Airway Segmentation and Centerline Extraction from 17 Thoracic CT Scans

Data from a study compares three airway segmentation methods using CT lung datasets from 17 patients. The dataset includes quantitative measures such as total branch length, number of branches, number of generations, and 3D model volume for each method. It was authored by Pall Jens Reynisson and published in 2020.

Use Cases

• Compare the number of branches and generations identified by different segmentation methods like TSF and MPM.
• Analyze the relationship between the total length of branches and the volume of the 3D airway model across patient datasets.
• Evaluate the level of automation by correlating the number of user clicks required with segmentation quality metrics.
• Benchmark segmentation accuracy using reference segmentation comparison averages and standard deviations provided for MPM and TSF.

Strengths

• Includes data from 17 distinct patient CT lung datasets.
• Provides multiple quantitative measures per method: total branch length, number of branches, number of generations, and 3D model volume.
• Compares three established segmentation approaches: Mimics (Basic and Pulmonology modules), OsiriX, and the Tube Segmentation Framework (TSF).
• Dataset is published under a permissive CC0 1.0 public domain license.

Limitations

• Small sample size of 17 patients limits statistical power and generalizability of findings.
• Dataset is from a single study focused on methodological comparison, not a large-scale clinical repository.
• No raw CT image data or columnar sample data is provided, only the comparative results and metrics from the study.

Provenance

Source

Dryad digital repository.

Collection Method

CT lung datasets from 17 patients were processed and compared using Mimics software (Basic and Pulmonology modules), OsiriX, and the Tube Segmentation Framework (TSF).

Freshness

Last updated in 2020.