Currently, I’m working on applying probabilistic learning techniques to improve automatic tuning and boundary detection of lutetium oxyorthosilicate (LSO) scintillation crystal arrays in the High Resolution Research Tomograph (HRRT) brain PET scanner. Manufacturer supplied tuning software performs poorly in situations where there is low signal-to-noise characteristics or when there is significant “cross-talk” from neighboring scintillators in the array.
Other Interests
I am interested in applying AI and machine learning techniques to medical imaging data, modeling brain circuitry (both nuclei and interconnections between nuclei) abnormalities in a variety of neurological disorders. Further interests include biomedical informatic methods for enabling large-scale distributed data federations. For the past 6 years I’ve been working on tools to enable the documentation and federation of distributed fMRI data across multi-site collaboratories.
•High Resolution PET Listmode Motion Correction Using 3D Motion Data. Lee K.; Hong I; Potkin S.; Burbar Z.;Keator, D.. Abstract and poster presentation at Joint Molecular Imaging Conference, 2007.
•A General and Extensible Multi-Site Database and XML based Informatics System for the Storage, Retrieval, Transport, and Maintenance of Human Brain Imaging and Clinical Data.Keator D; Ozyurt BI; Wei D; Gadde S; Potkin SG; Brown G; MBIRN; FBIRN; Grethe J. Abstract and poster (253 TH-AM) presentation at Organization of Human Brain Mapping, 2006.