1525 A novel and affordable DCE-MRI phantom : Prospective assessment of DCE-MRI breast protocols

Introduction T1-weighted dynamic contrast-enhanced (DCE) MRI breast sequences are a well-established component of breast MRI screening protocols. However, although international guidelines provide standardized recommendations [1], the wide array of possible sequence parameters results in a broad range of compliant sequences. These sequence differences can produce contrast agent (CA) enhancement curves with variable screening sensitivity [2]. Although standard test objects can be used to verify the dependence of image intensity on CA concentration, it is very difficult to prospectively assess the impact of common sequence parameter changes on CA uptake curves. This work used a novel test object, able to create reproducible CA enhancement curves, to prospectively evaluate the effect of common protocol alterations on two clinical breast protocols. Materials and Methods MRI Protocols: Two current clinical breast DCE-MRI protocols, performed with 3D fat-suppressed spoiled gradient-echo sequences, were selected for evaluation (Protocol 1: Siemens Aera 1.5T; Protocol 2: Philips Achieva 3.0T; see Table 1). Each dynamic series consisted of one pre-contrast and eight post-contrast axial images with CA (DOTAREM; Guerbet, France) administered at the end of the pre-contrast image. Experimental Set-Up: The test-object consists of a 40 mm inside-diameter (ID) sphere divided into four inter-linking compartments with a liquid input and output (each 3 mm ID) (see Figure 1). The input was connected to an automated clinical injector (MedRad, USA) and a standardised procedure for injection of water and CA was adopted: [i] 8 mL water at 0.14 mL/s; [ii] 2 mL of 0.05 mM CA at 0.5 mL/s; [iii] 72 mL water at 0.14 mL/s ensuring that CA was injected following the first dynamic image acquisition. The signal intensity of each image was evaluated from a rectangular region of interest (ROI) placed on the output (see Figure 1): the dilution of CA within the sphere modifies the input CA concentration curve, resulting in a CA enhancement curve with wash-out characteristics [3]. In a separate reproducibility study at high temporal resolution (3.3s), variability was measured at ±3% at peak enhancement. Sequence Investigation: Two sequence alterations were selected for investigation: CA injection timing and k-space sampling pattern. (i) For each protocol, CA was injected at 15s and then 30s earlier; (ii) A motion compensation strategy was applied to each protocol whilst maintaining the length of the dynamic sequence. Protocol 1 was altered to a random k-space sampling pattern (dynamic acquisition = 56.6s) whilst radial sampling was applied to Protocol 2 (dynamic acquisition = 55.7s). Data Analysis: Relative enhancement (RE) (percentage increase of signal intensity: ([SI – SIpre-contrast]/SIpre-contrast x 100) was calculated for each dynamic sequence from the signal intensity of the output ROI. Baseline repeatability was assessed for each clinical breast protocol from three subsequent dynamic acquisitions, and expressed as the relative standard deviations (%RSD) at peak and final relative enhancement (see Table 2). Test-retest repeatability was evaluated on Protocol 1 on two separate occasions under the same experimental conditions (see Figure 2).