Metrics

Metrics are grouped together according to which PyFibre objects they are calculated from (fibre network, fibre segment, cell segment, global image). The 4 database types are:

Fibre Network

Contain metrics calculated from each network of fibres extracted by the FIRE algorithm:

<file-name>_network_metric.h5
<file-name>_network_metric.xls
Table 2 Fibre Network Metrics

Metric

Description

Category

No. Fibres

Number of extracted fibres

Content

Mean Fibre Waviness

Average fibre waviness (length / displacement)

Content

Mean Fibre Length

Average fibre pixel length

Content

Fibre Angle SDI

Angle spectrum SDI (mean / max) of fibre angles

Content

Fibre Network Degree

Average fibre network number of edges per node

Network

Fibre Network Eigenvalue

Max eigenvalue of network adjacency matrix

Network

Fibre Network Connectivity

Average fibre network connectivity

Network

Fibre Network Cross-Link Density

Average cross-links per fibre

Network

Fibre Segment

Contain metrics calculated from each fibre segment identified in the segmentation process:

<file-name>_fibre_metric.h5
<file-name>_fibre_metric.xls
Table 3 Fibre Segment Metrics

Metric

Description

Category

Fibre Segment Area

Area of fibre segment (in pixels)

Shape

Fibre Segment Coverage

Ratio of fibre segment area / bounding box area

Content

Fibre Segment Circularity

Measure of how circular the fibre segment is

Shape

Fibre Segment Eccentricity

Eccentricity shape metric of fibre segment

Shape

Fibre Segment SHG Angle SDI

Angle spectrum SDI (mean / max) for all SHG image pixels in fibre segment

Texture

Fibre Segment SHG Anisotropy

Anisotropy of mean structure tensor of all SHG image pixels in fibre segment

Texture

Fibre Segment SHG Local Anisotropy

Mean anisotropy of structure tensors of all SHG image pixels in fibre segment

Texture

Fibre Segment SHG Mean

Mean intensity of fibre segment pixels in SHG image

Texture

Fibre Segment SHG STD

Standard deviation of fibre segment pixels in SHG image

Texture

Fibre Segment SHG Entropy

Average Shannon entropy of fibre segment pixels in SHG image

Cell Segment

Contain metrics calculated from each cell segment identified in the segmentation process:

<file-name>_cell_metric.h5
<file-name>_cell_metric.xls
Table 4 Cell Segment Metrics

Metric

Description

Category

No. Cells

Number of cell segments

Content

Cell Segment Area

Area of cell segment (in pixels)

Shape

Cell Segment Coverage

Ratio of cell segment area / bounding box area

Content

Cell Segment Circularity

Measure of how circular the cell segment is

Shape

Cell Segment Eccentricity

Eccentricity shape metric of cell segment

Shape

Cell Segment PL Angle SDI

Angle spectrum SDI (mean / max) for all PL image pixels in cell segment

Texture

Cell Segment PL Anisotropy

Anisotropy of mean structure tensor of all PL image pixels in cell segment

Texture

Cell Segment PL Local Anisotropy

Mean anisotropy of structure tensors of all PL image pixels in cell segment

Texture

Cell Segment PL Mean

Mean intensity of cell segment pixels in PL image

Texture

Cell Segment PL STD

Standard deviation of cell segment pixels in PL image

Texture

Cell Segment PL Entropy

Average Shannon entropy of cell segment pixels in PL image

Global Image

Contain metrics calculated from the global image:

<file-name>_global_metric.h5
<file-name>_global_metric.xls

Contains averages of all network and segment metrics that are created during the analysis process

Output Files

PyFibre will create a folder with relevant analysis for each multi image that is loaded into the software. These contain database files containing all metrics for each identified object in the SHG-PL-Trans images in both hdf5 and xls formatting. PNG images displaying the analysis are also created if the --save_figures option is selected:

file-name-shg.tif
file-name-pl.tif
file-name-pyfibre-analysis/
│
├── data
│    │
│    ├── file-name_global_metric.h5
│    ├── file-name_global_metric.xls
│    ├── file-name_cell_metric.h5
│    ├── file-name_cell_metric.xls
│    ├── file-name_fibre_metric.h5
│    ├── file-name_fibre_metric.xls
│    ├── file-name_network_metric.h5
│    ├── file-name_network_metric.xls
│    ├── file-name_network.pkl
│    ├── file-name_fibre_segments.npy
│    └── file-name_cell_segments.npy
│
│
└── fig
     ├── file-name_cell_seg.png
     ├── file-name_fibre_seg.png
     ├── file-name_fibre.png
     ├── file-name_network.png
     ├── file-name_tensor.png
     ├── file-name_PL.png
     ├── file-name_SHG.png
     └── file-name_trans.png

The additional raw data files in the data directory contain serialised copies of the segment and network objects used in the analysis:

file-name_network.pkl

Contains a pickled copy of the NetworkX graph used to represent the fibre networks:

file-name_fibre_segments.npy
file-name_cell_segments.npy

Contain stacks of NumPy arrays representing pixel masks of each fibre and cell segment identified in the segmentation process.

Databases

Databases are also generated from all SHG-PL-Trans images that are loaded into the PyFibre software during a session. The names of these output files can be customised by setting the --database_name flag in the CLI or specifying a name in the “Save Database” GUI tool:

<database_name>_global.h5
<database_name>_global.xls

Contains a set of global metrics for each image:

<database_name>_network.h5
<database_name>_network.xls

Contains metrics for every fibre network in each image:

<database_name>_fibre.h5
<database_name>_fibre.xls

Contains metrics for every fibre segment in each image:

<database_name>_cell.h5
<database_name>_cell.xls

Contains metrics for every cell segment in each image.