Congratulations to Eugene Vaios, MD, MBA, who has been awarded a $50,000 yearlong Duke Cancer Institute Pilot Grant for the project "Leveraging Explainable AI to Develop a Non-invasive Radionecrosis Diagnostic Program for Patients with Brain Metastases." Co-PIs are Chunhao Wang, PhD, and neuroradiologist Evan Calabrese, MD, PhD.
Brain metastases from solid tumors are increasing in incidence and develop in more than 200,000 patients with cancer every year in the United States. Stereotactic radiosurgery is widely used for treatment, but approximately 20-30% of patients experience radiographic progression post-treatment, which can be difficult to distinguish from radionecrosis (RN). Accurate diagnosis of RN versus true tumor progression is critical, but biopsy is not always feasible. Artificial intelligence (AI), particularly deep learning, shows potential for distinguishing RN from recurrent tumors using routine MRI. This study aims to develop an Explainable AI platform using neural differential equations to integrate imaging, genomic and clinical data, providing insights into the diagnostic process and identifying genomic biomarkers associated with brain metastases RN after stereotactic radiosurgery.