-
Notifications
You must be signed in to change notification settings - Fork 3
Overview
Serial section microscopy results in set of sections that represent the sequential cross-sections of the tissue block. This software is organized around this concept.
A series consists of an ordered set of sections. This software allows the user to operate on one series at a time.
Data are grouped by section, with each section having an index indicating its position in the series.
Each section contains a pointer to an image. This image is used as the "background" for the section when viewing it. The images for the series are selected when the series is created.
Each section contains contour and trace data.
A trace is a single shape or curve on a section.
A contour is all traces with the same name on a SINGLE section.
An object is all traces with the same name throughout the ENTIRE series.
A ztrace is single curve or shape that exists on multiple sections.
A transform is a set of six numbers (a, b, c, d, e, f) that denotes an affine transformation.
The numbers a, b, d, and e describe the stretch/shear/rotation component of the transformation, while c and f describe the horizontal and vertical translation components of the transformation, respectively.
Each section has one transformation that is applied to the image and every trace. This means that the traces are fixed to the image. If the transformation for the section is changed, the traces and the image will move together.
An alignment is a set of transformations across a series, with each section having a single transform. (Transform is to section as alignment is to series).
If a series has multiple alignments, each section will have multiple transformations stored under these alignment names. While running the software, the user can switch between the different alignments.
For clarity, the different types of coordinates are defined here:
Field coordinates refer to the real-world location of an item.
Untransformed field coordinates refer to the location of an item before the section transformation has been applied.
Pixel coordinates refer to the coordinates in refernece to the image pixels.
Screen coordinates refer to the coordinates in refernece to the pixels on the output screen.
Data for each series are stored in a single series file (.jser).
The .jser file is in JSON format, which is a basic dictionary. It is split into two main keys: series and sections.
The value for the series key contains data pertaining to the series as a whole and its usage data. The key/value pairs in the series dictionary are as follows:
- current_section: the section index the series is focused on
- src_dir: the directory containing the image files (defined by the user)
- window: the x, y, width, and height of the viewing window for the series in field coordinates
- palette_traces: the traces that make up the trace palette
- ztraces: all of the ztraces in the series
- alignment: the name of the alignment setting (denotes a set of transforms)
- object_groups: a dictionary that contains information on which objects belong to which groups
- object_3D_modes: a dictionary that contains data for the 3D viewing settings for each object
- options: a dictionary containing various series options
The value for the sections key is a list, with each item in the list being a dictionary containing the data for a section. The key/value pairs in the section dictionary are as follows:
- src: the name of the image file (only the name -- not the path)
- brightness: user-adjusted brightness of the section (between -100 and 100)
- contrast: user-adjusted contrast of the section (between -100 and 100)
- mag: the magnification of the section (microns per image pixel)
- align_locked: true if the section is locked (meaning that its transformation cannot be changed)
- tforms: a dictionary of alignment name : transformation matrix pairs
- thickness: the thickness of the section
- contours: a dictionary of contour name : trace list pairs
- calgrid: set to true if the section is a calibration grid; false otherwise
Contours are stored within the section data. Contour data is stored as a dictionary, with each key being the name of a contour and each value being a list of the data for each trace corresponding to the contour name.
Traces are stored within the contours dictionary. A single trace is represented as a list. The items in the list are as follows (in the following order):
- the name of the trace (only for palette traces in the series data; traces within contours do NOT have a name)
- x-values for the trace points
- y-values for the trace points
- the color of the trace (rgb 0-255)
- whether or not the trace is closed
- whether or not the trace is negative
- whether or not the trace is hidden from view
- the fill style of the trace when viewing
- the tags of the trace
- the history of the trace
Ztraces are stored within the series file as a dictionary. Each key is the name of a ztrace, while each value is a dictionary containing ztrace data. The ztrace data dictionary is as follows:
- color: the color of the ztrace (rgb 0-255)
- points: the (x, y, section index) points of the ztrace
The following image formats are supported for use as section images: bmp, jpeg, jpg, pbm, pgm, png, ppm, tga, tif, tiff, wbmp, webp, xbm, xpm
Additionally, images can be stored in the zarr format. This is highly recommended for large images, as the software will load only the parts of the image that are needed (rather than the entire image every time the user changes sections).
Each image is stored in a directory with the extension .zarr (ex. ABC_images.zarr). This directory contains every image, which are individually stored as directories themselves. These image directories are named directly after the image file (ex. image.tif) and contain the zarr data for the untransformed (base) image.
The libraries that PyReconstruct uses are the following:
- lxml: used to import/export data from Legacy RECONSTRUCT
- numpy: used for higher-level number and array processing
- opencv-python: used to obtain exteriors/interiors of traces and reduce the number of points in a trace
- PyOpenGL: used to generate 3D models
- pyqtgraph: used to view the 3D models
- PySide6: the main engine for the GUI
- scikit-image: used to assist generating the 3D model
- trimesh: used to generate the 3D model
- zarr: used for understanding images in zarr format