The Radiation Oncology Institute is thrilled to announce this year’s new research award winners. Six talented teams will be investigating how to leverage artificial intelligence (AI) for radiation oncology. AI-based tools can synthesize numerous variables and detect patterns in data faster than humans and can be a powerful complement to care provided by clinicians and help improve outcomes for patients with cancer. The following teams were selected to receive grants from ROI through a comprehensive and competitive peer review process.
Chaitra Badve, MD, Serah Choi, MD, PhD, and Yong Chen, PhD, of the University Hospitals Cleveland Medical Center and Case Western Reserve University will use Magnetic Resonance Fingerprinting (MRF) and artificial intelligence-based neural networks to improve target delineation in post-operative radiotherapy for glioblastomas. Radiation helps control local recurrence but carries the risk of neurological deterioration, especially as more of the brain is treated. Clearly distinguishing areas of the brain where the tumor infiltrated beyond the part that could be removed through surgery is challenging with current imaging methods. MRF is a quantitative imaging tool developed by their group that allows for accurate and reproducible high-resolution measurement of multiple tissue properties in a single five-minute scan. By combining MRF with AI, Dr. Badve, Dr. Choi and Dr. Chen aim to map tumor infiltration more accurately for each patient to create personalized radiation treatment strategies that better target tumor cells and reduce neurological deterioration.
Ming Chao, PhD, of the Icahn School of Medicine at Mount Sinai and Jose Penagaricano, MD, of the H. Lee Moffitt Cancer Center and Research Institute will be Co-Principal Investigators on a study to test their unique spatial cluster model to predict xerostomia in head and neck cancer. Their model is based on percolation theory, one of the simplest probability models that predicts cluster formation in random networks. Dr. Chao and Dr. Penagaricano showed that incorporating spatial dose distribution into the cluster model improves the prediction of xerostomia over existing methods. To optimize the accuracy and sensitivity of the cluster model, big data analytics will be utilized. Their goal is to develop a better decision-making tool for treatment planning in head and neck cancer that can help improve patient quality of life by reducing or preventing xerostomia. The model could be translated to other organ systems for toxicity prediction as well.
Tafadzwa Chaunzwa, MD, MHS, and his team at Dana-Farber Brigham Cancer Center will use radiomics to develop imaging-based biomarkers that identify patients with advanced non-small cell lung cancer who are unlikely to respond to immunotherapy and will need intensified treatment with stereotactic ablative radiotherapy (SBRT) or chemotherapy. Together with mentors Hugo Aerts, PhD, and Raymond Mak, MD, Dr. Chaunzwa will analyze computed tomography (CT) scans of the chest routinely acquired before and during treatment to identify radiomic signatures that help predict the response of lung cancer to immune-checkpoint inhibitors. They will then combine this with clinical and demographic risk factors to determine which patients are likely to benefit from additional treatments, including radiotherapy. They aim to develop an advanced but inexpensive AI software that can improve outcomes for patients with lung cancer. Dr. Chaunzwa is this year’s James D. Cox Research Award winner, a special recognition for a resident pursuing a career in research that is generously supported by Dr. Ritsuko Komaki-Cox.
Allen Mo, MD, PhD,and his team at Montefiore Medical Center/Albert Einstein College of Medicine will use machine learning (ML) algorithms to aid clinical decision making for optimal treatment selection for patients with hepatocellular carcinoma (HCC). Building on prior work analyzing clinical, demographic, pathologic and radiographic features in their database of over 900 patients with HCC, Dr. Mo and mentor Rafi Kabarriti, MD, will develop decision support tools to evaluate progression free survival and risk of toxicity for patients in both the newly diagnosed and previously treated settings. First, they will develop ML models for progression free survival and liver toxicity using their curated retrospective database. This will be followed by a two-phase, prospective observational trial comparing ML decisions to clinical treatment decisions made in liver tumor board meetings at their institution. Through this work, they strive to reduce variability in clinical decision making, personalize patient care and improve outcomes.
Jinzhong Yang, PhD, Percy Lee, MD, and their team at the University of Texas MD Anderson Cancer Center and City of Hope are using MR-guided radiotherapy and customized AI tools to improve the management of cardiac toxicity in patients with non-small cell lung cancer. They will leverage the unique capabilities of MRI guided linear accelerators (MR-Linac) that allow for improved and real-time visualization of cardiac substructures during treatment and may reveal subtle changes to the heart due to radiation that might impact long-term toxicity. Using a deep learning segmentation model they have developed, Dr. Yang and Dr. Lee will optimize MRI sequences for online treatment planning and cardiac substructure auto-segmentation to create dosimetric constraints that minimize the dose to the substructures of the heart. They are also building a patch-based deep learning model to measure dose accumulation in the heart using hybrid CT/MR deformable registration. These new tools could enable real-time treatment plan adaptations that minimize the risk of cardiac toxicity during radiation treatment for lung cancer.
Simeng Zhu, MD, and his team at the Henry Ford Health System are developing an AI-based clinical decision support tool to guide the potential personalization of radiation treatment for patients with oropharyngeal cancer (OPC). Building on prior work, Dr. Zhu and mentors Farzan Siddiqui, MD, PhD, Indrin Chetty, PhD, and Hassan Bagher Ebadian, PhD, will expand an existing curated dataset of over 1,000 patients with OPC treated at multiple institutions. From this, they will build and test an innovative model that incorporates tumor imaging features in addition to clinical factors such as HPV status and tumor stage to predict a patient’s risk of local recurrence or distant metastasis. Treatment can be personalized by identifying which patients have a higher risk of recurrence and need heightened surveillance after treatment and those who may benefit from treatment de-escalation, which can reduce long-term radiation-induced toxicity.
ROI is funding these innovative research teams who are pushing the boundaries of radiation oncology with the support of our generous donors. With these six Leveraging Artificial Intelligence for Radiation Oncology awards, ROI is surpassing the $4 million mark in funding for research and educational programs. You are making a difference for radiation oncology researchers who are working to improve outcomes for patients, and we are grateful for your ongoing support.