4D liver tumor localization using cone-beam projections and a biomechanical model

Area of Science:

• Medical Imaging
• Computational Biology
• Radiotherapy Physics

Background:

• Accurate liver tumor localization is crucial for effective radiotherapy.
• Current methods for 4D tumor localization using cone-beam CT (CBCT) face challenges, particularly in low-contrast regions.
• Deformable image registration techniques are essential for tracking tumor motion during treatment.

Purpose of the Study:

• To develop and evaluate a biomechanical modeling-guided liver CBCT estimation (Bio-CBCT-est) technique for improved 4D liver tumor localization.
• To generate new CBCT images by deforming prior CT/CBCT images using deformation vector fields (DVFs).
• To enable automatic 4D tumor localization by propagating tumor contours using DVFs.

Main Methods:

• An iterative scheme combining intensity-driven 2D-3D deformation and biomechanical modeling-guided DVF regularization and optimization was employed.
• DVFs were solved by matching digitally reconstructed radiographs to 2D phase-sorted on-board projections.
• Finite element analysis-based biomechanical modeling was used to fine-tune intra-liver DVFs in low-contrast regions.

Main Results:

• The Bio-CBCT-est technique improved the average DICE similarity index from 0.48 to 0.77.
• Average center-of-mass error (COME) decreased from 7.7 mm to 2.2 mm.
• Average Hausdorff distance (HD) decreased from 10.6 mm to 5.9 mm, and cross-correlation (CC) increased from -0.004 to 0.422.

Conclusions:

• Bio-CBCT-est significantly enhances the accuracy of 4D liver tumor localization.
• The technique effectively propagates tumor contours, improving localization precision.
• This method holds promise for more accurate image-guided radiotherapy for liver cancer.