skills.rest
NVIDIA
NVIDIAOFFICIAL
Explore all skills by this author →
WebsiteGitHub

i4h-catheter-navigation-digital-twin

Builds a patient vasculature digital twin from CT volumes via preprocessing and vessel segmentation.

3.2k|370|Updated Feb 25, 2026
Search Tags:#digital twin#NIfTI#catheter navigation#isaac for healthcare#vessel segmentation#totalsegmentator#ct preprocessing
One-click install
npx skills add https://github.com/NVIDIA/skills --skill i4h-catheter-navigation-digital-twin
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: i4h-catheter-navigation-digital-twin
Source: https://github.com/NVIDIA/skills/tree/main/skills/i4h-catheter-navigation-digital-twin
Command: npx skills add https://github.com/NVIDIA/skills --skill i4h-catheter-navigation-digital-twin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Preparing patient-specific CT data for catheter navigation simulation requires multiple manual steps: locating or downloading a CT dataset, preprocessing the NIfTI volume into an attenuation cache, and segmenting the arterial tree into a vessel mask and centerline. This Skill automates that pipeline so agents produce a ready-to-use ct_cache for viewport and DRR rendering.

Core Features & Use Cases

  • CT Preprocessing: Converts a NIfTI CT volume into an attenuation cache (mu_volume.npy + metadata.json) using the i4h catheter_navigation preprocess_ct mode.
  • Vessel Segmentation: Runs segment_vessels against TotalSegmentator ground truth to produce the arterial mask and centerline artifacts.
  • Dataset Bootstrap: Downloads the documented public TotalSegmentator sample (~3.2 GB) when the user has no CT data of their own.
  • Use Case: A developer preparing a patient-specific simulation asks the agent to build a vasculature digital twin from a TotalSegmentator subject, then proceeds to the catheter viewport or DRR rendering skills using the generated cache.

Quick Start

Ask your agent to build a vasculature digital twin from your TotalSegmentator subject at a given path and write the cache to /tmp/ct_cache.

Frequently Asked Questions about i4h-catheter-navigation-digital-twin

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a vasculature digital twin from a CT scan?▼

Point SUBJ at an extracted TotalSegmentator subject containing ct.nii.gz and a segmentations directory, then run the i4h preprocess_ct mode followed by segment_vessels. The resulting cache holds mu_volume.npy, metadata.json, vessel mask, and centerline artifacts.

How to prepare CT data for catheter navigation simulation?▼

Preprocess the CT NIfTI into an attenuation cache with preprocess_ct, then segment the arterial tree with segment_vessels using TotalSegmentator ground truth. The generated ct_cache is consumed directly by the catheter viewport and DRR rendering stages.

What dataset works with the i4h catheter navigation digital twin workflow?▼

The documented public dataset is the TotalSegmentator small subset (~3.2 GB), downloadable via a Dropbox URL in the skill. Contrast-enhanced CTA subjects work best, and users must comply with the dataset license without committing patient data.

Why does segment_vessels fail with a --ts-gt-dir error?▼

The error occurs when the subject's segmentations directory is missing. Confirm that ${SUBJ}/segmentations exists and follows the TotalSegmentator ground truth layout before rerunning the segmentation step.

What are the limitations of CT-based digital twin generation?▼

The skill ships no data, so users must download or supply their own CT volume. Segmentation is CPU/GPU mixed and can take several minutes, large volumes need 32 GB or more RAM, and the Zenodo mirror is throttled so the Dropbox URL is preferred.

Related Skills

Explore More Software Engineering Skills→
imbue-aiimbue-ai
213

manually-qa-branch

Set up an isolated Sculptor environment to test a pull request commit.

Official
Advanced
langcarelangcare
53

procedure-note-template

Generate standardized bedside procedure notes from FHIR patient data.

Official
Advanced
jaechang-hitsjaechang-hits
298

nnunet-segmentation

Segment anatomical structures and lesions in medical images using nnU-Net.

Community
Advanced
sayalinvidiasayalinvidia

nv-segment-ct

Run NV-Segment-CT VISTA3D segmentation on CT volumes and output JSON evidence payloads.

Community
Advanced