Retracted: Reducing motion artifacts in 4D MR images using principal component analysis (PCA) combined with linear...

Area of Science:

• Medical Imaging
• Radiology
• Image Processing

Background:

• Four-dimensional MRI (4D-MRI) is crucial for accurate tumor delineation but often suffers from motion artifacts.
• These artifacts can compromise the precision of tumor targeting and the representation of surrounding organs.

Purpose of the Study:

• To develop and validate an approach using principal component analysis (PCA) and a linear polynomial fitting model to reduce motion artifacts in 4D-MRI.
• To remodel displacement vector fields (DVFs) obtained from deformable image registration (DIR) for improved 4D-MRI quality.

Main Methods:

• DVFs were calculated using DIR from 4D-MRI datasets of patients with liver cancer or non-small cell lung cancer.
• DVFs were preprocessed with polynomial fitting and then decomposed using PCA to capture regular respiratory motion.
• Synthetic 4D-MR images with reduced artifacts were generated by applying remodeled DVFs.

Main Results:

• The proposed method demonstrated high correlation coefficients (CC) between synthetic 4D-MRI and cine-MRI (0.98-0.99) and 4D CT (0.95-0.96).
• Differences in motion amplitude between synthetic 4D-MRI and other methods were minimal (0.15-0.76 mm).
• The technique effectively reduced motion artifacts, showing strong agreement with reference imaging modalities.

Conclusions:

• The combination of PCA and polynomial fitting effectively models respiratory motion and reduces artifacts in 4D-MRI.
• This approach shows significant potential for enhancing 4D-MRI quality in oncological imaging.
• Improved image quality can lead to more accurate tumor target delineation and treatment planning.