Click here to

Session: Therapy General ePoster Viewing [Return to Session]

Comparison of Manual, Isodose-Based, and Projection-Based PTV Contouring Methods for 3D Conformal Breast Cancer Treatment Planning

P Jensen1,2*, Z Chen1,2, (1) Yale-New Haven Health, New Haven, CT, (2) Yale University School of Medicine, New Haven, CT


PO-GePV-T-307 (Sunday, 7/10/2022)   [Eastern Time (GMT-4)]

ePoster Forums

Purpose: In current clinical practice, the planning target volume (PTV) of 3D conformal breast cancer treatment planning is contoured based on initial open fields placed by the physician. The PTV is created from these fields either by manually contouring them or by choosing an isodose line from the initial open field dose. Recently, a geometric projection algorithm was developed which contours the PTV based on the physical geometry of the initial open fields. This study compares the contours generated by these three PTV contouring methods.  

Methods: This algorithm's performance was tested on a dataset consisting of retrospective treatment plan data from 50 patients who had previously received breast radiotherapy at our institution. For each patient, the manual-, isodose-, and projection-based PTVs were contoured. The volumes, Dice similarity coefficients, and Jaccard indices of these structures were calculated.

Results: The averages and standard deviations for the manually-contoured PTVs, isodose-based PTVs, and projection-based PTVs were 1332±637 mL, 1246±652 mL, and 1367±627 mL, respectively. The Dice similarity coefficients comparing manual with isodose, manual with projection, and isodose with projection were 88.8±5.6%, 97.8±2.5%, and 88.9±5.6%, respectively. The Jaccard indices comparing manual with isodose, manual with projection, and isodose with projection were 80.0±9.4%, 95.5±5.3%, and 80.7±8.6%, respectively. A team of physicians has confirmed that the projection-based PTVs are more clinically appropriate than the manually-contoured PTVs and isodose-based PTVs.

Conclusion: Although the volume distributions seem similar, the Dice similarity coefficients and Jaccard indices indicate that these PTVs have significantly different spatial extents. Although the manually contoured PTVs should be identical to the projection-based PTVs by definition, imperfect human contouring can lead to large discrepancies which may lower cure rates by failing to target subclinical disease. Care should be taken to determine the PTV contouring method which provides the most meaningful metrics for target coverage evaluation.


Conformal Radiotherapy, Treatment Planning


TH- External Beam- Photons: treatment planning/virtual clinical studies

Contact Email