Answers are grounded in this paper and cite their sources.
Feasibility study demonstrating semi-automated 3D-printable breast prostheses, indicating enhanced personalization for patients post-mastectomy.
BACKGROUND: Breast reconstruction after mastectomy is limited by geographic disparities, surgical complexity, socioeconomic factors, and incomplete restoration of native anatomy. Customized prostheses offer a non-invasive alternative that could help mitigate some of the barriers to patients undergoing mastectomy. Objective: To develop Restora, a semi-automated software that converts clinical imaging into 3D-printable, patient-specific breast prostheses at the point of care. Methods: Twelve breast MRI datasets were segmented manually in 3D Slicer and semi-automated in Restora. Segmentation accuracy was evaluated using Dice similarity coefficients, volumetric comparisons, and surface distance metrics. Prostheses were 3D-printed using biocompatible polylactic acid (PLA), thermoplastic polyurethane (TPU), and resin to assess feasibility, print time, and material suitability. Results: Restora reduced manual intervention to a single modification per case versus 5-6 for manual segmentation. The Mean Dice coefficient was 0.82, with a mean surface deviation of 4.3 mm and volume discrepancies ranging from 5.9% to 54.7%. Prostheses were successfully printed across materials within 9-17 hours, demonstrating feasibility for clinical use. Limitations included flat posterior designs and localized geometric deviations, particularly in larger breasts. Conclusion: Restora enables rapid, semi-automated generation of anatomically consistent, 3D-printable breast prostheses, reducing labor, while maintaining acceptable geometric fidelity. Future work will integrate upright imaging, improved segmentation algorithms, and post-mastectomy chest wall modeling to enhance comfort, fit, and aesthetic outcomes, supporting personalized prosthetic care for diverse patient populations. .
No takes yet. Share an insight, caveat, or question.
Kittani et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: