TavrWorkbench

A 3D Slicer extension for reviewing TAVR (transcatheter aortic valve replacement) CT segmentations and measurements, rate quality case by case, leave comments, and track progress across a dataset.

What it does

• Loads CT volumes with their segmentations (Aorta, Aortic Root, Left Ventricle, Coronaries, Thoracic Aorta) one case at a time.
• Lets you rate each case overall, per segmentation label, and per measurement (✓ Acceptable / △ Minor correction / ✕ Not acceptable), plus a free-text comment.
• Tracks review progress (Total / Reviewed / Pending) and saves everything to disk as you go, so you can stop and resume later.
• Aligns the 2D slice views to a loaded contour (annulus, SOV, STJ), or slides a cross-section along a centerline curve.
• Includes a built-in Segment Editor and Markups panel for correcting segmentations directly.

Requirements

• 3D Slicer (with the Segmentations and Markups modules, and the SegmentStatistics module).
• Python packages pandas, numpy, SimpleITK, installed automatically on first run if missing.

Installing

1. Open Slicer → Developer Tools → Extension Wizard.
2. Click Select Extension and point it at the TavrWorkbench folder.
3. The TavrWorkbench module will appear in the module dropdown.

Input data

You can load a dataset in one of two ways:

Directory mode : pick a folder containing:

• *.nii / *.nii.gz / *.nrrd volumes, optionally paired with masks named <volume>_mask.nii.gz
• or a mapping.csv / mapping_unique.csv with img_path, mask_path (and subj_id for the "unique" variant)

Recommended:

JSON manifest mode : pick a .json file shaped like this (top-level metadata is optional/informational; only samples is read):

{
  "name": "Full Dataset for TAVR-related Structure Segmentation",
  "description": "101 CTA scans with Annotations for Aorta, Aortic Root and Left Ventricle annotations along with the measurements",
  "modality": "CT",
  "totaldatasample": 101,
  "samples": [
    {
      "case_id": "568",
      "image": "/Volumes/falcon/tavr_application/imgs/568.img.nii.gz",
      "label": "/Volumes/falcon/tavr_application/stj_labels_corrected_v1/568.seg.nrrd",
      "hinge_points": {
        "LCC": "/Volumes/falcon/tavr_application/hinge_points/v2/568/LCC.mrk.json",
        "RCC": "/Volumes/falcon/tavr_application/hinge_points/v2/568/RCC.mrk.json",
        "NCC": "/Volumes/falcon/tavr_application/hinge_points/v2/568/NCC.mrk.json"
      },
      "contours": {
        "annulus": "/Volumes/falcon/tavr_application/annulus_results_17/568/annulus_contour.mrk.json",
        "sov": "/Volumes/falcon/tavr_application/sov/568/sov_contour.mrk.json",
        "stj": "/Volumes/falcon/tavr_application/stj_contour/568/stj_contour.mrk.json"
      },
      "max_diameters": {
        "annulus": "/Volumes/falcon/tavr_application/annulus_results_17/568/annulus_max_d.mrk.json",
        "sov": "/Volumes/falcon/tavr_application/sov/568/sov_max_diameter_line.mrk.json",
        "stj": "/Volumes/falcon/tavr_application/stj_contour/568/stj_max_d.mrk.json"
      },
      "min_diameters": {
        "annulus": "/Volumes/falcon/tavr_application/annulus_results_17/568/annulus_min_d.mrk.json",
        "sov": "/Volumes/falcon/tavr_application/sov/568/sov_min_diameter_line.mrk.json",
        "stj": "/Volumes/falcon/tavr_application/stj_contour/568/stj_min_d.mrk.json"
      },
      "heights": {
        "stj": "/Volumes/falcon/tavr_application/stj_heights/568/STJ_H.mrk.json",
        "lch": "/Volumes/falcon/tavr_application/coronary_ostium_points/568/LCH.mrk.json",
        "rch": "/Volumes/falcon/tavr_application/coronary_ostium_points/568/RCH.mrk.json"
      }
    }
  ]
}

Per sample, case_id, image, and label are the essentials; hinge_points, contours, max_diameters, min_diameters, and heights each hold named .mrk.json markup files that get loaded and (aside from hinge_points) show up as rows in the Measurement Review table. A centerline group is also supported, for .mrk.json markup points plus optional .vtp model files (e.g. centerline.surface, centerline.curve) used by the Centerline Slicing panel. Any extra fields your data has, like dirs or aortic_angle, are simply ignored by the extension.

Output files

Written into the same directory as your input:

File	Contents
annotations.csv	One row per reviewed case, with all ratings and comments
detailed_reviews.json	Same review data, keyed by case ID
tavr_workbench.log	Run log for troubleshooting
<case_id>.seg.nrrd	Saved when you click Save Mask to overwrite a corrected segmentation

Contributors

Mohammed Khubaib (Saeed Lab, Florida International University), under the supervision of Dr. Fahad Saeed and Dr. Kaoutar Ben Ahmed.

License