Deformable 3D/3D CT-to-digital-tomosynthesis image registration in image-guided bronchoscopy interventions

Area of Science:

• Medical Imaging
• Interventional Pulmonology
• Image Registration

Background:

• Traditional bronchoscopy uses CT, radiography, and CBCT for lung biopsies, facing limitations like radiation dose and image distortion.
• Digital tomosynthesis (DTS) offers advantages over CBCT but lacks depth resolution, requiring integration with CT.
• CT-to-body divergence due to patient motion and anatomical changes impedes accurate CT guidance during bronchoscopy.

Purpose of the Study:

• To develop and evaluate a novel deformable 3D/3D CT-to-DTS registration algorithm.
• To mitigate CT-to-body divergence for improved image-guided bronchoscopy interventions.
• To enhance the accuracy of biopsies for peripheral lung lesions.

Main Methods:

• Proposed a multistage, multiresolution deformable registration algorithm using affine and elastic B-spline transformations.
• Utilized bone and lung mask images, a Gaussian image pyramid, and a multigrid strategy.
• Employed normalized correlation coefficient and a multimetric weighted cost function for registration accuracy.

Main Results:

• The algorithm achieved promising results on simulated and real patient bronchoscopy data.
• Quantitative assessment showed a mean Dice coefficient (DC) of 0.82 and 0.74, and mean Average Symmetric Surface Distance (ASSD) of 0.65mm and 0.93mm for simulated and real data, respectively.
• Visual inspection confirmed the qualitative accuracy of the registration.

Conclusions:

• The developed CT-to-DTS registration algorithm effectively addresses CT-to-body divergence in image-guided bronchoscopy.
• This method provides a foundation for CT-aided intraoperative DTS imaging, improving guidance for lung biopsies.
• Future work will focus on automated optimization of metric weights for enhanced performance.