sksurgerytf.utils.segmentation_statistics module

Module to implement various segmentation statistics for evaluation.

sksurgerytf.utils.segmentation_statistics.calculate_dice(gold_standard, segmentation)

Computes dice score of two boolean images.

Inspired by NiftyNet.

Parameters:

• gold_standard – gold standard / reference image.
• segmentation – segmented / predicted / inferred image.

Returns:

dice score

sksurgerytf.utils.segmentation_statistics.check_same_size(image_a, image_b)

Check shape the same.

Parameters:

• image_a – image
• image_b – image

Returns:

sksurgerytf.utils.segmentation_statistics.get_confusion_matrix(gold_standard, segmentation)

Compute the confusion matrix of 2 boolean images.

Inspired by NiftyNet.

Parameters:

• gold_standard – gold standard / reference image.
• segmentation – segmented / predicted / inferred image.

Returns:

2x2 confusion matrix, [[TN, FN],[FP,TP]].

sksurgerytf.utils.segmentation_statistics.get_sorted_files_from_dir(directory)

Retrieves all files in directory, sorted. :param directory: directory path name :return: list of file names

sksurgerytf.utils.segmentation_statistics.run_seg_stats(gold_standard, segmentation)

Compares segmentation image(s) to gold standard images(s).

Parameters:

• gold_standard – directory, or single image
• segmentation – directory, or single image

Returns:

list of stats