PhD student in Medical Radiation Physics: AI-Driven Precision Radiotherapy for Spinal Metastases

Project description

Spinal metastases represent the most common site of skeletal dissemination in advanced cancer, affecting an estimated 30% to 70% of oncology patients most frequently arising from primary lung, breast, or prostate tumors. These lesions carry a heavy clinical burden, often causing intractable pain, pathological fractures, and epidural spinal cord compression that result in permanent neurological deficits and severely degraded quality of life. Clinical management is typically multimodal, with decision-making guided by frameworks like NOMS and SINS to evaluate instability and collapse risk. Within this paradigm, Stereotactic Body Radiation Therapy (SBRT) has become a cornerstone modality. Delivering highly conformal, ablative radiation with sub-millimeter precision and steep dose gradients allows SBRT to achieve superior local control, direct tumor ablation, and fast pain relief while sparing adjacent, radiation-sensitive neural tissue.

Despite its therapeutic success, SBRT carries notable clinical risks, particularly radiation-induced vertebral compression fractures (VCF) in 10% to 15% of treated segments. Exceeding strict tissue tolerances can also cause irreversible complications like radiation myelopathy or radiculopathy. Developing robust predictive models for these adverse events is critical for patient stratification and proactive surgical planning. Accurate risk assessment empowers clinical teams to personalize dose-fractionation regimens, ensuring that the benefits of local tumor control are not compromised by secondary structural collapse or neurological morbidity.

This doctoral research focuses on establishing an integrated computational ecosystem to enhance precision management and personalized care across the spinal oncology pipeline. Specifically, the project develops advanced deep learning methodologies for automated and interactive target volume delineation, paired with multimodal architectures designed to quantify longitudinal post-SBRT lesion dynamics. In parallel, it implements robust prognostic models for clinical outcome and fracture risk prediction, while establishing a standardized, radiobiologically driven framework for voxel-based cumulative dose summation and toxicity constraint optimization in spinal reirradiation.

Ref. No. SU FV-2757-26

Closing date: 28 September 2026

Complete information here