The Veeraraghavan’s lab is broadly interested in developing and translating new artificial intelligence (AI) methodologies for automated longitudinal cancer treatment response monitoring, early prediction of treatment outcomes for safe and more effective personalized and risk-adaptive treatment of cancers. We combine advanced machine learning including deep learning, medical image analysis, and statistical estimation methods to enable reliable and robust automated analysis of a wide spectrum of medical images including low contrast during treatment cone-beam computed tomography (CT)s, diagnostic CTs, magnetic resonance imaging (MRI), ultrasound, and endoscopy images. We are focused on developing assistive technologies that increase the accuracy, efficiency and ultimately effectiveness of cancer treatments. This encompasses methodologies to automate quantification of tumor and tissue changes during treatment and at follow-up from medical images, new and robust radiomics biomarkers (e.g., tumor hypoxia, local inflammation due to radiotoxicity to tissues, and intra-/inter-tumor site imaging heterogeneity to early predict treatment effect as well as detect subsequent development of treatment complications for personalized treatment interventions.
Dr. Veeraraghavan’s group has advanced new AI methodologies such as cross-modality distillation learning for reliable segmentation of tumor volumes from low soft-tissue contrast cone-beam CTs acquired during radiation treatment as well as quantitative dose accumulation using simultaneous registration segmentation methods for cone-beam CT and new MRI-guided radiation treatments for tumors and tissues undergoing large deformations.
Dr. Veeraraghavan’s translational focus lies on developing and deploying normal organs and tumor auto-segmentation methods for automating radiotherapy treatment planning as well as longitudinal deformable radiation dose accumulation methods for objective estimation of delivered treatment doses for precise and personalized adaptation of radiation treatments. The Veeraraghavan Lab has released three AI methods for auto-segmentation of normal tissues for prostate, lung, and head and neck cancers deployed for routine clinical treatment planning at Memorial Sloan Kettering Cancer Center (MSK) since 2019, with a fourth one planned to be released for MRI-guided radiation treatment of pancreatic cancers.