Enhanced analysis of PET/CT imaging
We are developing methods for automatically quantifying and interpreting PET scans using deep learning. In particular, we have focused on how to measure quantitative PET imaging biomarkers for lymphoma in both adult and pediatric populations. We aim to improve patient outcomes by facilitating the measurement of imaging biomarkers that can better guide therapies.
Language modeling in radiology
Our lab explores the value of large language models (LLMs) in enhancing radiology workflows. We focus on creating new ways to improve report generation, improve education and training, and optimize clinical decision-making. We investigate how LLMs can understand and interpret medical language within the context of radiology.
Vision-language modeling
We are developing vision-language models (VLMs) and evaluating large multimodal models (LMMs) in accelerating and enhancing various tasks in radiology. In particular, we are interested in using medical text to guide medical image analysis and interpretation.
NucLex: nuclear medicine data and terminology standardization
Big data analytics requires that concepts and terminology are standardized. This is particularly important for biomedical text. Existing medical ontologies are inadequate with regards to nuclear medicine and nuclear oncology. NucLex is an effort to address this gap.
