A cardiac contouring atlas for radiotherapy

Area of Science:

• Radiotherapy
• Medical Imaging
• Cardiology

Background:

• Accurate contouring of cardiac substructures in radiotherapy CT-planning scans is challenging due to the heart's complex anatomy.
• Variability in contouring can impact radiation dose calculations and potentially affect clinical outcomes.

Purpose of the Study:

• To present a reproducible method for contouring left ventricular and coronary arterial segments on radiotherapy CT-planning scans.
• To establish a cardiac atlas to standardize segmentation for research and clinical applications.

Main Methods:

• Defined cardiac segments based on cardiology models, with consensus from two cardiologists.
• Developed a reference atlas with written guidelines and tested by six radiation oncologists.
• Assessed spatial variation using DICE similarity coefficient (DSC) and directed Hausdorff average distance (d→_H,avg).

Main Results:

• The atlas facilitated contouring of 15 cardiac segments with varying inter-observer agreement (DSC 0.60-0.73 for LV, 0.10-0.53 for coronary arteries).
• Inter-observer contour separation ranged from 1.5-2.2mm for LV segments and 1.3-5.1mm for coronary arteries.
• Spatial variation led to minimal dose variation (<1Gy) for most segments, but significant variation (1.2-21.8Gy) for segments near a field edge.

Conclusions:

• The developed cardiac atlas enables reproducible contouring of left ventricle and main coronary artery segments.
• This standardization is essential for future studies correlating cardiac radiation doses with clinical outcomes.
• Facilitates more accurate radiation therapy planning and outcome assessment in cardiac patients.