Characterization of Renal Masses: Is There a Threshold for Differentiating Noise from True Enhancement on Subtraction Imaging?

Samson Wong¹, Sooah Kim¹, Nicole Hindman¹, Daniel Sahlein¹, Vivian S. Lee¹

Postprocessing image subtraction has been used successfully in MR in characterizing and evaluating enhancement of renal lesions. Image noise, however, is a potential pitfall in MR subtraction imaging. Our objective was to evaluate if there is a threshold signal-to-noise ratio of subtraction images for distinguishing nonenhancing from enhancing renal lesions. We evaluated the approximate SNR (aSNR) of MR subtracted images of nonenhancing lesions (simple cysts and T1 hyperintense cysts) and renal neoplasms and found that the aSNR of cysts were significantly lower than renal neoplasms (p<.001). In addition, a cutoff aSNR value of 10 achieved a 100% specificity and sensitivity.