Impact of displacement, artifacts, thickness variations, and scanning protocols on cone-beam computed tomography image quality and registration accuracy in radiotherapy

Abstract

Objectives

To evaluate the impact of displacement, anatomical thickness variations, artifacts and scanning protocols on cone-beam computed tomography(CBCT) image quality and registration accuracy of plan CT(pCT)-CBCT in image-guided radiation therapy(IGRT).

Methods

A Catphan 504 phantom and an anthropomorphic head phantom were utilized to assess image quality and registration errors, respectively. Controlled phantom displacements (5 mm vertical, longitudinal, and lateral), addition of a 2 cm resin bolus or 0.25 mm lead apron to mimic displacement, anatomical thickness variations and artifacts, respectively. Nine preset CBCT scanning protocols, in addition to an adjustable lowest-dose protocol were tested under six mimic scenarios. Visual spatial resolution, MTF10 %, uniformity, and contrast-to-noise ratio [CNR] and registration errors (translation/rotation errors) were analyzed.

Results

A total of 180 catphan and 180 head phantom scans yielded 360 registration outcomes. Displacement, anatomical thickness variation, artifacts, and scanning protocols significantly impacted image quality (spatial resolution, uniformity, CNR; p < 0.05). Translational registration errors increased with displacement, thickness variation and artifacts, while scanning protocols and registration methods exhibited no significant effect. Rotational errors were influenced by displacement, artifacts, and registration methods (p < 0.05) but not by protocols or thickness variation. The lowest-dose protocol markedly degraded uniformity and CNR without compromising registration accuracy.

Conclusions

Displacement, anatomical thickness variations, and artifacts significantly impair CBCT image quality and registration accuracy. Scanning protocols modulate image quality, but do not detrimentally affect registration accuracy in head phantom. The preset CBCT scanning protocol of linear accelerator requires optimization to prioritize clinical objectives while balancing radiation exposure with registration performance.