Is the data size the actual first bottleneck? What are good tools to process big EM data? Opensource?
Consider using machine learning (dictionary based) for segmentation. (talk also to Daniel Sage). Talk also to Willy Kuo, PhD student of Vartan Kurtcuoglu at Uni. Zurich
Collagen network quantification: counting? Thickness? everything?
Treat myofibrils (or their skeleton) as time-series to pull out summary statistics.