Raw scans in, usable data out.

I trained as a biomedical engineer and spent two years as a neuroimaging research technician at Washington University School of Medicine before starting a Ph.D. there. Watching scan after scan degraded by head motion pulled me toward processing problems, above all correcting the distortion that motion leaves in echo-planar images.

My doctoral work produced MEDIC, a frame-wise distortion correction for multi-echo fMRI, released as the open-source warpkit library. Alongside it I maintain brainextractor and wbsurfer2 for neuroimaging, and jamb and difftrace for the engineering side: requirements traceability for regulated software, and change detection for monorepos.

Since then the work has been neuroimaging processing and machine learning: pipelines that take raw MRI to analysis-ready data, and deep-learning models for medical images, built to run reliably in clinical settings. The common thread is turning imaging research into software people can depend on.

The MRI physics demos on this site come from teaching rather than the day job. They are how I explain to newcomers where the images come from.