Automated Deformable Registration and Three-dimensional Margin Assessment for Predicting Local Recurrence after Lung...

Area of Science:

• Medical Imaging
• Computational Biology
• Oncology

Background:

• Lung thermal ablation is a key treatment for lung tumors.
• Accurate assessment of ablation margins is crucial for predicting treatment efficacy.
• Current methods for margin assessment can be challenging and time-consuming.

Purpose of the Study:

• To develop and validate a lung-specific deformable image registration algorithm for thermal ablation.
• To evaluate the predictive value of three-dimensional (3D) ablation margin assessment for local recurrence.
• To optimize lung thermal ablation techniques through improved margin analysis.

Main Methods:

• A four-stage deformable image registration framework was developed, including affine, lung-mask-guided, and local registration steps.
• Free-form B-spline transformations and cost function masking were employed for precise registration.
• Tumor, ablation zone, and lung segmentation preceded registration and margin quantification.
• Target registration error (TRE) assessed registration accuracy; distance-transform analysis quantified 3D margins.

Main Results:

• The algorithm achieved high accuracy with a mean TRE of 0.4 mm ± 0.3 mm.
• A mean ablation margin of 1.6 mm ± 2.1 mm was quantified.
• Larger ablation margins were significantly associated with a longer time to local recurrence (SHR 0.5 per mm increase, P < .001).
• A 2-mm margin threshold predicted a 2-year local recurrence rate of 3% (AUC 0.86).

Conclusions:

• The developed lung-optimized deformable image registration framework enables accurate, automated 3D ablation margin quantification.
• Ablation margin size is a significant independent predictor of local recurrence after lung thermal ablation.
• This method can enhance treatment planning and outcome prediction in interventional oncology.