Administrative Supplement for Probabilistic Deep Learning Cervical Lymph-Node Auto-Segmentation For Imaging-enhanced Evaluation of Extracapsular Extension Risk (PDL-CLASIFIER) - PROJECT SUMMARY/ABSTRACT Facial data in medical imaging poses a unique challenge, especially in the treatment of head and neck cancer. While these images are essential for planning and delivering radiation treatments, they also include identifiable features that raise privacy concerns. Techniques to obscure or remove facial features—known as facial de-identification or defacing —have been proposed to reduce the risk of patient re-identification. However, these methods often interfere with critical anatomical details needed for precise cancer treatment, particularly in radiotherapy. Currently, there is no clear standard for using these techniques in head and neck cancer imaging, creating uncertainty about the best way to balance privacy with maintaining the usefulness of these images. This administrative supplement will evaluate how different facial de-identification methods affect the quality of medical images used in cancer treatment. We will study how these methods impact important steps in the radiotherapy process, such as identifying tumors and planning treatment. By comparing original images with de-identified ones, we aim to identify which techniques preserve the most critical information while ensuring patient privacy. This work will provide valuable insights into the trade-offs between privacy and the usefulness of medical data, helping researchers and clinicians make more informed decisions. In addition, the project will engage with patients, clinicians, researchers, and other experts to explore their views on sharing medical images containing facial features. Through structured interviews, we will investigate ethical concerns, practical challenges, and potential solutions for responsibly sharing these sensitive datasets. This stakeholder-driven approach will highlight diverse perspectives and provide an evidence base to guide future discussions on privacy and data-sharing practices. This collaborative effort between Baylor College of Medicine and The University of Texas MD Anderson Cancer Center directly aligns with the goals of the NIH and the NIDCR to address ethical and privacy concerns in sharing medical data. By combining technical research with stakeholder engagement, our work will enhance the ability to share high-quality imaging data responsibly. This will not only foster innovation in cancer treatment but also ensure public trust, ultimately benefiting patients and advancing research.
