Room: Exhibit Hall
Purpose: To examine the clinical efficacy of a novel Multi-Source Adaptive Fusion method (MSAF) in enhancing tumor-to-tissue contrast to noise ratio (CNR) and reducing inter-observer variation in liver tumor delineation.
Methods: Twelve patients with liver cancer were retrospectively evaluated. The MRI sequences of T1 weighted portvenous phase(T1-w), T1-w19 mins delay post-contrast (T1-w 19mins), T2 weighted (T2-w), diffusion weighted images (DWI) were acquired by Philips 3T MRI scanner during patient breath-hold. All MRI sequences underwent deformable registration based on the planning CT, which were then imported into in-house developed MSAF software and compiled into a fused image, with adjustment of linear weightings from -1 to 1 with 0.2 interval.Tumor CNR were measured in Eclipse and compared between planning CT, four original MR sets and the fused MR set. Four observers delineated tumors on the planning CT, four original MR sets and fused MR set for all patients. Inter-observer variation in tumor volume was evaluated using the standard deviation (SD) across observers for each MR set.
Results: The fused MRI has the highest average tumor CNR comparing with planning CT and other original MR sets the average (±SD) tumor CNR was 0.44±0.22, 2.35±2.09, 3.66±1.74, 3.08±1.23, 3.68±2.00, and 4.96±2.2 for CT, T1+C-w, T1-w 19mins, T2-w, DWI and Fused MRI, respectively. The fused MRI has the smallest inter-observer variation in liver tumor delineation.The average tumor volume (±SD) was 147.8±8.9 (range:2.68-892) cmΛ3, 117.1±3.2 (range: 1.1-811.2) cmΛ3,119.7±3.1 (range: 2.1-819) cmΛ3, 121.4±4.3 (range: 2.5-817) cmΛ3, 120.4.0±2.4 (range: 2.1-813) cmΛ3 and 119.5±1.0 (range: 2.2-810)cmΛ3 for planning CT, T1-w, T1-w 19 mins, T2-w, DWI and Fused MRI, respectively.
Conclusion: The MSAF method holds great promises in enhancing tumor CNR and reducing inter-observer variation in liver tumor delineation.
Not Applicable / None Entered.
Not Applicable / None Entered.