Room: Exhibit Hall | Forum 7
Purpose: Retrospective cohort studies on therapy-related late effects involve longitudinal follow-up of cancer survivors across multiple institutions. These survivors (N>10,000 patients) were treated with historic/conventional radiotherapy (RT) techniques without CT-image based planning. Thus we use abstracted radiotherapy parameters from records and reconstruct treatments on in-house age-specific computational phantoms. Cardiac dose reconstructions are performed using a simple model developed using organ geometry from anatomy atlases. Purpose of this work was to develop a new detailed cardiac model with substructure definitions for age-specific computational phantoms used in late effects studies of childhood cancer survivors.
Methods: The atlas-based-heart model was refined and expanded using cardiac substructure contours from the computational human phantom series developed at the National Cancer Institute (NCI). Within a commercial treatment planning system, we registered the NCI phantom with our in-house phantom. Using a python script, we extracted whole heart and substructure contours and converted them to a grid of evenly-spaced points. Lastly, these points were scaled and translated to our in-house computational phantom. With this model, we can compute various average doses and dose-volume parameters, e.g., V5, V20 for large cohorts.
Results: We developed a new cardiac model with substructure definitions (aorta, ventricles, atriums and arteries) for our computational phantom. Importantly, the phantom (along with heart and substructures) can be scaled to age at RT for individual dose reconstructions, which is essential for pediatric cohorts. This new heart model has 20 times greater resolution than our previous model, which allows for excellent structural definitions, even for thinnest parts of cardiac arteries.
Conclusion: The detailed cardiac model developed here, enables dose estimation for individual cardiac substructures for pediatric cohorts whose historic RT was not planned with CT-images. Our model can be used for late cardiac toxicity studies that are based on cardiac substructure doses rather than whole-heart doses.
Not Applicable / None Entered.