create-gold-standard

Create a gold standard segmentation from the current Slicer session.

Usage

/create-gold-standard <name> [--algorithm <algo>] [--description "<desc>"]

Where:

• •name - Name for the gold standard (e.g., "MRBrainTumor1_tumor")
• •--algorithm - Algorithm used (default: watershed)
• •--description - Description of the gold standard

What This Skill Does

1. •Reads the current Slicer session state
2. •Prompts user to verify/refine the segmentation
3. •Records click locations and parameters
4. •Saves to GoldStandards/<name>/
5. •Generates reference screenshots

Prerequisites

1. •Slicer is running with the extension loaded
2. •A segmentation has been created that you want to save as gold standard
3. •The source volume is loaded (e.g., MRBrainTumor1)

Manual Steps

Since gold standards require manual verification, this skill provides guidance:

Step 1: Prepare the Segmentation

In Slicer:

1. •Load sample data (e.g., MRBrainTumor1)
2. •Create a segmentation using Adaptive Brush
3. •Refine until quality is acceptable
4. •Note the click locations used

Step 2: Save Using Python Console

Open Slicer's Python console (View > Python Interactor) and run:

from SegmentEditorAdaptiveBrushTesterLib import GoldStandardManager

manager = GoldStandardManager()

# Get current nodes
seg_node = slicer.util.getNode("Segmentation")
vol_node = slicer.util.getNode("MRBrainTumor1")

# Define click locations (RAS coordinates)
clicks = [
    {"ras": [-5.31, 34.77, 20.83]},
    {"ras": [-5.31, 25.12, 35.97]},
    # Add more as needed
]

# Save gold standard
manager.save_as_gold(
    segmentation_node=seg_node,
    volume_node=vol_node,
    segment_id="Segment_1",  # Or your segment ID
    name="MRBrainTumor1_tumor",
    click_locations=clicks,
    description="Tumor segmentation using watershed, 5 clicks",
    algorithm="watershed",
    parameters={
        "brush_radius_mm": 25.0,
        "edge_sensitivity": 40,
    }
)

Step 3: Capture Reference Screenshots

After saving, capture reference screenshots:

# Navigate to key slices and use the tester panel to capture screenshots
# Or use the ScreenshotCapture utility

Gold Standard Directory

Gold standards are saved to:

SegmentEditorAdaptiveBrushTester/GoldStandards/
└── MRBrainTumor1_tumor/
    ├── gold.seg.nrrd          # The segmentation
    ├── metadata.json          # Parameters, clicks, etc.
    └── reference_screenshots/

Verification

After creating a gold standard:

1. •Run /run-regression to verify it can be reproduced
2. •Check the Dice coefficient matches expectations
3. •Review reference screenshots for visual quality

Tips

• •Use consistent click locations that can be reproduced
• •Document the algorithm and parameters used
• •Include enough clicks to achieve > 90% Dice on reproduction
• •Take screenshots from multiple views for reference