Vienna applicator calibration and TRE simulation

Purpose

Intracavitary brachytherapy is commonly used in the treatment of cervical cancer. Accurate geometrical information of the tumour, surrounding structures, and brachytherapy applicator is required for the definition of an optimal treatment plan. Magnetic resonance (MR) imaging provides an excellent definition of the anatomical structures, but the possible source path inside the applicator is not directly visible with this image modality. Hence the goal of this research is to develop an automatic algorithm for recovering the full pose (position and orientation) of a brachytherapy ring applicator in MR images, which can be directly used to digitize the source path inside the 3D MRI dataset.

Methods

The pose of the applicator is recovered by registering a 3D MR image of the applicator to the patient’s MR image.

Synthetic MR image of the applicator: The applicator’s material typically does not respond to MR excitation and it appears as a dark region in the images. Localization markers and holes (that are filled with body fluids when inserted) are typically appear as light linear structures in the images. This this appearance is reproduced by generating an MR image volume from the 3D geometrical model of the applicator.

Registration: We use 3D Slicer (www.slicer.org), to register the synthetic applicator volume and the patient’s MR volume.

Test framework: A test framework generates synthetic patient MR image volumes with known ground applicator pose. Synthetic patient MR image volumes are created by merging real patient MR images with a synthetic volume of the applicator, in a known pose. To make the resulting image more realistic, random noise is added to the applicator voxels.