SU‐F‐J‐103: Assessment of Liver Tumor Contrast for Radiation Therapy: Inter‐Patient and Inter‐Sequence Variability

Abstract

Purpose:To determine the variation in tumor contrast between different MRI sequences and between patients for the purpose of MRI‐based treatment planning.Methods:Multiple MRI scans of 11 patients with cancer(s) in the liver were included in this IRB‐approved study. Imaging sequences consisted of T1W MRI, Contrast‐Enhanced T1W MRI, T2W MRI, and T2*/T1W MRI. MRI images were acquired on a 1.5T GE Signa scanner with a four‐channel torso coil. We calculated the tumor‐to‐tissue contrast to noise ratio (CNR) for each MR sequence by contouring the tumor and a region of interest (ROI) in a homogeneous region of the liver using the Eclipse treatment planning software. CNR was calculated (I_Tum‐I_ROI)/SD_ROI, where I_Tum and I_ROI are the mean values of the tumor and the ROI respectively, and SD_ROI is the standard deviation of the ROI. The same tumor and ROI structures were used in all measurements for different MR sequences. Inter‐patient Coefficient of variation (CV), and inter‐sequence CV was determined. In addition, mean and standard deviation of CNR were calculated and compared between different MR sequences.Results:Our preliminary results showed large inter‐patient CV (range: 37.7% to 88%) and inter‐sequence CV (range 5.3% to 104.9%) of liver tumor CNR, indicating great variations in tumor CNR between MR sequences and between patients. Tumor CNR was found to be largest in CE‐T1W (8.5±7.5), followed by T2W (4.2±2.4), T1W (3.4±2.2), and T2*/T1W (1.7±0.6) MR scans. The inter‐patient CV of tumor CNR was also the largest in CE‐T1W (88%), followed by T1W (64.3%), T1W (56.2%), and T2*/T1W (37.7) MR scans.Conclusion:Large inter‐sequence and inter‐patient variations were observed in liver tumor CNR. CE‐T1W MR images on average provided the best tumor CNR. Efforts are needed to optimize tumor contrast and its consistency for MRI‐based treatment planning of cancer in the liver.This project is supported by NIH grant: 1R21CA165384