Mar 15 06 (project meeting)

present: Tolga, Pavel, Liz, Bryan, Chi-Bin, Melissa, and Kristen Kwan from Chi-Bin's lab

Tolga and Pavel showed progress with feature-based registration and raised the possibility of doing feature-based mosaicking. The existing microneuroma data seems to have different tiles missing from different slices. It would be useful to have a test data set with more slices; it may be possible to start playing around with tracking. There was discussion of when the final datasets would arrive; this will have to wait for the new EM which will not be installed until the Moran II building opens.

Liz showed her current success with snakes-based tracking. Amazingly, she is able to track ~50% of the axons that Melissa hand-tracked, through about 100 slices of SBFSEM data. This is using the UINTA-cleaned dataset. There are some tweaks to try to improve the algorithm, including preventing the snakes from inverting their orientation and crossing themselves. It may also be useful to include information about the included area, and implement a criterion like "no membranes inside the snake". Liz has submitted a paper to MICCAI.

Chi-Bin suggested that it may be worth looking at the online supplementary data from the Horstmann and Denk paper, which may have higher-quality data in which some axons could be tracked.

Melissa is scheduled to go to Heidelberg in mid-May, and is hoping to acquire higher-quality SBFSEM images, as well as more biologically-relevant images (i.e. encompassing the lower portion of the optic tract).

Kristen showed some of the 4D timelapse confocal data that she has been taking of the developing zebrafish eye, labeling all nuclei in red and all cell membranes in green. There are problems of segmentation at individual timepoints, tracking through time, and visualization of resulting analyzed data. Tolga thought the problems looked interesting. There do not seem to be existing SCI tools that could be used to analyze this data. SCIrun was mentioned with derision. Kristen will first look at existing commercial visualization software. Perhaps Ross can take a look sometime.

There was little enthusiasm for writing meeting notes on this Wiki. Thus this slapdash account from Chi-Bin.