Purpose: Nowadays, many focal therapies use Magnetic Resonance (MR) imaging for treatment planning purpose. In this context, geometric distortions should be as low as possible. In this paper, we present a new device to assess image distortions in MRI.Methods: The phantom consists of a stack of plates made of Polymethyl Methacrylate (PMMA). The detection elements are balls forming MRI capsules and delimiting an internal volume filled with paraffin oil. When imaged, this system enables having a regular pattern with the same distance in the three space directions between each capsule center of mass.Correction of the distortions is done using a registration of the detected balls pattern with the theoretical pattern. Iterative Closest Points (ICP) algorithm is used for thus purpose.Results: Preliminary evaluations on different MR sequences have been conducted, the combination of device and ICP-based correction method shows the feasibility of the approach.Conclusion: From the promising preliminary results obtained, future developments will include an upgrade of the phantom and the implementation of more sophisticated correction algorithms.