) estimation.