Joey Schulz

I am a Medical Physics PhD candidate at the University of Wisconsin–Madison, working in Bryan Bednarz’s group on radiation therapy treatment planning, HDR brachytherapy, FLASH radiotherapy, clinical workflow automation, and translational research. My work focuses on developing practical systems that can move from research concepts into clinical use.

I have worked on patient-specific 3D-printed radiation therapy devices, including electron cutouts and photon blocks, automated treatment planning and QA workflows, virtual reality tools for patient education, and preclinical FLASH radiotherapy studies.

More recently, my research has focused on optimization methods for HDR brachytherapy, differentiable dose models, and machine-learning approaches that can make treatment planning faster, more flexible, and easier to validate. I am especially interested in problems where physics, software, biology, and clinical implementation meet. My goal is to develop radiation oncology tools that are accurate enough for clinical use, simple enough to deploy, and useful enough to change day-to-day practice.

Current research

• HDR brachytherapy optimization — integrated catheter position and dwell-time optimization for focal dose escalation in prostate HDR, and a robust formulation that folds catheter selection and dwell-time optimization into a single inverse-planning framework.
• Differentiable dose calculation — implicit neural representations (INRs) that replace discrete pre-computed dose kernels with continuous, differentiable models trained in PyTorch, accelerating gradient-based optimization through autodiff.
• FLASH radiotherapy — an open-source preclinical FLASH treatment planning system and Collimator Creator, plus the software-based conversion of a clinical Varian TrueBeam to FLASH mode.
• Machine learning for clinical workflows — LLM-assisted analysis of linac downtime, deep-learning autosegmentation for preclinical radiopharmaceutical therapy dosimetry, and foundation segmentation models for markerless patient positioning.

Background

Before starting my PhD I spent nearly three years as a Medical Physicist Assistant at Stanford Health Care, supporting clinical physics across SRS, SBRT, IMRT, VMAT, SGRT, and IGRT, and building Python and C# (ESAPI) tools for treatment planning and QA. Several of those projects reached clinical implementation, including a 3D-printed electron cutout program, non-toxic 3D-printed photon blocks, and the AVATAR audio-visual system for anesthesia avoidance, which was later licensed to Leo Cancer Care.

I hold a BS in Physics with a minor in Mathematics from Boston College.