Review figures

Two figures to look at a recording before trusting its numbers (fasteit.plotting). Colours and markers are the same in every figure.

Before preprocessing: the segments

from fasteit.models import Segment
from fasteit.parsers.loader import load_data
from fasteit.plotting import plot_segmentation

segments = [
    Segment(0, 290, "phase", "tidal"),        # baseline
    Segment(320, 1500, "phase", "peep_trial"),
    Segment(600, 640, label="movement"),       # exclusion
]
plot_segmentation(load_data("P01_T0.bin"), segments, title="P01, T0")

The global signal of the file (sum of all pixels) with the segments as bands: phases coloured by protocol (blue tidal, orange PEEP trial), exclusions grey and hatched, PEEP steps placed by hand with their level. Below, the airway pressure in cmH2O when the file has it.

After preprocessing: the breaths

from fasteit.plotting import plot_breaths

plot_breaths(trial)        # also a recording, TidalFeatures or PeepTrialFeatures

Here trial is the result of segment_peep_trial (analyse a recording).

Each breath is a marker at its end-inspiratory peak on the signal the breaths were found on:

Marker	Breath
filled blue circle	used for the indices: the valid breaths of the phase (tidal), the breaths analysed in each step (PEEP trial)
empty blue circle	valid but not used: outside the phase, or in the first seconds of a step, or beyond the last 10
orange triangle	flagged by a quality check
grey cross	excluded (the excluded parts are grey hatched bands)

For a PEEP trial the steps are shaded and labelled with their PEEP level of the protocol (peep_set, or the measured level without the protocol), and the PEEP of each breath is drawn below with the level of each step. On the right, the tidal image inside the lung mask (patient right on the left, ventral on top) and the count of each kind of breath.

Both functions return a matplotlib figure: fig.savefig("P01_T0_breaths.png") keeps one image per recording next to the exported tables (export). The times to write in the segments are found as in cut a recording. The figures work on reconstructed images only (Dräger .bin, Timpel).