Study of Normal Hip and Lumbar Bone Marrow With Dynamic Contrast Enhancement Magnetic Resonance Imaging

MRI protocol Patients were examined on a 3T MR scan (Ingenia, Philips Healthcare, The Netherlands). Conventional sequences were acquired depending on the clinical problem. A T1 spin echo sequence imaged the right hip in the coronal plane. A previously described Dynamic 3D T1 Spoiled Gradient Echo covered the right hip (8). Its main features were as follows. 94 axial slices covered a Field of View (FOV) of 228 x 130 x 169 mm. TR, TE, flip angle and bandwidth per pixel were respectively 4.5 and 2.1 ms, 10°, 389 Hz. Acquisition and reconstruction matrix were 64 x 66 and 128 x 128 respectively. Temporal resolution was 13.5 seconds.

Three variable flip angles (VFA) sequences (3°, 10° and 17°) were acquired before injection. Each acquisition lasted 55 seconds. Five baseline scans were acquired. 0.1 mmol/kg of gadoteric acid (DOTAREM, Guerbet, France) were injected at the beginning of the sixth scan at a rate of 2.5ml/sec followed by 20cc of saline flush. Twenty dynamic scans were collected. Total examination time was 9 minutes.

Post-processing The investigators analyzed DCE images with the open-source software Osirix and DCE tool software (http://kyungs.bol.ucla.edu/software/DCE_tool/DCE_tool.html). A ROI was deposed in the common femoral artery to determine Arterial Input Function. T1 map was calculated from VFA acquisitions. The precise r1 relaxivity (3.4) of the contrast agent was introduced. These elements were used to calculate the time / gadolinium concentration curve. Tofts model was used.

The morphology of the curve was assessed visually according to the description made by van Rijswik (10). For each ROI, semi-quantitative and pharmacokinetic parameters were calculated: initial slope (IS), area under the curve (AUC), time to peak (TTP), transfer constant (Ktrans), rate constant (Kep) and extravascular-extracellular space volume (Ve). IS calculation included points 5 to 15. AUC and TTP were calculated from points 5 to 25. Parametric maps were obtained for the illustration of this work, but were not used for the analysis itself in order to avoid bias in the deposition of ROIs.