Poster - 05: Automated analysis of MR distortion using a novel anthropomorphic phantom and open source software




MRI in stereotactic radiosurgery is the primary imaging modality, but due to inherent distorting effects MR images have significant geometric distortions. In this work we present framework to calculate, visualize, track MRI distortion using a combination of open source tools, and in-house software on a novel anthropomorphic head phantom.


The phantom used in this study was an anthropomorphic skull containing a 3D orthogonal grid of rods whose intersections are used as principle points for CT and MR image comparison. High resolution CT images were taken of the phantom. MRI images of the phantom were obtained on our 3T MRI system using a three-dimensional T1 weighted sequence. MRI data was collected with and without using Siemens model based 3D distortion correction method. MRI images were automatically rigidly registered to the CT images. For comparison, images were automatically rigidly registered using the Velocity software. The same process was repeated using deformable imaging methods. A point-by-point comparison of the centroids generates a distortion map which can be applied to subsequent patient MR images. This distortion map is compared to a baseline value and monitored over time using an open source tracking software.

Results and Conclusions:

We have developed in-house software to identify principle points within the phantom using CT imaging to a high spacing accuracy. With the current implemented imaging optimization, this open source software framework has promise in creating accurate distortion maps, and and thus could be used in a stereotactic radiosurgery setting for routine quality assurance.