Robustness of textural analysis features in quantitative ^{99 m}Tc and ¹⁷⁷Lu SPECT-CT phantom acquisitions

Area of Science:

• Medical Imaging
• Radiochemistry
• Quantitative Analysis

Background:

• Robustness and volume independence of Textural Analysis features are crucial for clinical molecular imaging.
• Recent guidelines suggest using phantoms to identify reliable radiomic features.
• This study evaluates phantoms for SPECT quantification of 99mTc and 177Lu.

Purpose of the Study:

• To assess the utility of phantoms for evaluating Textural Analysis features in SPECT.
• To determine feature robustness and volume dependency for 99mTc and 177Lu.
• To identify features suitable for clinical quantitative SPECT studies.

Main Methods:

• Utilized a uniform phantom for volume dependency testing and a custom 'Revolver' phantom for repeatability assessment.
• Acquired SPECT data using 99mTc and 177Lu.
• Extracted and analyzed 67 Textural Analysis features for robustness and volume dependency.

Main Results:

• Features with high volume dependency or poor repeatability were excluded.
• 39 features remained suitable for 99mTc, and 33 for 177Lu.
• The phantoms successfully identified features for quantitative SPECT.

Conclusions:

• Phantoms are effective for assessing Textural Analysis features in quantitative SPECT for 99mTc and 177Lu.
• Feature selection is influenced by center-specific camera performance and protocols.
• This approach aids in developing robust radiomic analysis for SPECT clinical applications.