[OA197] Normal ranges of cardiac imaging radiomic metrics in healthy volunteers 24 h post ultra-small particles of iron oxide (USPIO) administration

Purpose Radiomics is a novel post-processing imaging methodology by which statistical features are extracted from medical images. The radiomic features of a myocardial region of interest open the possibility that novel descriptive patterns, found in images, may selectively identify phenotypic features of myocardial disease. Ultra-small particles of iron oxide (USPIO) uptake in the myocardium may show insights into a variety of pathologies that involve macrophage activation. While myocardial uptake may be low, any MRI signal changes due to USPIO uptake may be difficult to detect using global measures. Hence the potential need for more subtle metrics. The study aims to investigate USPIO uptake by innate macrophages resistant in normal myocardium using radiomics. Methods 11 healthy volunteers were imaged using a 3T Philips MRI scanner pre and exactly 24 h post administration of 2 ( n  = 7) or 4 mg/kg ( n  = 4) of USPIO. A short axis stack of entire left ventricular myocardium was acquired using a multi-echo gradient sequence (10 mm thickness, 0 mm gap). To delineate the myocardium, endocardial and epicardial border contours were manually traced on the first echo in the sequence (TE = 1.15 ms). First order and second order radiomic features were extracted from each slice of the stacks. Differences between feature values pre and post USPIO administration were calculated using paired samples t -test. Results In the 4 mg/kg USPIO group significant differences were found in Mean ( p  = 0.027), Large Area Emphasis (LAE) ( p  = 0.027), Zone Variance (ZV) ( p  = 0.027), Large Area Low Gray Level Emphasis (LALGLE) ( p  = 0.027), and Large Area High Gray Level Emphasis (LAHGLE) ( p  = 0.027). For the 2 mg/kg group, significant differences were found for Informal Measure of Correlation 1 ( p  = 0.009), LAE ( p  = 0.014), ZV ( p  = 0.014), Zone Percentage ( p  = 0.014), LALGLE ( p  = 0.014), and LAHGLE ( p  = 0.014). Conclusions Radiomics assessment of multi-echo gradient magnetic resonance images of the myocardium is feasible. Compared to pre USPIO images, post USPIO radiomics was able to detect a reduction in voxel intensity, an increase in intensity and run length homogeneity, and an increase in coarseness; with a smaller distribution of shorter run lengths.