A modified McKinnon-Bates (MKB) algorithm for improved 4D cone-beam computed tomography (CBCT) of the lung

Area of Science:

• Medical Imaging
• Radiotherapy Physics
• Computational Imaging

Background:

• Four-dimensional (4D) cone-beam computed tomography (CBCT) is crucial for lung radiotherapy motion management.
• Traditional reconstruction algorithms like Feldkamp-Davis-Kress (FDK) produce artifacts due to sparse sampling.
• The McKinnon-Bates (MKB) algorithm reduces streaking but introduces ghosting artifacts.

Purpose of the Study:

• To identify and correct shortcomings in the McKinnon-Bates (MKB) algorithm for 4D CBCT.
• To develop an improved algorithm for reducing artifacts in lung 4D CBCT reconstruction.

Main Methods:

• A modified McKinnon-Bates (mMKB) algorithm was developed by destreaking the prior image.
• A 4D bilateral filter was applied for noise suppression and edge preservation (mMKB^bf).
• Algorithms were tested using the 4D XCAT phantom and in vivo thorax studies.

Main Results:

• The mMKB algorithm significantly reduced ghosting artifacts and increased contrast-to-noise ratios (CNRs).
• mMKB^bf achieved up to a 300% CNR improvement over FDK-PC reconstructions.
• In vivo studies confirmed artifact reduction, enabling visualization of tumors with significant motion.

Conclusions:

• The modified McKinnon-Bates (mMKB) algorithm with bilateral filtering provides high-quality 4D CBCT images.
• These improved images may offer sufficient detail for patient verification in radiotherapy.
• The mMKB algorithm represents a significant advancement in artifact reduction for 4D CBCT.