I started this project in 2006 with the anticipation of observing the onset of a migraine while a person is in an MRI scanner. I eventually got IRB approval from Hartford Hospital to do the study and a departmental award from Vince Calhoun to pay the subjects. My initial recruitment goal was 15 subjects (6 migraine w/aura, 6 migraine w/out aura, 3 healthy controls). As of Fall 2007 I had recruited 12, scanned 7 (including test subjects to tweak protocol), and successfully induced and observed a migraine in one patient. I gave a talk at the Olin Neuropsychiatry Research Center on September 25, 2007 about the study and a summary of current migraine pathogenesis understanding. The talk is available in PDF.

The most interesting finding for the single subject who got a migraine was a negative activation in the trigeminal ganglion during the visual stimulation from the flashing checkerboards. The subject who did not get a migraine had no activation there. The MR angiogram produced no signoficant changes in blood vessel diameter.

I started writing a DICOM viewer in 2004 and expanded it to read raster, analyze, and nifti images. I also experimented with volume rendering, MIP views, and displaying 4D data. The viewer's webpage is at gbooksoft.com where it can be downloaded.

The hyperscanning project is part of a Eureka grant application to study player interaction during a game of Dominos. My piece is to create a multiplayer game of Dominos. This was an interesting experience as I hadn't created a networked program before.

The Olin center runs 100 MRI subjects on average per month and before 2005 we had no way of accessing older data without going to our backup DVD source. I began a system to store all our MRI data online and it gained the name All Data Online (adoserver). By 2007, the system had outgrown the 4 servers I had built for it, and we moved it to two newer, larger servers. I also rewrote the interface, storage system, unified the search process, and used a MySQL database to catalog all the data. Searching and retrieving became much easier. Currently the servers hold nearly 85,000 MRI datasets or about 6.4TB unzipped.

I created a system for analyzing pupillary dilation in a similar way to the way event related potentials (ERPs) are processed. I did a group analysis using the Fusion ICA Toolbox developed by Vince Calhoun with 19 subjects who completed the auditory oddball task while simultaneously collecting pupil diameter information. The analysis was submitted to the CNS 2008 conference.

Starting back in 2003 I was writing some C++ programs and I was looking for a good language reference on a single card. I couldn't find one, so I decied to create one, and I started selling them. I expanded the offering of cards to include chemistry, PHP, and organic chemistry. I've sold 350 cards so far. The website for the reference cards is gbookcards.com

While at Travlers, I've had the opportunity to solve a geolocation problem and in the process was able to generate some interesting street maps. I had first set out to create an application that would locate the closest vendors to a claimant. A simple radius search that has been done many times before by various companies. After finding a source of free geographic data from the Census Bureau, I found I could make street maps with the data as well. Some of the material I created may be copyrighted by Travelers, so I've limited this section to just results of the map project.
   I came across the Census Bureau's data while I was looking for a source of data from which to create driving directions. So I decided creating maps would be the best way to learn how I could create driving directions. I quickly found that it was not a good source of data for driving directions because it does not contain three important things: exit numbers, one way street indicator, street intersection data. So I stuck with making maps, hopefuly for the purpose of showing a user where the vendors are in relation to their address. The maps shown here do not have street labels, which keeps them from becoming cluttered and unreadable. I am currently working on methods of improving the map drawing to allow for select street names when many streets are present. When street labels are added, they follow the angle of the street segment.
View Maps in Connecticut:

06183 (Hartford, CT) 1.57MB (2600x1950)
06269 (Storrs, CT) 820KB (2600x1950)
06615 (Stratford, CT) 770KB (2600x1950)
06770 (Naugatuck, CT) 511KB (2600x1950)

I experimented with eye gaze tracking for my senior design project at UConn.

I apply the fisheye view to a radial tree graph of websites.

I created a program to view vertex/edge data using a fisheye view. We tried it against a directory tree and a subway map.