A motion-compensated reconstruction algorithm improves reader-assessed cone-beam CT image quality during TACE
Katherine L McNamara1, Austin Ng1, Alexander M Vezeridis1
1Department of Radiology, Stanford University, Palo Alto, United States.
Background:
Cone-beam computed tomography (CBCT) is a critical component of liver directed therapies used to identify tumor feeding vessels. CBCT images are often limited by breathing motion artifacts. This study evaluates if post-processing CBCT images with a motion compensation algorithm improves image-quality for transarterial chemoembolization (TACE).
Methods:
A motion compensated reconstruction algorithm was applied to CBCT images from 51 patients who underwent TACE with full-strength of half-strength contrast. All cases demonstrated respiratory motion artifact. Subjective analysis of CBCT images before and after motion compensation was performed by four blinded radiologists.
Results:
Inter-reader agreement was substantial (Gwet's AC1 > 0.60). Across all cases, vessel sharpness (p = 0.023), confidence in identifying all feeding vessels (p = 0.036), and overall image quality (p = 0.024) were improved with the motion compensated reconstruction. The mean confidence in identifying all feeding vessels was improved in both the full-strength and half-strength contrast cases. Tumor conspicuity (p = 0.112) and the number of feeding vessels identified (p = 0.133) were not significantly different between groups.
Conclusions:
The motion compensated reconstruction algorithm improved CBCT image quality scores across multiple criteria and improved reader confidence in the imaging findings. These findings support further prospective evaluation to determine if this translates into reduced repeat imaging, lower radiation exposure, shorter procedure time, or improved treatment selectivity.
