A Comparison of MRI Distortion Between Scanners and Sequences for Radiotherapy Purposes
A Walker1,2*, L Holloway1,2,3,4,5, G Liney3, P Metcalfe1, (1) Centre for Medical Radiation Physics, University of Wollongong, Wollongong, (2) Liverpool and Macarthur Cancer Therapy Centres, Liverpool, NSW, (3) Ingham Institute, Liverpool, NSW, (4)School of Physics, University of Sydney, NSW, (5)South Western Sydney Clinical School, University of New South Wales, NSWTH-C-141-9 Thursday 10:30AM - 12:30PM Room: 141
Purpose: To quantify the magnitude and distribution of system related geometric distortion present in MR images and determine the clinical radiotherapy sites for which this distortion is likely to have most impact.
Methods:A commercially developed phantom conforming to AAPM standards for testing uniformity and linearity was imaged on 4 different MRI scanners. Similar T1 and T2 weighted clinical protocols were assessed. Each sequence and scanner had varying protocol parameters. Additionally, the direct impact of altering the receiver bandwidth (7.69 - 76.8kHz) on distortion and signal to noise was conducted on a standard spin echo sequence. The position of each phantom grid point was determined using in-house MATLAB code, then compared to the known position as verified by a CT scan of the phantom to obtain the deformation matrix. Positions of anatomical sites (breast, lung, cervix, prostate and H&N) were determined by analysing radiotherapy contoured data for 10 patients per site to establish the range of distortions occurring within these regions.
Results:Distortions were observed to range from 0mm - 8.9mm in magnitude. Maximum distortion values measured across one of the similar clinical spin echo sequences tested varied from 2.75mm - 8.94mm. Similarly, comparison of a gradient echo sequence showed maximum distortions varied from 2.56mm - 3.33mm. Increasing the bandwidth reduced the maximum distortion present in the image by 3.49mm, at the cost of SNR. While distortion magnitudes and distributions varied, in most sequences the distortions began exceeding 2mm beyond a radial distance of 100mm from isocentre, where the oral cavity, lungs and breast lie. For regions where the prostate, cervix and brain stem lie, distortions were below 2mm.
Conclusion:This work displays the variations in distortions present in MR images for different sequences and scanners. Distortions are of greatest magnitude in areas where the lung and breast would be located for radiotherapy purposes.