Through my career, I have acquired extensive experience in neuroinformatics, with a focus in the application of deep learning in the domain of neuroimaging. As part of an ongoing project, we have developed a framework using 3-Dimensional Convolutional Neural Networks to classify structural malformations in neonatal MRI. I built and currently administer the research infrastructure in the Pediatric Imaging Research Center at Children’s Hospital of Pittsburgh. In a modern, multi-modal and multi-centered research environment, the efficient use of technology to facilitate and enhance collaboration and data throughput is indispensable. Our infrastructure includes automated de-identification of imaging, real-time MRS quantification, and secure, HIPAA compliant image transfer across multiple institutions. This has helped facilitate several on-going prospective and retrospective research projects involving dozens of investigators. Current efforts are geared towards streamlining the process of large-scale imaging dataset generation within our institution.

Additionally, a major area of my research is in quantitative imaging, primarily applied to pediatric brain tumors. This work particularly highlights the importance of cross-vendor and cross-institutional consistency and reproducibility in quantitative imaging. My Master’s thesis was in the development of open source software for the quantitative analysis of longitudinal diffusion in pediatric brain tumors. This work was specifically applied in the emerging field of immunotherapy, where it is imperative to distinguish tumor growth from the inflammatory response expected from the treatment using precise quantitative measures. Our work shows great promise in using Apparent Diffusion Coefficient (ADC) as a potential biomarker for detecting treatment response. Currently we are working towards using multi-modal imaging, including contrast enhancement and spectroscopy to further improve our sensitivity in this precariously difficult population. Our future work will be applying our refined methods to novel domains, including breast and liver cancer.