-
Notifications
You must be signed in to change notification settings - Fork 9
OMERO Hadoop Distributed Processing
Image feature calculation with WND-CHARM (https://github.com/wnd-charm/wnd-charm) is one of the use cases. The eventual goal is for a user to trigger feature calculation for multiple images, and for this to be done on a Hadoop cluster instead of on a Script Processor service as at present. Hadoop effectively acts as a scheduler and queue manager which avoids a lot of the complexity involved in working with traditional HPC clusters, however this still requires significant work in OMERO.server to handle the potentially long delay between submission of a job and receiving the results.
The immediately short term goal is to be able to trigger feature calculation using WND-CHARM for multiple images from an OMERO.script.
SPIM is a relatively new imaging modality which involves imaging a whole plane of a specimen using a light sheet. The sample is rotated to obtain views from multiple angles, and a reconstruction algorithm is required to build the final 3D image/model. This algorithm requires computationally intensive numerical calculations, and is therefore an ideal candidate for offloading onto a HPC cluster.