Main Menu (Mobile)- Block

Main Menu - Block

Data Acquisition

janelia7_blocks-janelia7_fake_breadcrumb | block
Data Acquisition
node_title | node_title
Data Acquisition
node_body | node_body

 

There are a variety of electron microscopy approaches for acquiring 3D image volumes suitable for reconstructing neural circuitry.  Each faces trade-offs between throughput and resolution detail, and all data faces a massive challenge of reconstruction where effort can be measured in hundreds of man-years for even modest volumes. Guided by the fly EM’s vision to speed the processing by maximizing automation in the reconstruction, we choose an EM imaging approach that offers the best 3D resolution that is most compatible with automation, while sacrificing some throughput. Specifically an approach called focused ion beam scanning electron microscopy, FIB-SEM, can easily achieve 8-10 nm resolution not only in the XY imaging plane but more importantly FIB-SEM retains this resolution in the vertical Z direction, the third dimension.  Used widely in the semiconductor industry, FIB-SEM images the surface of a block of prepared neural tissue, then ablates ~ 2-10 nm from the surface, followed by further cycles of imaging the slightly deeper surface and further ablation. By repeating this cycle tens to hundreds of thousands of times, the volumetric data set is acquired with sizes of ~100 microns. To acquire such large volumes required special modification of the FIB-SEM so that we can achieve the needed stable long-term operation of several months.

 

Multiple such volumes have been generated, focusing on different neuropil of the fly brain.   Several data sets of the optical lobe targeted the medulla initially and later to encompass the medulla, lobula, and lobula plate.  This latter data set seen in the video spans 160 microns with 8x8x8 nm3 voxels and was acquired over a 3 months period.  More recently several antenna lobes and volumes containing the alpha and alpha prime mushroom body have been acquired and are being reconstructed.

 

FIB-SEM does have a 100 micron limit on milling depth, which prevents one from acquiring larger neuropils in one imaging session.  This barrier has been overcome by portioning larger samples into 20 micron thick tabs and the quality of the hot knife partition is such that less than an effective 40 nm of material is lost and processes can be traced across this “hot knife” partition. Larger brain volumes are now being partitioned into tabs suitable for FIB-SEM and will be employed for the current goal of acquiring the entire central complex as well as a future goal of acquiring the whole fly brain.