These data tools provide a means for data discovery, search, and access with capabilities for geolocation (identifying location based on user location via a data collection mechanism), reprojection of datasets (from one coordinate system to another), and mapping of datasets on a grid.
Grid Cartographer 4 key serial
Geohash is a public domain geocode system invented in 2008 by Gustavo Niemeyer[1] which encodes a geographic location into a short string of letters and digits. Similar ideas were introduced by G.M. Morton in 1966.[2] It is a hierarchical spatial data structure which subdivides space into buckets of grid shape, which is one of the many applications of what is known as a Z-order curve, and generally space-filling curves.
For exact latitude and longitude translations Geohash is a spatial index of base 4, because it transforms the continuous latitude and longitude space coordinates into a hierarchical discrete grid, using a recurrent four-partition of the space. To be a compact code it uses base 32 and represents its values by the following alphabet, that is the "standard textual representation".
The most important property of Geohash for humans is that it preserves spatial hierarchy in the code prefixes. For example, in the "1 Geohash digit grid" illustration of 32 rectangles, above, the spatial region of the code e (rectangle of greyish blue circle at position 4,3) is preserved with prefix e in the "2 digit grid" of 1024 rectangles (scale showing em and greyish green to blue circles at grid).
Renewed interest in ssEM as a high-resolution 3D tool for neuroscience has led to improvements over the last decade in this otherwise time-, skill-, and labor-intensive approach [34], [35]. Recent studies [16], [18] have benefitted from newly developed methods based on an SEM platform using backscatter imaging from a tissue block surface that is successively removed by the diamond knife (serial block-face SEM, or SBFSEM; [36]) or a focused ion beam (FIB-SEM; [37], [38]). Unfortunately, these approaches may not yield the level of lateral resolution or contrast necessary for unequivocal identification of the nanoscale subcellular structures as discussed above. Furthermore, these approaches are destructive, so that sections cannot be retrieved for subsequent viewing at higher resolution.
The serial ultrathin sections were imaged with either a JEOL JEM-1400 TEM (Tokyo, Japan) or a Zeiss SUPRA 40 field-emission (FE) SEM (Oberkochen, Germany). The TEM is equipped with a charge coupled device (CCD) camera with the field size of 4,0804,080 (or 16.65106) pixels (Gatan UltraScan 4000; Pleasanton, CA), controlled by DigitalMicrograph software (Gatan). For TEM, the slot grids containing serial ultrathin sections were loaded into grid cassettes that were individually loaded into a Gatan 650 CC specimen holder that allow the grid to be rotated inside the chamber. The holder accommodates one grid at a time, and requires manual exchange between grids. At 6,000magnification at 2 nm pixel size with the accelerating voltage of 120 kV, serial section images were manually acquired as the Gatan proprietary.dm3 files, which were later batch converted into 8-bit JPEG files with DigitalMicrograph software. Conversion into JPEG was originally done to save space in our database. No practical differences in identification of key structures were found compared to the same.dm3 images converted into TIFF format.
If the size of each image field needs to be extended beyond 32,76832,768 pixels, the operator can set up mosaics by specifying the target dimensions of the image field and the amount of overlap between image tiles. ATLAS then automatically determines the number of image tiles per field, based on the pixel size and the size of each image tile. For example, an image field of 360 µm wide60 µm tall can be set up as a 61 mosaic of image tiles measuring 32,76832,768 pixels each at 2 nm pixel size (Fig. 1E). The operator is required to mark only the center of mosaic field (* in Fig. 1E), versus (# in Fig. 1E) in the single-frame images. The ATLAS system can also be used to acquire large frame images with secondary detectors (on our system; Fig. 1A) or backscatter detectors (not on our system). On occasions where we needed lower magnification views of the overall grid or section (e.g., Fig. 1E), the specimens were imaged with secondary electron detectors mounted on the side of the chamber or inside the final lens (see Fig. 1A).
We had hoped to use a carbon replica grating to compare high-order distortions in the TEM and tSEM; however, the tSEM scan field is much larger than the individual grid support window, within which the grating replica also revealed large scale wrinkles (Fig. 2A). Instead, we imaged an unused integrated circuit (IC) chip, which has a regular pattern and is etched onto a very stiff substrate and is therefore very flat. The substrate is electron opaque, which makes the IC ineligible for use in calibrating the TEM. In the tSEM, we may image it under the same conditions that we might use for transmission imaging, with the exception that we use a secondary electron detector (Fig. 2B). We wrote software in Matlab to measure the high-order geometric distortion due to imaging with an electron microscope (available from: ). Unfortunately, we have no prior knowledge to guide our expectation of how the sample should appear. Visual inspection tells us that the pattern consists of squares that are regular over a parallelogram with an inner angle of approximately 60 (Fig. 2C). To estimate the imaging distortion, we found the imaged locations of the units of this pattern and compared them to their expected locations. A match kernel representing a single unit was selected manually from the image and used for normalized cross-correlation with the original image, resulting in a map in which local maxima represent the precise image locations of these repeated units (Fig. 2D). To extract the maxima, we perform a simple threshold (Fig. 2E). Thresholding is advantageous in that it is simple to implement, but may occasionally result in spurious detections. We correct for these and extract our model for the expected pattern at the same time using RANSAC (RANdom SAmple Consensus) [50].
Image distortions can affect calibration of pixel size and section thickness, which are critical steps in quantitative 3D analysis of tissue volumes. Pixel size was calibrated based on a grating replica image (Fig. 2A inset) that was imaged along with the serial section series. Section thickness is estimated with the cylindrical mitochondria method, which uses the ratio of the maximum diameter of longitudinally sectioned mitochondria (or other cylindrical objects) to the number of serial sections they span [51]. For our typical series acquired on either tSEM or TEM, the voxel size obtained through these methods is about 2 nm2 nm45 nm (xyz). This calibration is applied to the entire tissue volume for three-dimensional quantitative analysis of reconstructed neuropil and synapse structures (e.g., counting, lengths, area, volume, z-distances) that are sampled based on well-defined sets of structural criteria (see Discussion).
In addition, the grid exchanges are minimized in the new tSEM system because the current specimen holder accommodates up to 12 grids at a time (easily 200 serial sections) and there is certainly room for a larger holder in the chamber. The attended operation involves simply locating, adjusting scan rotation, and marking the center of the imaging field on each serial section, which takes several hours (not days) for 200 serial sections. The rest of the image acquisition process is automated for image focus and brightness optimization, acquisition, and stage translation from one field to the next. Since the fields are so large, slight shifts in positioning of the center mark do not dramatically affect the tissue volume available for subsequent 3D analysis. Furthermore, operator time remains constant even when montaging to enlarge the field because only the center of the entire field need be marked per section to guide the automated acquisition (Fig. 1E).
Because tSEM uses serial ultrathin sections to achieve the desired axial resolution, the same limitations of TEM apply regarding handling of fragile serial section ribbons and grids [31]. A long ribbon of serial sections must be broken into shorter segments to fit them within the slot of a TEM grid. Sections must be supported on low-structure, electron transparent support film, such as Pioloform. Depending on the quality of the knife and skill of the operator, cutting forces can produce compression or knife marks and tears, settling of the sections on the film can produce folds, and sections can be lost, especially during ribbon breakup. These artifacts can interfere with accurate local alignment, making 3D reconstructions challenging or even impossible. Hence, in the past with a TEM, sections were first viewed through the entire series to find small regions where the fields could be imaged across serial sections while minimizing encounters with artifacts. The larger tSEM imaging field makes it harder to avoid these potential artifacts. The new elastic alignment tool in TrakEM2 [43] is much less sensitive to the section artifacts than prior alignment strategies, and provides accurate alignment across sections that would otherwise be distorted (Movie S1). Thus, the combination of automated image acquisition, afforded by the tSEM, and the enhanced elastic alignment tool, quickly provides much larger volumes for quantitative ultrastructural analysis.
The main disadvantage of the tSEM approach is that it still involves collection of ultrathin sections requiring substantial operator skill. Difficulties of serial ultrathin sectioning include: (1) Generating a long continuous ribbon of serial ultrathin sections with uniform thickness. (2) Dividing the ribbon into shorter segments that fit on the slot grids. (3) Avoiding breakage or folds in the fragile electron transparent substrates used to fill the slot and support the sections. (4) Reducing compression due to contact with the diamond knife. (5) Achieving uniform section thickness and avoiding contamination from local environment or poor preparation of post-section heavy metal stains. We have published numerous methods to avoid or minimize these potential flaws in serial ultrathin sections [31], [45]. Ultrathin sections supported on film materials such as polyimide and silicon nitride (e.g., [55]) that are more rigid and/or stable might reduce folds, and decrease further the fragility of the support. Such materials may also allow for larger specimen windows, reducing the need to break up ribbons into smaller segments. 2ff7e9595c
Comments