Quantification of reconstructed SPECT non-uniformity.

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Processing

Background:

• Assessing image quality is crucial for accurate diagnosis in nuclear medicine.
• Quantifying non-uniformity in single photon emission computed tomography (SPECT) is essential for reliable image interpretation.

Purpose of the Study:

• To present a simple, reproducible, and sensitive method for quantifying reconstructed image non-uniformity in SPECT.
• To evaluate the method's performance with varying data acquisition parameters and simulated non-uniformity.

Main Methods:

• Reconstructed transaxial SPECT slices are converted to polar images using radial profiles.
• Radial and tangential uniformity are calculated from intensity variations in polar images.
• Axial uniformity is assessed by comparing slice-averaged radial profiles to the overall average profile.

Main Results:

• The proposed indices demonstrate reproducibility and sensitivity to system response changes.
• Optimal performance is achieved with prefiltered data exceeding 1 million counts per slice.
• The method effectively detects clinically relevant imaging artifacts.

Conclusions:

• The developed method provides a reliable tool for quantifying SPECT image non-uniformity.
• It is suitable for quality control programs and performance comparisons between SPECT systems.