spateo.segmentation.benchmark#

Functions to help segmentation benchmarking, specifically to compare two sets of segmentation labels.

Module Contents#

Functions#

adjusted_rand_score(→ float)

Compute the Adjusted Rand Score (ARS).

iou(→ scipy.sparse.csr_matrix)

Compute intersection-over-union (IOU).

average_precision(→ float)

Compute average precision (AP).

classification_stats(→ Tuple[float, float, float, ...)

Calculate pixel classification statistics by considering labeled pixels

labeling_stats(→ Tuple[float, float, float, float])

Calculate labeling (cluster) statistics.

compare(), seed)

Compute segmentation statistics.
spateo.segmentation.benchmark.adjusted_rand_score(y_true: numpy.ndarray, y_pred: numpy.ndarray) → float[source]#

Compute the Adjusted Rand Score (ARS).

Re-implementation to deal with over/underflow that is common with large datasets.

Parameters:
y_true

True labels

y_pred

Predicted labels

Returns:

Adjusted Rand Score

spateo.segmentation.benchmark.iou(labels1: numpy.ndarray, labels2: numpy.ndarray) → scipy.sparse.csr_matrix[source]#

Compute intersection-over-union (IOU).

Parameters:
labels1

First set of labels

labels2

Second set of labels

Returns:

Sparse matrix where the first axis corresponds to the first set of

labels and vice-versa.

spateo.segmentation.benchmark.average_precision(iou: scipy.sparse.csr_matrix, tau: float = 0.5) → float[source]#

Compute average precision (AP).

Parameters:
iou

IOU of true and predicted labels

tau

IOU threshold to determine whether a prediction is correct

Returns:

Average precision

spateo.segmentation.benchmark.classification_stats(y_true: numpy.ndarray, y_pred: numpy.ndarray) → Tuple[float, float, float, float, float, float, float][source]#

Calculate pixel classification statistics by considering labeled pixels as occupied (1) and unlabled pixels as unoccupied (0).

Parameters:
y_true

True labels

y_pred

Predicted labels

Returns:

  • true negative rate

  • false positive rate

  • false negative rate

  • true positive rate (a.k.a. recall)

  • precision

  • accuracy

  • F1 score

Return type:

A 7-element tuple containing the following values

spateo.segmentation.benchmark.labeling_stats(y_true: numpy.ndarray, y_pred: numpy.ndarray) → Tuple[float, float, float, float][source]#

Calculate labeling (cluster) statistics.

Parameters:
y_true

True labels

y_pred

Predicted labels

Returns:

  • adjusted rand score

  • homogeneity

  • completeness

  • v score

Return type:

A 4-element tuple containing the following values

spateo.segmentation.benchmark.compare(adata: anndata.AnnData, true_layer: str, pred_layer: str, data_layer: str = SKM.X_LAYER, umi_pixels_only: bool = True, random_background: bool = True, ap_taus: Tuple[int, Ellipsis] = tuple(np.arange(0.5, 1, 0.05)), seed: int | None = None) → pandas.DataFrame[source]#

Compute segmentation statistics.

Parameters:
adata

Input Anndata

true_layer

Layer containing true labels

pred_layer

Layer containing predicted labels

data_layer

Layer containing UMIs

umi_pixels_only

Whether or not to only consider pixels that have at least one UMI captured (as determined by data_layer).

random_background

Simulate random background by randomly permuting the pred_layer labels and computing the same statistics against true_layer. The returned DataFrame will have an additional column for these statistics.

ap_taus

Tau thresholds to calculate average precision. Defaults to 0.05 increments starting at 0.5 and ending at (and including) 0.95.

seed

Random seed.

Returns:

Pandas DataFrame containing classification and labeling statistics