Calculate Jaccard

calculate_jaccard - Segmentation Overlap Measurement Tool

Part of the micaflow processing pipeline for neuroimaging data.

This module calculates the Jaccard similarity index (intersection over union) between two segmentation volumes. The Jaccard index is a statistic used for comparing the similarity and diversity of sample sets, with values ranging from 0 (no overlap) to 1 (perfect overlap). It is particularly useful for evaluating the quality of segmentations against a ground truth or comparing results from different methods.

Features:

Support for multi-label segmentations with per-ROI analysis
Global Jaccard calculation across the entire volume
Optional masking to restrict comparison to specific regions
Configurable threshold for probabilistic segmentations
CSV output format for easy integration with analysis workflows

API Usage:

micaflow calculate_jaccard: –input <path/to/segmentation.nii.gz> –reference <path/to/ground_truth.nii.gz> –output <path/to/results.csv> [–mask <path/to/mask.nii.gz>] [–threshold <value>]

Python Usage:

>>> from micaflow.scripts.calculate_jaccard import main
>>> main(
...     image="segmentation.nii.gz",
...     reference="ground_truth.nii.gz",
...     output_file="jaccard_results.csv",
...     threshold=0.5,  # optional
...     mask_path="brain_mask.nii.gz"  # optional
... )

Command Line Usage

micaflow calculate_jaccard [options]

Source Code

View the source code: GitHub Repository

Description

This script calculates the Jaccard similarity index (intersection over union): between two segmentation volumes, either globally or for each ROI.

Full Help

╔════════════════════════════════════════════════════════════════╗
║                    JACCARD INDEX CALCULATOR                    ║
╚════════════════════════════════════════════════════════════════╝

This script calculates the Jaccard similarity index (intersection over union)
between two segmentation volumes, either globally or for each ROI.

────────────────────────── USAGE ──────────────────────────
  micaflow calculate_jaccard [options]

─────────────────── REQUIRED ARGUMENTS ───────────────────
  --input, -i     : Path to the first input segmentation volume (.nii.gz)
  --reference, -r : Path to the reference segmentation volume (.nii.gz)
  --output, -o    : Output path for the CSV file with Jaccard indices

─────────────────── OPTIONAL ARGUMENTS ───────────────────
  --mask, -m      : Optional mask to restrict comparison to a specific region
  --threshold, -t : Threshold value for probabilistic segmentations (default: 0.5)

──────────────────── EXAMPLE USAGE ──────────────────────
  micaflow calculate_jaccard \
    --input segmentation1.nii.gz \
    --reference ground_truth.nii.gz \
    --output jaccard_metrics.csv

  # With mask and custom threshold:
  micaflow calculate_jaccard \
    --input segmentation1.nii.gz \
    --reference ground_truth.nii.gz \
    --output jaccard_metrics.csv \
    --mask brain_mask.nii.gz \
    --threshold 0.75

────────────────────────── NOTES ─────────────────────────
- For multi-label segmentations, the Jaccard index is computed for each label
- Values range from 0 (no overlap) to 1 (perfect overlap)
- A global Jaccard index is calculated across all labels