Click here to

Session: Multi-Disciplinary General ePoster Viewing [Return to Session]

Impact of Image Distortion On Target Tracking in 0.35T MRgRT

H Zhang1*, S Marasini1, T Kim1, (1) Washington University in St Louis, St. Louis, MO


PO-GePV-M-243 (Sunday, 7/10/2022)   [Eastern Time (GMT-4)]

ePoster Forums

Purpose: MR-guided tumor tracking is advantageous to radiotherapy because of real-time cine imaging with superior soft-tissue contrast. This study is to investigate the impact of cine image distortion on the target tracking in 0.35 T MRgRT.

Methods: A cuboid in QUASAR MRI4D Motion Phantom was tracked at two positons: isocenter (setup 1) and off-center (setup 2). At each setup, cine images on the sagittal plane with phantom at static positions and gantry=3000 were acquired as the reference. The motion of the phantom was set to sinusoidal mode (amplitude=10 mm, 15 BPM) to mimic the human breathing motion. The target was tracked with an imaging rate=8 FPS. Tracking contour was turned on and off for one minute each, and cine images were screen-recorded at 30 fps. This procedure was repeated with the linac gantry at 0 ~ 3300 (300 interval).

Results: The screen-recorded cine images were processed with openCV2 package. For the cine images without tracking contours, Otsu’s auto-threshold method was used to contour the target. The average area of the auto-threshold contours with target position 10 mm, 0 mm, and -10 mm were analyzed and utilized as an index for image distortion. The average area of the tracking contours was used as an index for the impact of image distortion on target tracking performance. The result shows that cine images can be distorted by up to 4.5% with various gantry angles. The distortion will change the tracking contours by up to 25%, and target in motion was more severely distorted than that in static status.

Conclusion: Using the contour area as the index for image distortion in 0.35 T MRgRT, this study shows that the cine images can be distorted by up to 4.5% with different gantry position. The image distortion caused the tracking contour distortion by up to 25%.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email