Improved identification of abdominal aortic aneurysm using the Kernelized Expectation Maximization algorithm

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Cardiovascular Imaging

Background:

• Abdominal aortic aneurysm (AAA) rupture risk is typically assessed by diameter, but growth is unpredictable.
• ¹⁸F]-NaF Positron Emission Tomography (PET) shows promise for AAA monitoring by measuring calcified lesion uptake.
• Low spatial resolution in PET imaging can limit diagnostic accuracy.

Purpose of the Study:

• To investigate the effectiveness of anatomically guided kernelized expectation maximization (KEM) and hybrid KEM (HKEM) for AAA diagnosis.
• To statistically evaluate the improvements offered by KEM and HKEM over standard methods.
• To assess the potential for improved quantification of small lesions in clinical applications.

Main Methods:

• Reconstruction of 61 AAA patient and 11 control patient datasets using OSEM, HKEM, and KEM algorithms.
• Analysis of reconstructed images using the target-to-blood-pool ratio.
• Application of statistical tests to determine the significance of observed improvements.

Main Results:

• All tested algorithms demonstrated comparable diagnostic power.
• HKEM and KEM significantly improved the recovery of lesion uptake compared to OSEM.
• Diagnostic accuracy was enhanced by up to 22% using HKEM and KEM.

Conclusions:

• HKEM and KEM offer significant improvements in PET-based AAA diagnosis and growth prediction.
• These advanced reconstruction methods enhance the accuracy of lesion quantification.
• Potential for similar improvements in quantifying small lesions in other clinical areas, such as oncology.