Comparison of 3D two-point Dixon and standard 2D dual-echo breath-hold sequences for detection and quantification of fat content in renal angiomyolipoma

Abstract

To assess the utility of a 3D two-point Dixon sequence with water-fat decomposition for quantification of fat content of renal angiomyolipoma (AML). 84 patients underwent renal MRI including 2D in-and-opposed-phase (IP and OP) sequence and 3D two-point Dixon sequence that generates four image sets [IP, OP, water-only (WO), and fat-only (FO)] within one breath-hold. Two radiologists reviewed 2D and 3D images during separate sessions to identify fat-containing renal masses measuring at least 1cm. For identified lesions subsequently confirmed to represent AML, ROIs were placed at matching locations on 2D and 3D images and used to calculate 2D and 3D SI(index) [(SI(IP)-SI(OP))/SI(IP)] and 3D fat fraction (FF) [SI(FO)/(SI(FO)+SI(WO))]. 2D and 3D SI(index) were compared with 3D FF using Pearson correlation coefficients. 41 AMLs were identified in 6 patients. While all were identified using the 3D sequence, 39 were identified using the 2D sequence, with the remaining 2 AMLs retrospectively visible on 2D images but measuring under 1cm. Among 32 AMLs with a 3D FF of over 50%, both 2D and 3D SI(index) showed a statistically significant inverse correlation with 3D FF (2D SI(index): r=-0.63, p=0.0010; 3D SI(index): r=-0.97, p<0.0001). 3D two-point Dixon sequence may provide a reasonable alternative to 2D dual-echo sequence for detection of renal AML and may have additional value for quantification of fat content of these lesions given the observation that 3D FF, unlike 2D and 3D SI(index), is not limited by ambiguity of water or fat dominance. This may assist clinical management of AML given evidence that fat content predicts embolization response.