Evaluating scintigraphic dyssynchrony as radiomic parameters for dynamic laminar and turbulent fluids in compartmental phantom systems: A signal processing study

Abstract Introduction : Radiomics quantify radiological data to correlate with clinical findings. Dyssynchrony, a proposed radiomic parameter measured via phase images, reflects the temporal discoordination of ventricular contraction, which can impair overall cardiac efficiency. This study assessed the consistency and reliability of dyssynchrony in laminar and turbulent flow compartments under varying image acquisition. It also evaluated the relationship between dyssynchrony and fluid dynamics alterations. Methods : The dataset included 64 dynamic images using gamma camera (128,000 frames) generated using an in-home phantom, representing combinations of flow velocity, count, and frame rates. Phase and amplitude images were generated and analyzed to calculate synchrony, entropy, approximate entropy (ApEn), and bounded-ApEn for different rotation directions. Entropy values were examined under parameter changes, with comparisons using Pearson’s test, ANOVA, logistic regression, and receiver operating characteristic (ROC) analysis. Results : Images were categorized by activity concentrations: Group 1 (37 MBq), Group 2 (29.5 MBq), and Group 3 (18.5 MBq). Group 1 showed a strong negative correlation between entropy and frame rates (r = −0.991, p < 0.001), while Group 3 displayed positive correlations between frame rate, ApEn, gray count, and pixel count. Logistic regression predicted turbulence (AUC = 0.93) and direction (AUC = 0.96) using bounded-ApEn. Regression analysis indicated ApEn and bounded-ApEn significantly predicted vortex parameters (R² = 93%). Conclusion : Dyssynchrony metrics, including entropy, ApEn, and bounded-ApEn, demonstrated consistent measurements across varying conditions. These findings highlight their potential for enhancing diagnostic accuracy and guiding personalized therapeutic strategies for conditions influenced by blood flow patterns