Skip to content

Tools for quantitative evaluation of prediction segmentations (Dice, Jaccard, volume similarity, false_negative, false_positive, mean, median, max, and std surface distances)

License

Notifications You must be signed in to change notification settings

brianmanderson/Segmentation_Evaluation_Tools

Repository files navigation

Segmentation_Evaluation_Tools

There are a set of tools for creating quantitative comparison metrics based on ground truth and prediction SITK Image handles

Installation

pip install SegmentationEvaluationTools

Usage

from SegmentationEvaluationTools.SIKOverlapTools import calculate_overlap_measures, determine_sensitivity, 
determine_false_positive_rate_and_false_volume, sitk

truth_handle_base = sitk.ReadImage(image_path)
prediction_handle_base = sitk.ReadImage(prediction_path)

overlap_measures = calculate_overlap_measures(prediction_handle_base, truth_handle_base, measure_as_multiple_sites=False, perform_distance_measures=False)

fp_measures = determine_false_positive_rate_and_false_volume(prediction_handle_base, truth_handle_base)

sensitivity_measures = determine_sensitivity(prediction_handle=prediction_handle_base, truth_handle_base)

About

Tools for quantitative evaluation of prediction segmentations (Dice, Jaccard, volume similarity, false_negative, false_positive, mean, median, max, and std surface distances)

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages