Comparing filtered backprojection and ordered-subsets expectation maximization for small-lesion detection and...

Area of Science:

• Nuclear Medicine
• Medical Imaging Analysis

Background:

• Iterative reconstruction (IR) methods like ordered-subsets expectation maximization (OSEM) are emerging alternatives to filtered backprojection (FBP) for single-photon emission computed tomography (SPECT) image reconstruction.
• Conflicting evidence exists regarding the diagnostic performance improvement of IR over FBP.

Purpose of the Study:

• To compare the diagnostic performance of OSEM versus FBP for detecting and localizing small lesions in simulated thoracic gallium SPECT images.
• To evaluate the impact of attenuation correction (AC) on lesion detection and localization accuracy for both OSEM and FBP.

Main Methods:

• Simulated thoracic gallium SPECT images were generated using an analytic projector and a cardiac torso phantom.
• Lesion detection and localization accuracy were assessed using localization receiver operating characteristic (LROC) methodology by 5 nuclear medicine physicians.
• Statistical analysis involved two-way ANOVA and Scheffé multiple comparisons test.

Main Results:

• Filtered backprojection (FBP) with or without attenuation correction (AC) showed minimal differences in lesion detection and localization accuracy.
• Ordered-subsets expectation maximization (OSEM) with AC significantly extended the contrast range for accurate lesion detection and localization.
• No significant difference in reconstruction accuracy was observed between OSEM and FBP when AC was not applied.

Conclusions:

• OSEM with AC demonstrates potential for improved detection and localization of thoracic gallium-labeled lesions compared to FBP reconstruction.
• The findings suggest that OSEM with AC may enhance diagnostic performance in specific clinical scenarios.