Impact of Acquisition Method and Region of Interest Placement on Inter-observer Agreement and Measurement of Tumor Response to Targeted Therapy Using Dynamic Contrast-Enhanced Ultrasound

This study evaluated the impact of different acquisition methods, user-directed region of interest placement and post-processing steps on the quantification of dynamic contrast-enhanced ultrasound measurements of blood volume in 29 patients with renal cancer, pre- and post-treatment. Specifically, we compared tumor quantification using multiple planes versus a single plane, breathhold versus free breathing and large region of interest versus a region targeting the area of highest vascularity. Performance was evaluated using area under the receiver operating characteristic curves to identify the method that best predicts progression-free survival. The intra-class correlation coefficient was also used to investigate how the same parameters affect inter-observer agreement. Of the different methods used to quantify blood volume in this study, the combination that had the highest level of inter-observer agreement (intra-class correlation coefficient = 0.8-0.97) and was the best predictor of progression-free survival was the change in blood volume measured (area under receiver operating characteristic curve = 0.77, p = 0.04) by a multiplane average, acquired during quiet breathing, quantified using a region of interest that encompassed the entire tumor.