A heterogeneous phantom study for investigating the stability of PET images radiomic features with varying reconstruction settings

The purpose of this work was to assess the capability radiomic features in distinguishing PET image regions with different uptake patterns. Furthermore, we assessed stability varying reconstruction settings. An in-house phantom designed and constructed, consisting homogenous heterogenous artificial inserts. Four artificially constructed inserts were placed into a water filled levels radioactivity simulate homogeneous heterogeneous imaged for 80 min. images reconstructed whilst parameters. parameters adjusted included, number ordered subsets, iterations, use time-of-flight filter cut off. Regions interest (ROI) established by segmentation from images. In total seventy eight 3D each ROI unique extracted. Friedman test used determine statistical power feature differentiating hetero/homogeneous configurations. Coefficient Variation (COV) feature, respect setting stability. Forty three out found be stable (COV ≤ 5%) against all To provide any utility, are required differentiate between hetro/homogeneity. Of forty features, fifteen (35%) showed statistically significant difference Such included GLCM (Difference average, Difference entropy, Dissimilarity Inverse difference), GLRL (Long run emphasis, Grey level non uniformity Run percentage) NGTDM (Complexity Strength). finding suggests that capable radioactive distribution patterns demonstrate heterogeneity. Therefore, could serve as an adjuvant diagnostic tool along traditional imaging. However, choice needs account variability introduced when settings used. Standardization across sites performing analysis multi-centre trials should considered.