Task-Based Image Quality Assessment Comparing Classical and Iterative Cone Beam CT Images on Halcyon^®

Area of Science:

• Medical Physics
• Radiotherapy Imaging
• Image Reconstruction

Background:

• Filtered back projection (FBP) is standard for cone-beam CT (CBCT) in radiotherapy, despite advances in iterative reconstruction (IR) for diagnostic imaging.
• Varian's Halcyon® V2.0 accelerator introduces a novel IR algorithm for CBCT reconstruction.

Purpose of the Study:

• To evaluate and compare the image quality of CBCT images reconstructed using FBP versus a new IR algorithm.
• Assess the impact of reconstruction method on image quality metrics relevant to radiotherapy.

Main Methods:

• Three CBCT acquisition modes (head, thorax, pelvis large) on a Halcyon® system were tested.
• Image quality was assessed using a phantom with five acquisitions per mode, reconstructed with both FBP and IR.
• Task-based image quality metrics including noise power spectrum (NPS), task-based transfer function (TTF), and detectability index (d') were analyzed.

Main Results:

• IR reconstruction resulted in lower noise magnitude and NPS peak frequency compared to FBP across all modes.
• Higher 50% TTF values were observed with IR for both low and high-contrast inserts.
• Detectability index (d') values for simulated lesions were superior with IR, indicating improved lesion detection.
• Patient CBCT images reconstructed with IR were less noisy and smoother than those with FBP.

Conclusions:

• The IR algorithm significantly enhances CBCT image quality in radiotherapy compared to FBP.
• Improved image quality with IR can potentially increase the accuracy of online image registration.
• Adoption of IR for CBCT may lead to more precise patient positioning and reduced treatment errors.