Incorporating sensitive cardiac substructure sparing into radiation therapy planning

Area of Science:

• Medical Physics
• Radiation Oncology
• Cardiovascular Imaging

Background:

• Cardiac substructures are radiosensitive but difficult to visualize on standard treatment planning CTs (TPCTs).
• Current treatment planning often overlooks cardiac substructure sparing due to visualization limitations.
• Integrating MRI data can improve soft tissue contrast for better delineation of cardiac substructures.

Purpose of the Study:

• To leverage low-field MRI on an MR-linac for enhanced cardiac substructure visualization.
• To integrate MRI-derived soft tissue contrast into TPCTs via image registration.
• To enable and evaluate improved cardiac substructure sparing in radiation treatment planning.

Main Methods:

• Retrospective evaluation of 16 upper thoracic patients treated on a 0.35T MR-linac at various breathing states.
• Utilized a hybrid MR/CT atlas and deep learning (3D U-Net) to propagate 13 cardiac substructures to TPCTs.
• Radiation oncologists revised contours using registered MRIs; plans were re-optimized for dose reduction and evaluated dosimetrically.

Main Results:

• Cardiac sparing plans reduced mean heart dose (mean reduction 0.7 Gy) and left anterior descending artery (LADA) doses.
• Significant reductions observed in left ventricular (LV) maximum dose and LV volume receiving 5Gy (LV-V5) for several patients.
• No statistical difference in treatment time, monitor units (MU), or clinical endpoints to organs at risk (OARs) like lung or esophagus.

Conclusions:

• Integrating 0.35T MRIs with MR-linac technology allows for effective cardiac substructure segmentation and sparing in CT-based planning.
• This approach offers potential for more precise radiation delivery with limited increase in planning complexity.
• Further validation in larger cohorts may help reduce radiation-related cardiotoxicities.