Encrypted login | home

Program Information

A Phantom Study of MR Geometric Distortion and Its Orientation Dependency On a 1.5T MR-Simulator

no image available
M Law

M Law*, J Yuan , G Lo , O Wong , A Ding , K Cheng , K Chan , K Cheung , S Yu , Hong Kong Sanatorium & Hospital, Hong Kong

Presentations

TH-CD-207-12 (Thursday, July 16, 2015) 10:00 AM - 12:00 PM Room: 207


Purpose: Gradient non-linearity and B0 inhomogeneity introduce system-dependent MR image distortion that violates geometry fidelity required for radiotherapy. In addition, gradient non-linearity induced distortion can be acquisition orientation dependent, while this effect has rarely been investigated. This study aimed to quantitatively assess 3D geometric distortion of MR images acquired in different orientations with a dedicated 1.5T wide-bore (70cm diameter) MR-simulator, using a customized geometry accuracy phantom.

Methods: A polyurethane phantom (WxLxH=55x55x37.5cm3) was constructed with 4276 spherical markers (6mm diameter, MR/CT visible) arranged on a 3D grid (25mm isotropic-interval). Phantom reference images were acquired on a CT-simulator. Axial, coronal and sagittal MR images were individually acquired with 3D fast-spoiled-gradient-echo sequence (TR/TE=5.8/2.5ms, FOV=50cm, isotropic voxel size=1.3mm, flip angle=60o, NEX=4, receiver bandwidth=62.5kHz, with 3D geometric correction) within a single scan. Centroid positional deviations between MR images and CT reference were computed using a customized MatLab script to build a 3D displacement map for distortion quantification.

Results: Marker displacements of axial, coronal and sagittal acquisitions (mean±SD) were 2.46±2.66mm, 2.13±2.32mm and 2.48±3.18mm respectively. The minimum diameter-spherical-volume (DSV) that reported 1mm maximum/mean displacements on axial, coronal and sagittal images were 72mm/288mm, 72mm/244mm and 150mm/316mm respectively. Thus, sagittal acquisition provided the largest DSV that had a geometric accuracy comparable to a CT-simulator under the AAPM-TG66 requirement (distortion induced displacement<1mm).

Conclusion: We rigorously examined the acquisition orientation dependent geometric distortion of a wide-bore MR-simulator with a geometry accuracy phantom. The DSVs with 1mm maximum/mean displacements of the sagittal acquisition were notably larger than the others, implying the importance of considering acquisition orientation in clinical applications. Because of the relatively small DSV, MR geometric fidelity for large FOV applications has yet to be further improved. In vivo study is further needed to investigate subject-dependent distortion due to tissue susceptibility and chemical shift.



Contact Email: