Quantitative IVIM parameters, molecular diffusion coefficient [Ds], fraction of the diffusion linked to microcirculation of the blood [f] and perfusion-related diffusion coefficient [Df] are being used as biomarkers for clinical diagnosis and treatment monitoring [1-2]. However, verification of the accuracy and reproducibility of IVIM is not performed routinely as it requires a flow phantom and a MR-safe flow pump with accurate and constant output. Otherwise, long extension of tubing would be required. Such system is complicated and costly and therefore no quality assurance of IVIM has been performed to our best knowledge. To provide a solution for such unmet needs, we have developed a practical IVIM quality assurance method based on a simple and compact flow phantom driven by a power injector which is available on most clinical MR scanners.

A flow phantom (180x180x180mm³) was made of acrylic and filled with distilled water as stationary fluid as shown in Fig.1(top). Six polyethylene terephthalate tubes with inner diameter of 10mm connected by L-shaped connectors were suspended inside the chamber and immersed in water. A Spectris Solaris EP MR Injection system (Medrad, Inc., Warrendale, PA) equipped with the MR scanner was used to inject up to 120 mL of distilled water through the tube at a precise rate of 0.1mL/s. Such settings generated a constant flow for about 20 minutes. The outflow of water was directed into a small reservoir and can be reused in multiple measurements or to verify the flow rate if necessary. Studies were performed on a Siemens Tim Trio 3T MR scanner. Eight slices perpendicular to the straight sections of connected tubes were prescribed to acquire IVIM images, as shown in Fig.1 (bottom).

The imaging parameters were: 1500ms TR, 88.4ms TE, 256x256mm² FOV, 256x256 acquisition matrix size, 5mm slice thickness. 16 b-values (0,25,50,75,100,150,175,200,225,250,350,450,550,700,850 and 1000s/mm²). A phased array head coil was used for image acquisition and the total acquisition time was 18.08 minutes, shorter than the duration of constant flow. As shown in Fig.1, three region-of-interests (ROIs) were placed on the diffusion weighted images: matching the cross-section area of the tube (ROI1, red circle), including both the tube and the stationary water (ROI2, yellow circle), and in the region of stationary water only (ROI3, green circle). Ds, Df and f were computed by two-step bi-exponential fitting of the change of MRI signal with b-values using the Levenberg-Marquardt algorithm. To demonstrate the feasibility of such IVIM QA method, three repeat measurements were performed on the same day and another three repeat measurements were performed one week later.

Fig.2-4 shows the typical fittings of IVIM model for the average signal intensity in ROIs 1–3. The IVIM parameters from repeated measurements are shown in Table1. Both slow and fast diffusion components are observed for ROI1 and ROI2 while only the slow component for ROI3 as expected. The Ds of ROI2 and ROI3 are almost identical and close to the known water diffusion coefficient at room temperature. For ROI1, the Ds and Df are different from those of ROI1 and ROI2. However, the f of ROI1 is also very high (~70%). Consistent results were obtained from multiple scans with the same session as indicated by small standard deviations and between week 1 and week 2 for ROI1 and ROI2 where f is at 22%~24%, close to physiological range 26%~31%[1].

The primary results demonstrate that a consistent combination of perfusion and diffusion can reproduced to serve as the standard for IVIM quality control. Such standard can then be used to test accuracy of different IVIM imaging techniques and protocols. The proposed method of MR IVIM quality assurance takes advantage of availability of power injector which is designed and periodically tested to produce accurate flow rate. Using the power injector avoids a dedicated flow pump. Since the power injector can be positioned near the MR scanner, it also allows short tubes, less fluid and easy setup. Although the power injector's minimum slow rate is 0.1ml/s, we have demonstrated the current rate is slow enough to acquire the data to achieve highly reproducible parameters. The quantification of IVIM parameters maybe further improved by averaging multiple slices to improve SNR and a more robust fitting algorithm. Given its simplicity and availability, we plan to use such QA method to test and compare different scanner types for IVIM imaging.

we have demonstrated that a fast, low-cost and effective quality control procedure for IVIM using a simple phantom and a power injector is feasible and can be easily implemented and performed on a typical clinical or research scanner.