Fuzzy clustering-based segmented attenuation correction in whole-body PET imaging

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Processing

Background:

• Segmented attenuation correction is crucial in positron emission tomography (PET) to minimize noise from transmission scans.
• Accurate attenuation correction is vital for reliable quantification and image interpretation in PET.

Purpose of the Study:

• To introduce a novel Fuzzy C-Means (FCM) algorithm-based method for segmenting transmission images in whole-body PET scanning.
• To improve noise reduction in attenuation correction maps while accurately accounting for varying tissue attenuation coefficients.

Main Methods:

• Employs a median filtering procedure followed by the FCM algorithm to segment PET transmission images into clusters representing different tissues (air, lungs, soft tissue).
• The unsupervised and adaptive method processes both pre- and post-injection transmission images from various sources (68Ge, 137Cs).
• Incorporates a merging process for redundant clusters and anatomical knowledge to refine segmentation, generating attenuation correction factors via forward projection.

Main Results:

• Demonstrated significant improvement in image quality and a clear reduction in noise propagation in phantom and whole-body clinical PET studies.
• Enabled a reduction in transmission scan duration without compromising image quality.
• Showed high correlation (R2 = 0.96) in maximum standardized uptake values (SUVs) for lung nodules, with a statistically significant decrease in SUV (17.03% +/- 8.4%, P < 0.01) using the segmented method.

Conclusions:

• The FCM algorithm offers a robust and effective method for segmenting PET transmission images, enhancing attenuation correction.
• The technique shows potential for improving diagnostic accuracy and efficiency in clinical PET imaging.
• Further research can explore limitations and prospective applications of this segmentation method in PET.