pyfli.data_vnp.mono_bi_classifier#

Classify mono- versus bi-exponential fit agreement and visualize parameter correlations.

This module belongs to pyfli.data_vnp and is part of PyFLI visualization, normalization, plotting, and mono-versus-bi-exponential comparison tools. Public API includes classes MonoBiClassifier and ParamCorrelationMatrix.

Classes

`MonoBiClassifier`(b_bool_mask[, names, ...])	Classify agreement between mono- and bi-exponential fits across one or more datasets.
`ParamCorrelationMatrix`(all_datasets, bool_mask)	Visualize pairwise parameter relationships under an agreement mask.

class MonoBiClassifier(b_bool_mask, names=None, alpha_upper=0.95, alpha_lower=0.05, tau_tol=0.01, coord=None, figsize=None)[source]#

Bases: object

Classify agreement between mono- and bi-exponential fits across one or more datasets. It builds masks from lifetime and fraction criteria, summarizes agreement, and plots parameter comparisons.

Parameters:

b_bool_mask (np.ndarray) – Mask array used to select or label pixels.
names (Any | None) – Names used to label datasets, classes, or plotted groups.
alpha_upper (float) – Upper alpha-fraction threshold used by the classifier.
alpha_lower (float) – Lower alpha-fraction threshold used by the classifier.
tau_tol (float) – Lifetime tolerance used by the mono/bi classifier.
coord (Any | None) – Pixel or ROI coordinate used for lookup and plotting.
figsize (np.ndarray | None) – Figure size passed to Matplotlib.

CMAP_NAMES = ('jet', 'Spectral', 'Spectral_r')#

PALETTE: ClassVar[list[str]] = ['#5DADE2', '#EC7063', '#58D68D', '#F4D03F', '#AF7AC5', '#EB984E', '#48C9B0', '#52BE80', '#AAB7B8', '#F1948A', '#BB8FCE', '#7FB3D5', '#76D7C4']#

classify_one(res, name='Dataset')[source]#

Classify one.

Parameters:

res (Any) – Fit result object or dataset result dictionary.
name (str) – Dataset, experiment, figure, or output name.

Returns:

Dictionary containing the data produced by classify one.

Return type:

dict[Any, Any]

display_one(result)[source]#

Display one.

Parameters:: result (Any) – Classification or fitting result to display.
Returns:: No object is returned; the function display one.
Return type:: None

classify(all_datasets, names=None, display=True)[source]#

Classify every dataset; store self.results and self.all_datasets.

Parameters:

all_datasets (ndarray)
names (Any | None)
display (bool)

Return type:

Any

summary()[source]#

Run the summary routine.

Returns:: Object produced by summary.
Return type:: Any

agreement(metric='jaccard', classes_to_show=('mono', 'bi'), cmap='viridis', figsize=None, show=True)[source]#

Pairwise agreement between methods on the mono/bi classification.

metric‘count’ raw # pixels both methods call this class: ‘jaccard’ |A∩B| / |A∪B| (symmetric, 0..1) ‘fraction’ |A∩B| / |A| (asymmetric, read by row)

Returns (dict {class: NxN matrix}, fig).

Parameters:

metric (str)
classes_to_show (tuple[str, ...])
cmap (str)
figsize (ndarray | None)
show (bool)

Return type:

tuple[Any, …]

param_scatter_matrix(param='tau1_map', cls='mono', agree='pairwise', max_points=3000, colors=None, figsize=None, rng=None, show=True)[source]#

Cross-method correlation of ONE parameter over the pixels of ONE class.

diagonal (i, i) histogram of param for method i over its cls pixels off-diag (i, j) scatter of method_i (y) vs method_j (x), 1:1 line + r

agree‘pairwise’ (cell uses pixels both i & j call cls): ‘all’ (every cell uses pixels ALL methods agree on)

Parameters:

param (str)
agree (str)
max_points (int)
colors (Any | None)
figsize (ndarray | None)
rng (Any | None)
show (bool)

Return type:

ndarray

agreed_param_table(cls='mono', params=('alpha1_map', 'tau1_map', 'tau2_map'))[source]#

Long-form table of parameter values at pixels where ALL methods agree on cls. One row per (pixel, method) -> groupby(‘method’).describe().

Parameters missing from a dataset (e.g. Phasor has only ‘tau_map’, mono-exp datasets lack ‘alpha1_map’/’tau2_map’) are filled with NaN instead of raising KeyError.

Parameters:: params (tuple[str, ...])
Return type:: Any

class ParamCorrelationMatrix(all_datasets, bool_mask, names=None)[source]#

Bases: object

Visualize pairwise parameter relationships under an agreement mask. It supports subsampling, scalar and distribution-valued maps, uncertainty display, and pairwise scatter panels.

Parameters:

all_datasets (np.ndarray) – Sequence or mapping of fitted datasets to compare.
bool_mask (np.ndarray) – Boolean mask selecting pixels included in the analysis.
names (Any | None) – Names used to label datasets, classes, or plotted groups.

PALETTE: ClassVar[list[str]] = ['#5DADE2', '#EC7063', '#58D68D', '#F4D03F', '#AF7AC5', '#EB984E', '#48C9B0', '#52BE80', '#AAB7B8', '#F1948A', '#BB8FCE', '#7FB3D5', '#76D7C4']#

scatter_matrix(param='tau1_map', agree='pairwise', max_points=3000, colors=None, figsize=None, rng=None, show=True)[source]#

N × N cross-method scatter matrix for a single parameter.

Diagonal : histogram of pixel means within the ROI. Off-diagonal: dataset_i (y) vs dataset_j (x) with identity line + r.

Scalar data → scatter points. Distribution data → mean circle with ± std error bars.

Parameters:

param (str Key present in the dataset dicts.)
agree (str 'pairwise' — each cell uses pixels finite in both) – datasets. ‘all’ — restricted to pixels finite in ALL datasets simultaneously.
max_points (int Cap on plotted points (random sub-sample).)
colors (list Per-dataset colours (cycles PALETTE by default).)
figsize (tuple Figure size; auto-scaled to N if None.)
rng (np.random.Generator For reproducible sub-sampling.)
show (bool Call plt.show() when True.)

Returns:

fig

Return type:

matplotlib.figure.Figure

pairwise_scatter(idx_a, idx_b, params=('tau1_map', 'tau2_map', 'alpha1_map'), max_points=3000, colors=None, figsize=None, rng=None, show=True)[source]#

Scatter plots for multiple parameters between exactly two datasets.

Scalar parameter (H, W) → plain scatter point per pixel. Distribution parameter (H, W, N) → mean circle with ± std error bars.

Parameters:

idx_a (int or str) – Dataset for y-axis (idx_a) and x-axis (idx_b). Accepts integer position or name string.
idx_b (int or str) – Dataset for y-axis (idx_a) and x-axis (idx_b). Accepts integer position or name string.
params (sequence of str) – Parameter keys to compare; missing keys produce blank panels.
max_points (int Cap on plotted points (random sub-sample).)
colors (list One colour per parameter panel (cycles PALETTE).)
figsize (tuple Figure size; auto-sized to number of params if None.)
rng (np.random.Generator For reproducible sub-sampling.)
show (bool Call plt.show() when True.)

Returns:

fig

Return type:

matplotlib.figure.Figure