Detection of Small Metastatic Brain Tumors Comparison of 3D Contrast-Enhanced Whole-Brain Black-Blood Imaging and MP-RAGE Imaging

Abstract

Objective: Early and accurate diagnosis of small metastatic brain tumors may affect outcomes and treatment strategies. For this reason, 3-dimensional (3D) thin-section imaging is preferred. However, with conventional contrast-enhanced (CE) 3D imaging, such as magnetization-prepared rapid gradient echo (MP-RAGE), many visually enhanced vessels may mimic small metastatic tumors, hindering tumor detection. CE black-blood single-slab 3D turbo-spin echo imaging (BB-ssTSE) was recently developed, which uses variable refocusing flip angles and flow-sensitizing gradient schemes, to enhance metastatic brain tumors while selectively suppressing blood vessels. The purpose of this work was to investigate the efficiency of the proposed CE BB-ssTSE in detecting small metastatic brain tumors as compared with conventional MP-RAGE. Materials and Methods: Numerical comparisons of MP-RAGE and BB-ssTSE were performed by simulation studies to investigate the signal/contrast behaviors of flowing blood and stationary CE tumors. For in vivo studies, we enrolled 35 patients (18 women; mean age, 58.1 years) with breast or lung cancer who underwent brain magnetic resonance imaging. After administering a double dose of contrast medium, whole-brain 2-dimensional T1-weighted imaging followed by high-resolution isotropic 3D BB-ssTSE and MP-RAGE was performed at 3.0 T. Two reviewers independently evaluated the presence of metastatic brain tumors using: (1) MP-RAGE; (2) BB-ssTSE; and (3) MP-RAGE + BB-ssTSE sequentially in 3 review sessions, 2 weeks apart. The lesions were classified by size into 2 groups: large (>= 5 mm) and small (<5 mm). Both reviewers marked all tumors detected at each session. Another reviewer combined the results of the 2 reviewers and compared the detection rates of metastatic brain tumors between BB-ssTSE and MP-RAGE by using follow-up imaging. Intraclass correlation coefficients between the 2 reviewers were measured. Results: Numerical simulations showed that the proposed BB-ssTSE effectively attenuated the signal intensity of flowing blood over the entire echo train, resulting in CE tumor-to-white matter contrast comparable with conventional MP-RAGE. The combined evaluation of MP-RAGE + BB-ssTSE showed 242 tumors in 28 patients. Of these, 153 lesions were <5 mm. MP-RAGE found 111 small metastatic brain tumors, BB-ssTSE found 150, and MP-RAGE + BB-ssTSE found 153. Significantly, more small tumors were detected by BB-ssTSE than MP-RAGE (P = 0.001, Wilcoxon signed-rank test). All large tumors were detected similarly by both MP-RAGE and BB-ssTSE. By combined results for MP-RAGE + BB-ssTSE, sensitivities for detection of small metastatic tumors were 72.5% for MP-RAGE and 98.0% for BB-ssTSE (P < 0.0001, McNemar test). Intraclass correlation coefficients between the 2 reviewers were 0.826 for MP-RAGE and 0.954 for BB-ssTSE. Conclusion: Compared with conventional MP-RAGE, the proposed CE BB-ssTSE imaging, which enhances tumors while selectively suppressing blood vessels, leads to significantly better detection of small metastatic brain tumors <5 mm.