12. Image sequence properties

12.1. Cell unique id

Properties are computed over a temporal series of 3D images. The time point corresponding to an image is encoded into the image name by a string _t<ttt> where <ttt> are three digits. Time points may start at 0 (ie _t000), 1 (ie _t001), or later, but are a continuous series of integers.

In one 3D image, or snapshot, objects of interest (cells) are individualized by a label (ie an integer number). By convention, the background used the 1 label, and the cell labels are strictly larger than 1, but they are not necessarily a continuous series of integer numbers starting at 2.

A same label can be used in different snapshots. To differentiate cells from the different snapshots, each cell is being given an unique id, defined by \(T \times 10000 + label\), where T is the time point of the image and label the cell identifier in the image. Thus the id 100256 designates the cell of label 256 in the snapshot of time point 10.

12.2. Properties file format

The handled format are

  • pkl: a native python format. It can be used for its efficiency. Properties are stored as a python dictionary, each entry corresponding to a given property.

  • xml: as a text format, it can be easily parsed. Properties are enclosed in the element defined by the data tag as above.

<data>
  ...
</data>

12.3. Properties

Each given property are identified by a tag in xml files, which is also the key for the same properties in the python dictionary stored in pkl files.

The command

$ astec_embryoproperties -i file.pkl --print-content

displays the properties stored in a property file. The same stands for xml files.

In pkl files, each property is also a dictionary whose keys are cell unique ids, while in xml files, each property is an element tagged by cell with the cell-id attribute.

12.3.1. Properties computed by astec_astec

Cf section astec_astec.

  • cell_volume: cell volume (in voxel unit)

    <data>
  ...
  <cell_volume>
    ...
    <cell cell-id="100256">1234000</cell>
    ...
  </cell_volume>
  ...
</data>

    indicates that cell labeled 256 in image of time 10 is made of 1234000 voxels. For a volume in \(\mu m^3\), this measure has to be multiplied by the cube of the voxel size (assuming voxels are isotropic).

  • cell_lineage: relationships between cells of successive snapshots

    <data>
  ...
  <cell_lineage>
    ...
    <cell cell-id="100256">[110257]</cell>
    <cell cell-id="100257">[110300, 110301]</cell>
    ...
  </cell_lineage>
  ...
</data>

    indicates that

    • the cell labeled 256 in image of time 10 has been labeled 300 in image of time 11,

    • the cell labeled 257 in image of time 10 divided into the cells labeled 300 and 301 in image of time 11.

12.3.2. Properties computed by astec_embryoproperties

Additional properties can be computed by astec_embryoproperties (cf section astec_embryoproperties).

  • cell_surface: cell surface (in pixel unit)

    <data>
  ...
  <cell_surface>
    ...
    <cell cell-id="100256">42000.42</cell>
    ...
  </cell_surface>
  ...
</data>

    indicates that cell labeled 256 in image of time 10 has a total surface of 42000.42 pixels. For a surface in \(\mu m^2\), this measure has to be multiplied by the square of the voxel size (assuming voxels are isotropic).

  • cell_barycenter: cell center of mass (in voxel unit)

    <data>
  ...
  <cell_barycenter>
    ...
    <cell cell-id="100256">[311.75, 288.18, 148.37]</cell>
    ...
  </cell_barycenter>
  ...
</data>

    indicates that the center of mass of cell labeled 256 in image of time 10 has coordinates \([311.75, 288.18, 148.37]\) in voxel units.

  • cell_principal_values: principal value are issued from the diagonalization of the cell covariance matrix (in voxel unit). They are sorted in decreasing order.

    <data>
  ...
  <cell_principal_values>
    ...
    <cell cell-id="100256">[2175.17, 936.26, 799.08]</cell>
    ...
  </cell_principal_values>
  ...
</data>

  • cell_principal_vectors: principal vectors are issued from the diagonalization of the cell covariance matrix (in voxel unit). The vector ordering corresponds to the one of the principal values.

    <data>
  ...
  <cell_principal_vectors>
    ...
    <cell cell-id="100256">[[-0.42, -0.74, -0.51], [-0.07, 0.59, -0.80], [0.90, -0.29, -0.30]]</cell>
    ...
  </cell_principal_vectors>
  ...
</data>

  • cell_contact_surface: contact surfaces with the adjacent cells (in pixel unit).

    <data>
  ...
  <cell_contact_surface>
    ...
    <cell cell-id="100256">
      <cell cell-id="100001">1234.00</cell>
      <cell cell-id="100257">4567.00</cell>
      <cell cell-id="100300">13579.00</cell>
      ...
    </cell>
    ...
  </cell_contact_surface>
  ...
</data>

    indicates that the cell labeled 256 in image of time 10 is adjacent to

    • the background (cell id = \(100001 = 10 \times 10000 + 1\)) with a contact surface of 1234.00

    • cell labeled 257 (in image of time 10) with a contact surface of 4567.00

    • cell labeled 300 (in image of time 10) with a contact surface of 13579.00

    • …