Step 2 (S2): Localize particles¶
Inputs for S2¶
Inputs for S2 are provided through a YAML file containing parameters. An example is provided in examples/s2_params_example.yaml. These parameters are described in detail below:
dataset_name is also the name of the directory where all the outputs will be saved.
inputs:
[
{
segmentation: <path_to_segmentation_1>,
particle_cluster_id: 1, # Cluster index of the particle cluster
lower_z-slice_limit: <upper_zslice_where_the_lamella_starts>, #[Optional]#
upper_z-slice_limit: <lower_zslice_where_the_lamella_ends> #[Optional]#
},
{
segmentation: <path_to_segmentation_1>,
particle_cluster_id: 0,
lower_z-slice_limit: <upper_zslice_where_the_lamella_starts>, #[Optional]#
upper_z-slice_limit: <lower_zslice_where_the_lamella_ends> #[Optional]#
},
]
inputs is a list (enclosed within square brackets) that can be expanded with similar entries, enclosed in curly brackets as shown above.
segmentation and the corresponding particle_cluster_id are obtained from the visualizing segmentations.
lower_z-slice_limit and upper_z-slice_limit denote the upper and lower bounds on the Z-slices where the tomogram is likely to contain particles.
Note
These bounds define the bounds on the region from which particles will be picked. These bounds can be more relaxed than the ones used for generating semantic segmentations. See also Fig 2.
Fig. 2: Z-slice bounds for the two steps in PickET
particle_extraction_params:
[
{mode: connected_component_labeling},
{mode: watershed_segmentation, min_distance: 15}
]
Similar to the inputs, particle_extraction_params is also a list of dictionaries. Each dictionary defined in this list defines a particle extraction mode. Here, we provide two particle extraction modes connected_component_labeling and watershed_segmentation.
First, for mode: connected_component_labeling, there are no hyperparameters. This mode is fast and works well for less crowded datasets.
Second, for mode: watershed_segmentation, there is one hyperparameter. This mode uses watershed segmentation for converting the semantic segmentation into instance segmentation. It uses the min_distance hyperparameter that defines the minimum separation between two detected particles in voxels.
As the name suggests, output_dir describes the path to the directory where the output segmentations will be saved.
Note
The extracted particle centroid coordinates will be saved as .yaml files in output_dir/dataset_name/predicted_particles/ directory.