Development of in-house software to process real-time cine magnetic resonance images acquired during 1.5 T MR-guided radiation therapy

Abstract

This study aimed to develop and publicly release an in-house software package that converts the binary file format of 2D cine magnetic resonance (MR) images acquired through the Treatment Session Manager (TSM) of MOSAIQ on the Elekta Unity (Elekta AB, Stockholm, Sweden) into standard readable data formats. The software was developed using MATLAB (MathWorks, Natick, MA, USA) and includes an automatic image-sorting algorithm to classify the images into coronal, sagittal, and axial planes. To verify the geometric accuracy of the converted images, they were acquired from an MRgRT motion management QA phantom, both with and without motion. For the converted images without motion, the geometric size of the phantom was measured and compared with the known values. For the images acquired with motion, the magnitudes measured using the converted cine MR images were compared with the artificially introduced motions. The results showed that the 2D cine MR images in the binary file format were successfully converted into the metadata/DICOM format and accurately classified into different planes. The accuracies of the geometric size and motion in the converted 2D cine MR images were greater than 99.6% and 94.2%, respectively. This software is expected to be useful for 1.5 T MR-linac users in analyzing internal organ motion during radiation treatment.

Keywords: 2D cine MR image; MATLAB; MR guided radiotherapy; MR-linac.

Fig. 1

Overall workflow of converting cine MR images from binary file format into metadata and DICOM file formats.

Fig. 2

Illustration of plane-specific intensity patterns in the upper right corner of the cine MR image, used to automatically identify the imaging plane of each converted image.

Fig. 3

ZEUS MRgRT motion management QA phantom with the dimensions-of-interest presented.

Fig. 4

Experimental setup for two simulated motions: (a) respiratory motion of a moving target in the lung, (b) translational motion of the thorax region, including vertebral bodies, by shifting the phantom.

Fig. 5

Converted 2D cine MR images acquired during MR-guided radiation treatment of a prostate cancer patient displayed in MIM software.

Fig. 6

(a) Daily MR images included PTV structure (green line), (b) the converted 2D cine MR images, and (c) registration of two images in the sagittal planes.

Fig. 7

Manual image registration results of a (a) converted 2D cine MR image and (b) 3D T1 image, and (c) comparison between the two images as a checkerboard image using 3D Slicer.