At the Viswanath Lab, we develop the next generation of artificial intelligence (AI) approaches, specifically new image analytics, radiomics, pathomics, and machine/deep learning methods, to enable precision medicine by “unlocking” embedded information captured by different data modalities, in an intuitive, interpretable, and generalizable fashion.
The novel computational imaging methods being developed in our group are designed to capture biologically relevant and clinically intuitive measurements from a variety of data types, spanning radiological imaging, digital pathology, gene & protein expression, as well as spatial transcriptomics. Uniquely, we attempt to integrate information across multiple length scales of biomedical data by spatially resolving and cross-linking imaging (macro-scale) with molecular and pathology (micro-, nano- scales) data. Toward clinical translation, we not only ensure our AI models are reproducible across institution- or scanner-specific variations but also interrogate the inner workings of our AI tools to gain a deeper, interpretable understanding of what they capture and how they work. Our methods are being designed for oncological and non-oncological conditions, spanning both adult and pediatric populations.