Target volume and artifact evaluation of a new data-driven 4D CT

Area of Science:

• Radiology and Medical Imaging
• Radiation Oncology
• Medical Physics

Background:

• Four-dimensional computed tomography (4D CT) is crucial for defining internal gross target volumes (IGTVs) in lung cancer radiation therapy.
• Traditional 4D CT relies on external surrogates for motion management.
• Data-driven 4D CT presents a novel, image-based approach.

Purpose of the Study:

• To describe the data-driven 4D CT technique.
• To compare IGTV contours generated by data-driven 4D CT with those from standard 4D CT.
• To evaluate image artifacts and reproducibility between the two methods.

Main Methods:

• Cine CT data from 35 patients undergoing stereotactic body radiation therapy were analyzed.
• Internal gross target volumes (IGTVs) were contoured using both standard and data-driven 4D CT methods.
• Reproducibility, contour similarity (Dice similarity coefficient, Hausdorff distance), and image artifacts were quantitatively assessed.

Main Results:

• No significant difference in IGTV volume between data-driven and standard 4D CT (2.1% smaller, P=.08).
• High contour similarity (Dice 0.949 ± 0.006, mean surface separation 0.4 ± 0.1 mm).
• Data-driven 4D CT demonstrated significantly fewer and less severe artifacts (artifact score +0.37, P=1.5 × 10⁻⁵).

Conclusions:

• Data-driven 4D CT provides comparable IGTV definitions to standard 4D CT, suitable for clinical use without IGTV adjustments.
• The method shows potential clinical advantages due to artifact reduction and simplified setup.
• Observed patient-specific differences were mainly due to automatic contouring limitations or artifact variations.